Merge pull request #1315 from FunAudioLLM/dev/lyuxiang.lx

Dev/lyuxiang.lx

README.md

@@ -126,7 +126,7 @@ import torchaudio
 
 **CosyVoice2 Usage**
 ```python
-cosyvoice = CosyVoice2('pretrained_models/CosyVoice2-0.5B', load_jit=False, load_trt=False, fp16=False, use_flow_cache=False)
+cosyvoice = CosyVoice2('pretrained_models/CosyVoice2-0.5B', load_jit=False, load_trt=False, fp16=False)
 
 # NOTE if you want to reproduce the results on https://funaudiollm.github.io/cosyvoice2, please add text_frontend=False during inference
 # zero_shot usage

cosyvoice/bin/export_jit.py

@@ -61,8 +61,7 @@ def main():
         model = CosyVoice(args.model_dir)
     except Exception:
         try:
-            # NOTE set use_flow_cache=True when export jit for cache inference
-            model = CosyVoice2(args.model_dir, use_flow_cache=True)
+            model = CosyVoice2(args.model_dir)
         except Exception:
             raise TypeError('no valid model_type!')
 
@@ -93,9 +92,9 @@ def main():
     else:
         # 3. export flow encoder
         flow_encoder = model.model.flow.encoder
-        script = get_optimized_script(flow_encoder, ['forward_chunk'])
+        script = get_optimized_script(flow_encoder)
         script.save('{}/flow.encoder.fp32.zip'.format(args.model_dir))
-        script = get_optimized_script(flow_encoder.half(), ['forward_chunk'])
+        script = get_optimized_script(flow_encoder.half())
         script.save('{}/flow.encoder.fp16.zip'.format(args.model_dir))
         logging.info('successfully export flow_encoder')
 

cosyvoice/bin/export_onnx.py

@@ -62,135 +62,58 @@ def main():
         model = CosyVoice(args.model_dir)
     except Exception:
         try:
-            # NOTE set use_flow_cache=True when export jit for cache inference
-            model = CosyVoice2(args.model_dir, use_flow_cache=True)
+            model = CosyVoice2(args.model_dir)
         except Exception:
             raise TypeError('no valid model_type!')
 
-    if not isinstance(model, CosyVoice2):
-        # 1. export flow decoder estimator
-        estimator = model.model.flow.decoder.estimator
-        estimator.eval()
-
-        device = model.model.device
-        batch_size, seq_len = 2, 256
-        out_channels = model.model.flow.decoder.estimator.out_channels
-        x, mask, mu, t, spks, cond = get_dummy_input(batch_size, seq_len, out_channels, device)
-        torch.onnx.export(
-            estimator,
-            (x, mask, mu, t, spks, cond),
-            '{}/flow.decoder.estimator.fp32.onnx'.format(args.model_dir),
-            export_params=True,
-            opset_version=18,
-            do_constant_folding=True,
-            input_names=['x', 'mask', 'mu', 't', 'spks', 'cond'],
-            output_names=['estimator_out'],
-            dynamic_axes={
-                'x': {2: 'seq_len'},
-                'mask': {2: 'seq_len'},
-                'mu': {2: 'seq_len'},
-                'cond': {2: 'seq_len'},
-                'estimator_out': {2: 'seq_len'},
-            }
-        )
-
-        # 2. test computation consistency
-        option = onnxruntime.SessionOptions()
-        option.graph_optimization_level = onnxruntime.GraphOptimizationLevel.ORT_ENABLE_ALL
-        option.intra_op_num_threads = 1
-        providers = ['CUDAExecutionProvider' if torch.cuda.is_available() else 'CPUExecutionProvider']
-        estimator_onnx = onnxruntime.InferenceSession('{}/flow.decoder.estimator.fp32.onnx'.format(args.model_dir),
-                                                      sess_options=option, providers=providers)
-
-        for _ in tqdm(range(10)):
-            x, mask, mu, t, spks, cond = get_dummy_input(batch_size, random.randint(16, 512), out_channels, device)
-            output_pytorch = estimator(x, mask, mu, t, spks, cond)
-            ort_inputs = {
-                'x': x.cpu().numpy(),
-                'mask': mask.cpu().numpy(),
-                'mu': mu.cpu().numpy(),
-                't': t.cpu().numpy(),
-                'spks': spks.cpu().numpy(),
-                'cond': cond.cpu().numpy()
-            }
-            output_onnx = estimator_onnx.run(None, ort_inputs)[0]
-            torch.testing.assert_allclose(output_pytorch, torch.from_numpy(output_onnx).to(device), rtol=1e-2, atol=1e-4)
-        logging.info('successfully export estimator')
-    else:
-        # 1. export flow decoder estimator
-        estimator = model.model.flow.decoder.estimator
-        estimator.forward = estimator.forward_chunk
-        estimator.eval()
-
-        device = model.model.device
-        batch_size, seq_len = 2, 256
-        out_channels = model.model.flow.decoder.estimator.out_channels
-        x, mask, mu, t, spks, cond = get_dummy_input(batch_size, seq_len, out_channels, device)
-        cache = model.model.init_flow_cache()['decoder_cache']
-        cache.pop('offset')
-        cache = {k: v[0] for k, v in cache.items()}
-        torch.onnx.export(
-            estimator,
-            (x, mask, mu, t, spks, cond,
-             cache['down_blocks_conv_cache'],
-             cache['down_blocks_kv_cache'],
-             cache['mid_blocks_conv_cache'],
-             cache['mid_blocks_kv_cache'],
-             cache['up_blocks_conv_cache'],
-             cache['up_blocks_kv_cache'],
-             cache['final_blocks_conv_cache']),
-            '{}/flow.decoder.estimator.fp32.onnx'.format(args.model_dir),
-            export_params=True,
-            opset_version=18,
-            do_constant_folding=True,
-            input_names=['x', 'mask', 'mu', 't', 'spks', 'cond', 'down_blocks_conv_cache', 'down_blocks_kv_cache', 'mid_blocks_conv_cache', 'mid_blocks_kv_cache',
-                         'up_blocks_conv_cache', 'up_blocks_kv_cache', 'final_blocks_conv_cache'],
-            output_names=['estimator_out', 'down_blocks_conv_cache_out', 'down_blocks_kv_cache_out', 'mid_blocks_conv_cache_out', 'mid_blocks_kv_cache_out',
-                          'up_blocks_conv_cache_out', 'up_blocks_kv_cache_out', 'final_blocks_conv_cache_out'],
-            dynamic_axes={
-                'x': {2: 'seq_len'},
-                'mask': {2: 'seq_len'},
-                'mu': {2: 'seq_len'},
-                'cond': {2: 'seq_len'},
-                'down_blocks_kv_cache': {3: 'cache_in_len'},
-                'mid_blocks_kv_cache': {3: 'cache_in_len'},
-                'up_blocks_kv_cache': {3: 'cache_in_len'},
-                'estimator_out': {2: 'seq_len'},
-                'down_blocks_kv_cache_out': {3: 'cache_out_len'},
-                'mid_blocks_kv_cache_out': {3: 'cache_out_len'},
-                'up_blocks_kv_cache_out': {3: 'cache_out_len'},
-            }
-        )
-
-        # 2. test computation consistency
-        option = onnxruntime.SessionOptions()
-        option.graph_optimization_level = onnxruntime.GraphOptimizationLevel.ORT_ENABLE_ALL
-        option.intra_op_num_threads = 1
-        providers = ['CUDAExecutionProvider' if torch.cuda.is_available() else 'CPUExecutionProvider']
-        estimator_onnx = onnxruntime.InferenceSession('{}/flow.decoder.estimator.fp32.onnx'.format(args.model_dir),
-                                                      sess_options=option, providers=providers)
-
-        for iter in tqdm(range(10)):
-            x, mask, mu, t, spks, cond = get_dummy_input(batch_size, random.randint(16, 512), out_channels, device)
-            cache = model.model.init_flow_cache()['decoder_cache']
-            cache.pop('offset')
-            cache = {k: v[0] for k, v in cache.items()}
-            output_pytorch = estimator(x, mask, mu, t, spks, cond, **{k: v.clone() for k, v in cache.items()})
-            ort_inputs = {
-                'x': x.cpu().numpy(),
-                'mask': mask.cpu().numpy(),
-                'mu': mu.cpu().numpy(),
-                't': t.cpu().numpy(),
-                'spks': spks.cpu().numpy(),
-                'cond': cond.cpu().numpy(),
-            }
-            output_onnx = estimator_onnx.run(None, {**ort_inputs, **{k: v.clone().cpu().numpy() for k, v in cache.items()}})
-            if iter == 0:
-                # NOTE why can not pass first iteration check?
-                continue
-            for i, j in zip(output_pytorch, output_onnx):
-                torch.testing.assert_allclose(i, torch.from_numpy(j).to(device), rtol=1e-2, atol=1e-4)
-        logging.info('successfully export estimator')
+    # 1. export flow decoder estimator
+    estimator = model.model.flow.decoder.estimator
+    estimator.eval()
+
+    device = model.model.device
+    batch_size, seq_len = 2, 256
+    out_channels = model.model.flow.decoder.estimator.out_channels
+    x, mask, mu, t, spks, cond = get_dummy_input(batch_size, seq_len, out_channels, device)
+    torch.onnx.export(
+        estimator,
+        (x, mask, mu, t, spks, cond),
+        '{}/flow.decoder.estimator.fp32.onnx'.format(args.model_dir),
+        export_params=True,
+        opset_version=18,
+        do_constant_folding=True,
+        input_names=['x', 'mask', 'mu', 't', 'spks', 'cond'],
+        output_names=['estimator_out'],
+        dynamic_axes={
+            'x': {2: 'seq_len'},
+            'mask': {2: 'seq_len'},
+            'mu': {2: 'seq_len'},
+            'cond': {2: 'seq_len'},
+            'estimator_out': {2: 'seq_len'},
+        }
+    )
+
+    # 2. test computation consistency
+    option = onnxruntime.SessionOptions()
+    option.graph_optimization_level = onnxruntime.GraphOptimizationLevel.ORT_ENABLE_ALL
+    option.intra_op_num_threads = 1
+    providers = ['CUDAExecutionProvider' if torch.cuda.is_available() else 'CPUExecutionProvider']
+    estimator_onnx = onnxruntime.InferenceSession('{}/flow.decoder.estimator.fp32.onnx'.format(args.model_dir),
+                                                  sess_options=option, providers=providers)
+
+    for _ in tqdm(range(10)):
+        x, mask, mu, t, spks, cond = get_dummy_input(batch_size, random.randint(16, 512), out_channels, device)
+        output_pytorch = estimator(x, mask, mu, t, spks, cond)
+        ort_inputs = {
+            'x': x.cpu().numpy(),
+            'mask': mask.cpu().numpy(),
+            'mu': mu.cpu().numpy(),
+            't': t.cpu().numpy(),
+            'spks': spks.cpu().numpy(),
+            'cond': cond.cpu().numpy()
+        }
+        output_onnx = estimator_onnx.run(None, ort_inputs)[0]
+        torch.testing.assert_allclose(output_pytorch, torch.from_numpy(output_onnx).to(device), rtol=1e-2, atol=1e-4)
+    logging.info('successfully export estimator')
 
 
 if __name__ == "__main__":

cosyvoice/bin/train_dpo.py

@@ -0,0 +1,187 @@
+# Copyright (c) 2024 Alibaba Inc (authors: Xiang Lyu)
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#   http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import print_function
+import argparse
+import datetime
+import logging
+logging.getLogger('matplotlib').setLevel(logging.WARNING)
+from copy import deepcopy
+import os
+import torch
+import torch.distributed as dist
+import deepspeed
+
+from hyperpyyaml import load_hyperpyyaml
+
+from torch.distributed.elastic.multiprocessing.errors import record
+
+from cosyvoice.utils.executor_dpo import Executor
+from cosyvoice.utils.train_utils_dpo import (
+    init_distributed,
+    init_dataset_and_dataloader,
+    init_optimizer_and_scheduler,
+    init_summarywriter, save_model,
+    wrap_cuda_model, check_modify_and_save_config)
+
+
+def get_args():
+    parser = argparse.ArgumentParser(description='training your network')
+    parser.add_argument('--train_engine',
+                        default='torch_ddp',
+                        choices=['torch_ddp', 'deepspeed'],
+                        help='Engine for paralleled training')
+    parser.add_argument('--model', required=True, help='model which will be trained')
+    parser.add_argument('--config', required=True, help='config file')
+    parser.add_argument('--train_data', required=True, help='train data file')
+    parser.add_argument('--cv_data', required=True, help='cv data file')
+    parser.add_argument('--checkpoint', help='checkpoint model')
+    parser.add_argument('--model_dir', required=True, help='save model dir')
+    parser.add_argument('--tensorboard_dir',
+                        default='tensorboard',
+                        help='tensorboard log dir')
+    parser.add_argument('--ddp.dist_backend',
+                        dest='dist_backend',
+                        default='nccl',
+                        choices=['nccl', 'gloo'],
+                        help='distributed backend')
+    parser.add_argument('--num_workers',
+                        default=0,
+                        type=int,
+                        help='num of subprocess workers for reading')
+    parser.add_argument('--prefetch',
+                        default=100,
+                        type=int,
+                        help='prefetch number')
+    parser.add_argument('--pin_memory',
+                        action='store_true',
+                        default=False,
+                        help='Use pinned memory buffers used for reading')
+    parser.add_argument('--use_amp',
+                        action='store_true',
+                        default=False,
+                        help='Use automatic mixed precision training')
+    parser.add_argument('--deepspeed.save_states',
+                        dest='save_states',
+                        default='model_only',
+                        choices=['model_only', 'model+optimizer'],
+                        help='save model/optimizer states')
+    parser.add_argument('--timeout',
+                        default=60,
+                        type=int,
+                        help='timeout (in seconds) of cosyvoice_join.')
+    parser.add_argument('--dpo',
+                        action='store_true',
+                        default=False,
+                        help='Use Direct Preference Optimization')
+    parser.add_argument('--beta',
+                        default=0.01,
+                        type=float,
+                        help='beta of dpo training')
+    parser = deepspeed.add_config_arguments(parser)
+    args = parser.parse_args()
+    return args
+
+
+@record
+def main():
+    args = get_args()
+    logging.basicConfig(level=logging.DEBUG,
+                        format='%(asctime)s %(levelname)s %(message)s')
+    # gan train has some special initialization logic
+    gan = True if args.model == 'hifigan' else False
+
+    override_dict = {k: None for k in ['llm', 'flow', 'hift', 'hifigan'] if k != args.model}
+    if gan is True:
+        override_dict.pop('hift')
+    with open(args.config, 'r') as f:
+        configs = load_hyperpyyaml(f, overrides=override_dict)
+    if gan is True:
+        configs['train_conf'] = configs['train_conf_gan']
+    configs['train_conf'].update(vars(args))
+
+    # Init env for ddp
+    init_distributed(args)
+
+    # Get dataset & dataloader
+    train_dataset, cv_dataset, train_data_loader, cv_data_loader = \
+        init_dataset_and_dataloader(args, configs, gan)
+
+    # Do some sanity checks and save config to arsg.model_dir
+    configs = check_modify_and_save_config(args, configs)
+
+    # Tensorboard summary
+    writer = init_summarywriter(args)
+
+    # load checkpoint
+    model = configs[args.model]
+    ref_model = None
+    if args.dpo:
+        ref_model = deepcopy(model)
+    start_step, start_epoch = 0, -1
+    if args.checkpoint is not None:
+        if os.path.exists(args.checkpoint):
+            state_dict = torch.load(args.checkpoint, map_location='cpu')
+            model.load_state_dict(state_dict, strict=False)
+            if args.dpo:
+                ref_model.load_state_dict(state_dict, strict=False)
+            if 'step' in state_dict:
+                start_step = state_dict['step']
+            if 'epoch' in state_dict:
+                start_epoch = state_dict['epoch']
+        else:
+            logging.warning('checkpoint {} do not exsist!'.format(args.checkpoint))
+
+    # Dispatch model from cpu to gpu
+    model = wrap_cuda_model(args, model)
+    if args.dpo:
+        ref_model = wrap_cuda_model(args, ref_model)
+
+    # Get optimizer & scheduler
+    model, optimizer, scheduler, optimizer_d, scheduler_d = init_optimizer_and_scheduler(args, configs, model, gan)
+    if args.dpo:
+        ref_model, _, _, _, _ = init_optimizer_and_scheduler(args, configs, ref_model, gan)
+    scheduler.set_step(start_step)
+    if scheduler_d is not None:
+        scheduler_d.set_step(start_step)
+
+    # Save init checkpoints
+    info_dict = deepcopy(configs['train_conf'])
+    info_dict['step'] = start_step
+    info_dict['epoch'] = start_epoch
+    save_model(model, 'init', info_dict)
+
+    # Get executor
+    executor = Executor(gan=gan, dpo=args.dpo, beta=args.beta)
+    executor.step = start_step
+
+    # Init scaler, used for pytorch amp mixed precision training
+    scaler = torch.cuda.amp.GradScaler() if args.use_amp else None
+    print('start step {} start epoch {}'.format(start_step, start_epoch))
+    # Start training loop
+    for epoch in range(start_epoch + 1, info_dict['max_epoch']):
+        executor.epoch = epoch
+        train_dataset.set_epoch(epoch)
+        dist.barrier()
+        group_join = dist.new_group(backend="gloo", timeout=datetime.timedelta(seconds=args.timeout))
+        if gan is True:
+            executor.train_one_epoc_gan(model, optimizer, scheduler, optimizer_d, scheduler_d, train_data_loader, cv_data_loader,
+                                        writer, info_dict, scaler, group_join)
+        else:
+            executor.train_one_epoc(model, optimizer, scheduler, train_data_loader, cv_data_loader, writer, info_dict, scaler, group_join, ref_model)
+        dist.destroy_process_group(group_join)
+
+
+if __name__ == '__main__':
+    main()

cosyvoice/cli/cosyvoice.py

@@ -140,7 +140,7 @@ class CosyVoice:
 
 class CosyVoice2(CosyVoice):
 
-    def __init__(self, model_dir, load_jit=False, load_trt=False, fp16=False, use_flow_cache=False):
+    def __init__(self, model_dir, load_jit=False, load_trt=False, fp16=False, trt_concurrent=1):
         self.instruct = True if '-Instruct' in model_dir else False
         self.model_dir = model_dir
         self.fp16 = fp16
@@ -162,9 +162,9 @@ class CosyVoice2(CosyVoice):
         if torch.cuda.is_available() is False and (load_jit is True or load_trt is True or fp16 is True):
             load_jit, load_trt, fp16 = False, False, False
             logging.warning('no cuda device, set load_jit/load_trt/fp16 to False')
-        self.model = CosyVoice2Model(configs['llm'], configs['flow'], configs['hift'], fp16, use_flow_cache)
+        self.model = CosyVoice2Model(configs['llm'], configs['flow'], configs['hift'], fp16, trt_concurrent)
         self.model.load('{}/llm.pt'.format(model_dir),
-                        '{}/flow.pt'.format(model_dir) if use_flow_cache is False else '{}/flow.cache.pt'.format(model_dir),
+                        '{}/flow.pt'.format(model_dir),
                         '{}/hift.pt'.format(model_dir))
         if load_jit:
             self.model.load_jit('{}/flow.encoder.{}.zip'.format(model_dir, 'fp16' if self.fp16 is True else 'fp32'))

cosyvoice/cli/model.py

@@ -1,4 +1,5 @@
 # Copyright (c) 2024 Alibaba Inc (authors: Xiang Lyu)
+#               2025 Alibaba Inc (authors: Xiang Lyu, Bofan Zhou)
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -13,6 +14,7 @@
 # limitations under the License.
 import os
 from typing import Generator
+import queue
 import torch
 import numpy as np
 import threading
@@ -22,6 +24,7 @@ from contextlib import nullcontext
 import uuid
 from cosyvoice.utils.common import fade_in_out
 from cosyvoice.utils.file_utils import convert_onnx_to_trt
+from cosyvoice.utils.common import TrtContextWrapper
 
 
 class CosyVoiceModel:
@@ -30,12 +33,14 @@ class CosyVoiceModel:
                  llm: torch.nn.Module,
                  flow: torch.nn.Module,
                  hift: torch.nn.Module,
-                 fp16: bool = False):
+                 fp16: bool = False,
+                 trt_concurrent: int = 1):
         self.device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
         self.llm = llm
         self.flow = flow
         self.hift = hift
         self.fp16 = fp16
+        self.trt_concurrent = trt_concurrent
         if self.fp16 is True:
             self.llm.half()
             self.flow.half()
@@ -82,20 +87,18 @@ class CosyVoiceModel:
 
     def load_trt(self, flow_decoder_estimator_model, flow_decoder_onnx_model, fp16):
         assert torch.cuda.is_available(), 'tensorrt only supports gpu!'
-        if not os.path.exists(flow_decoder_estimator_model):
+        if not os.path.exists(flow_decoder_estimator_model) or os.path.getsize(flow_decoder_estimator_model) == 0:
             convert_onnx_to_trt(flow_decoder_estimator_model, self.get_trt_kwargs(), flow_decoder_onnx_model, fp16)
-        if os.path.getsize(flow_decoder_estimator_model) == 0:
-            raise ValueError('{} is empty file, delete it and export again!'.format(flow_decoder_estimator_model))
         del self.flow.decoder.estimator
         import tensorrt as trt
         with open(flow_decoder_estimator_model, 'rb') as f:
-            self.flow.decoder.estimator_engine = trt.Runtime(trt.Logger(trt.Logger.INFO)).deserialize_cuda_engine(f.read())
-        assert self.flow.decoder.estimator_engine is not None, 'failed to load trt {}'.format(flow_decoder_estimator_model)
-        self.flow.decoder.estimator = self.flow.decoder.estimator_engine.create_execution_context()
+            estimator_engine = trt.Runtime(trt.Logger(trt.Logger.INFO)).deserialize_cuda_engine(f.read())
+        assert estimator_engine is not None, 'failed to load trt {}'.format(flow_decoder_estimator_model)
+        self.flow.decoder.estimator = TrtContextWrapper(estimator_engine, trt_concurrent=self.trt_concurrent)
 
     def get_trt_kwargs(self):
         min_shape = [(2, 80, 4), (2, 1, 4), (2, 80, 4), (2, 80, 4)]
-        opt_shape = [(2, 80, 200), (2, 1, 200), (2, 80, 200), (2, 80, 200)]
+        opt_shape = [(2, 80, 500), (2, 1, 500), (2, 80, 500), (2, 80, 500)]
         max_shape = [(2, 80, 3000), (2, 1, 3000), (2, 80, 3000), (2, 80, 3000)]
         input_names = ["x", "mask", "mu", "cond"]
         return {'min_shape': min_shape, 'opt_shape': opt_shape, 'max_shape': max_shape, 'input_names': input_names}
@@ -231,7 +234,9 @@ class CosyVoiceModel:
             self.mel_overlap_dict.pop(this_uuid)
             self.hift_cache_dict.pop(this_uuid)
             self.flow_cache_dict.pop(this_uuid)
-        torch.cuda.empty_cache()
+        if torch.cuda.is_available():
+            torch.cuda.empty_cache()
+            torch.cuda.current_stream().synchronize()
 
 
 class CosyVoice2Model(CosyVoiceModel):
@@ -241,19 +246,21 @@ class CosyVoice2Model(CosyVoiceModel):
                  flow: torch.nn.Module,
                  hift: torch.nn.Module,
                  fp16: bool = False,
-                 use_flow_cache: bool = False):
+                 trt_concurrent: int = 1):
         self.device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
         self.llm = llm
         self.flow = flow
+        # NOTE default setting for jit/onnx export, you can set to False when using pytorch inference
+        self.flow.encoder.streaming = True
+        self.flow.decoder.estimator.streaming = True
         self.hift = hift
         self.fp16 = fp16
-        self.use_flow_cache = use_flow_cache
+        self.trt_concurrent = trt_concurrent
         if self.fp16 is True:
             self.llm.half()
             self.flow.half()
-        # stream related params, check examples/libritts/cosyvoice2/conf/cosyvoice2.yaml
+        # NOTE must matching training static_chunk_size
         self.token_hop_len = 25
-        self.flow_decoder_required_cache_size = 0 if use_flow_cache is False else 1 * self.token_hop_len * self.flow.token_mel_ratio
         # hift cache
         self.mel_cache_len = 8
         self.source_cache_len = int(self.mel_cache_len * 480)
@@ -261,59 +268,31 @@ class CosyVoice2Model(CosyVoiceModel):
         self.speech_window = np.hamming(2 * self.source_cache_len)
         # rtf and decoding related
         self.llm_context = torch.cuda.stream(torch.cuda.Stream(self.device)) if torch.cuda.is_available() else nullcontext()
+        self.trt_context_pool = queue.Queue(maxsize=trt_concurrent)
+        for _ in range(trt_concurrent):
+            self.trt_context_pool.put(torch.cuda.stream(torch.cuda.Stream(self.device)) if torch.cuda.is_available() else nullcontext())
         self.lock = threading.Lock()
         # dict used to store session related variable
         self.tts_speech_token_dict = {}
         self.llm_end_dict = {}
-        self.flow_cache_dict = {}
         self.hift_cache_dict = {}
-
-    def init_flow_cache(self):
-        encoder_cache = {'offset': 0,
-                         'pre_lookahead_layer_conv2_cache': torch.zeros(1, 512, 2).to(self.device),
-                         'encoders_kv_cache': torch.zeros(6, 1, 8, 0, 64 * 2).to(self.device),
-                         'upsample_offset': 0,
-                         'upsample_conv_cache': torch.zeros(1, 512, 4).to(self.device),
-                         'upsample_kv_cache': torch.zeros(4, 1, 8, 0, 64 * 2).to(self.device)}
-        decoder_cache = {'offset': 0,
-                         'down_blocks_conv_cache': torch.zeros(10, 1, 2, 832, 2).to(self.device),
-                         'down_blocks_kv_cache': torch.zeros(10, 1, 4, 2, self.flow_decoder_required_cache_size, 512, 2).to(self.device),
-                         'mid_blocks_conv_cache': torch.zeros(10, 12, 2, 512, 2).to(self.device),
-                         'mid_blocks_kv_cache': torch.zeros(10, 12, 4, 2, self.flow_decoder_required_cache_size, 512, 2).to(self.device),
-                         'up_blocks_conv_cache': torch.zeros(10, 1, 2, 1024, 2).to(self.device),
-                         'up_blocks_kv_cache': torch.zeros(10, 1, 4, 2, self.flow_decoder_required_cache_size, 512, 2).to(self.device),
-                         'final_blocks_conv_cache': torch.zeros(10, 2, 256, 2).to(self.device)}
-        if self.fp16 is True:
-            for cache in [encoder_cache, decoder_cache]:
-                for k, v in cache.items():
-                    if isinstance(v, torch.Tensor):
-                        cache[k] = v.half()
-        cache = {'encoder_cache': encoder_cache, 'decoder_cache': decoder_cache}
-        return cache
+        self.trt_context_dict = {}
 
     def load_jit(self, flow_encoder_model):
         flow_encoder = torch.jit.load(flow_encoder_model, map_location=self.device)
         self.flow.encoder = flow_encoder
 
-    def get_trt_kwargs(self):
-        min_shape = [(2, 80, 4), (2, 1, 4), (2, 80, 4), (2, 80, 4), (1, 4, 2, 0, 512, 2), (12, 4, 2, 0, 512, 2), (1, 4, 2, 0, 512, 2)]
-        opt_shape = [(2, 80, 200), (2, 1, 200), (2, 80, 200), (2, 80, 200), (1, 4, 2, 100, 512, 2), (12, 4, 2, 100, 512, 2), (1, 4, 2, 100, 512, 2)]
-        max_shape = [(2, 80, 1500), (2, 1, 1500), (2, 80, 1500), (2, 80, 1500), (1, 4, 2, 200, 512, 2), (12, 4, 2, 200, 512, 2), (1, 4, 2, 200, 512, 2)]
-        input_names = ["x", "mask", "mu", "cond", 'down_blocks_kv_cache', 'mid_blocks_kv_cache', 'up_blocks_kv_cache']
-        assert self.use_flow_cache is True, "get_trt_kwargs is set for flow cache mode. If you want to use trt with use_flow_cache=False, please set higher max_shape"
-        return {'min_shape': min_shape, 'opt_shape': opt_shape, 'max_shape': max_shape, 'input_names': input_names}
-
-    def token2wav(self, token, prompt_token, prompt_feat, embedding, uuid, finalize=False, speed=1.0):
-        with torch.cuda.amp.autocast(self.fp16):
-            tts_mel, self.flow_cache_dict[uuid] = self.flow.inference(token=token.to(self.device),
-                                                                      token_len=torch.tensor([token.shape[1]], dtype=torch.int32).to(self.device),
-                                                                      prompt_token=prompt_token.to(self.device),
-                                                                      prompt_token_len=torch.tensor([prompt_token.shape[1]], dtype=torch.int32).to(self.device),
-                                                                      prompt_feat=prompt_feat.to(self.device),
-                                                                      prompt_feat_len=torch.tensor([prompt_feat.shape[1]], dtype=torch.int32).to(self.device),
-                                                                      embedding=embedding.to(self.device),
-                                                                      cache=self.flow_cache_dict[uuid],
-                                                                      finalize=finalize)
+    def token2wav(self, token, prompt_token, prompt_feat, embedding, token_offset, uuid, finalize=False, speed=1.0):
+        with torch.cuda.amp.autocast(self.fp16), self.trt_context_dict[uuid]:
+            tts_mel, _ = self.flow.inference(token=token.to(self.device),
+                                             token_len=torch.tensor([token.shape[1]], dtype=torch.int32).to(self.device),
+                                             prompt_token=prompt_token.to(self.device),
+                                             prompt_token_len=torch.tensor([prompt_token.shape[1]], dtype=torch.int32).to(self.device),
+                                             prompt_feat=prompt_feat.to(self.device),
+                                             prompt_feat_len=torch.tensor([prompt_feat.shape[1]], dtype=torch.int32).to(self.device),
+                                             embedding=embedding.to(self.device),
+                                             finalize=finalize)
+        tts_mel = tts_mel[:, :, token_offset * self.flow.token_mel_ratio:]
         # append hift cache
         if self.hift_cache_dict[uuid] is not None:
             hift_cache_mel, hift_cache_source = self.hift_cache_dict[uuid]['mel'], self.hift_cache_dict[uuid]['source']
@@ -348,34 +327,30 @@ class CosyVoice2Model(CosyVoiceModel):
         with self.lock:
             self.tts_speech_token_dict[this_uuid], self.llm_end_dict[this_uuid] = [], False
             self.hift_cache_dict[this_uuid] = None
-            self.flow_cache_dict[this_uuid] = self.init_flow_cache()
+            self.trt_context_dict[this_uuid] = self.trt_context_pool.get()
         if source_speech_token.shape[1] == 0:
             p = threading.Thread(target=self.llm_job, args=(text, prompt_text, llm_prompt_speech_token, llm_embedding, this_uuid))
         else:
             p = threading.Thread(target=self.vc_job, args=(source_speech_token, this_uuid))
         p.start()
         if stream is True:
-            assert self.use_flow_cache is True, "set use_flow_cache=True if you want to use stream inference to avoid OOM"
-            # NOTE in cache mode, trim flow_prompt to same size as flow_decoder_required_cache_size
-            flow_prompt_speech_token = flow_prompt_speech_token[:, -int(self.flow_decoder_required_cache_size / self.flow.token_mel_ratio):]
-            prompt_speech_feat = prompt_speech_feat[:, -self.flow_decoder_required_cache_size:]
+            token_offset = 0
+            prompt_token_pad = int(np.ceil(flow_prompt_speech_token.shape[1] / self.token_hop_len) * self.token_hop_len - flow_prompt_speech_token.shape[1])
             while True:
                 time.sleep(0.1)
-                if len(self.tts_speech_token_dict[this_uuid]) >= self.token_hop_len + self.flow.pre_lookahead_len:
-                    this_tts_speech_token = torch.tensor(self.tts_speech_token_dict[this_uuid][:self.token_hop_len + self.flow.pre_lookahead_len]).unsqueeze(dim=0)
+                this_token_hop_len = self.token_hop_len + prompt_token_pad if token_offset == 0 else self.token_hop_len
+                if len(self.tts_speech_token_dict[this_uuid]) - token_offset >= this_token_hop_len + self.flow.pre_lookahead_len:
+                    this_tts_speech_token = torch.tensor(self.tts_speech_token_dict[this_uuid][:token_offset + this_token_hop_len + self.flow.pre_lookahead_len]).unsqueeze(dim=0)
                     this_tts_speech = self.token2wav(token=this_tts_speech_token,
                                                      prompt_token=flow_prompt_speech_token,
                                                      prompt_feat=prompt_speech_feat,
                                                      embedding=flow_embedding,
+                                                     token_offset=token_offset,
                                                      uuid=this_uuid,
                                                      finalize=False)
-                    # NOTE in cache inference mode, we only use flow_prompt_speech_token/prompt_speech_feat in first chunk
-                    flow_prompt_speech_token = torch.zeros(1, 0, dtype=torch.int32).to(self.device)
-                    prompt_speech_feat = torch.zeros(1, 0, 80).to(self.device)
+                    token_offset += this_token_hop_len
                     yield {'tts_speech': this_tts_speech.cpu()}
-                    with self.lock:
-                        self.tts_speech_token_dict[this_uuid] = self.tts_speech_token_dict[this_uuid][self.token_hop_len:]
-                if self.llm_end_dict[this_uuid] is True and len(self.tts_speech_token_dict[this_uuid]) < self.token_hop_len + self.flow.pre_lookahead_len:
+                if self.llm_end_dict[this_uuid] is True and len(self.tts_speech_token_dict[this_uuid]) - token_offset < this_token_hop_len + self.flow.pre_lookahead_len:
                     break
             p.join()
             # deal with remain tokens, make sure inference remain token len equals token_hop_len when cache_speech is not None
@@ -384,18 +359,19 @@ class CosyVoice2Model(CosyVoiceModel):
                                              prompt_token=flow_prompt_speech_token,
                                              prompt_feat=prompt_speech_feat,
                                              embedding=flow_embedding,
+                                             token_offset=token_offset,
                                              uuid=this_uuid,
                                              finalize=True)
             yield {'tts_speech': this_tts_speech.cpu()}
         else:
             # deal with all tokens
-            assert self.use_flow_cache is False, "set use_flow_cache=False for nonstream inference"
             p.join()
             this_tts_speech_token = torch.tensor(self.tts_speech_token_dict[this_uuid]).unsqueeze(dim=0)
             this_tts_speech = self.token2wav(token=this_tts_speech_token,
                                              prompt_token=flow_prompt_speech_token,
                                              prompt_feat=prompt_speech_feat,
                                              embedding=flow_embedding,
+                                             token_offset=0,
                                              uuid=this_uuid,
                                              finalize=True,
                                              speed=speed)
@@ -404,5 +380,8 @@ class CosyVoice2Model(CosyVoiceModel):
             self.tts_speech_token_dict.pop(this_uuid)
             self.llm_end_dict.pop(this_uuid)
             self.hift_cache_dict.pop(this_uuid)
-            self.flow_cache_dict.pop(this_uuid)
-        torch.cuda.empty_cache()
+            self.trt_context_pool.put(self.trt_context_dict[this_uuid])
+            self.trt_context_dict.pop(this_uuid)
+        if torch.cuda.is_available():
+            torch.cuda.empty_cache()
+            torch.cuda.current_stream().synchronize()

cosyvoice/dataset/processor_dpo.py

@@ -0,0 +1,443 @@
+# Copyright (c) 2024 Alibaba Inc (authors: Xiang Lyu)
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+import logging
+import random
+
+import pyarrow.parquet as pq
+from io import BytesIO
+import torch
+import torchaudio
+from torch.nn.utils.rnn import pad_sequence
+import torch.nn.functional as F
+import pyworld as pw
+
+
+AUDIO_FORMAT_SETS = {'flac', 'mp3', 'm4a', 'ogg', 'opus', 'wav', 'wma'}
+
+
+def parquet_opener(data, mode='train', tts_data={}):
+    """ Give url or local file, return file descriptor
+        Inplace operation.
+
+        Args:
+            data(Iterable[str]): url or local file list
+
+        Returns:
+            Iterable[{src, stream}]
+    """
+    for sample in data:
+        assert 'src' in sample
+        url = sample['src']
+        try:
+            for df in pq.ParquetFile(url).iter_batches(batch_size=64):
+                df = df.to_pandas()
+                for i in range(len(df)):
+                    if mode == 'inference' and df.loc[i, 'utt'] not in tts_data:
+                        continue
+                    sample.update(dict(df.loc[i]))
+                    if mode == 'train':
+                        # NOTE do not return sample directly, must initialize a new dict
+                        yield {**sample}
+                    else:
+                        for index, text in enumerate(tts_data[df.loc[i, 'utt']]):
+                            yield {**sample, 'tts_index': index, 'tts_text': text}
+        except Exception as ex:
+            logging.warning('Failed to open {}, ex info {}'.format(url, ex))
+
+
+def filter(data,
+           max_length=10240,
+           min_length=10,
+           token_max_length=200,
+           token_min_length=1,
+           min_output_input_ratio=0.0005,
+           max_output_input_ratio=1,
+           mode='train'):
+    """ Filter sample according to feature and label length
+        Inplace operation.
+
+        Args::
+            data: Iterable[{key, wav, label, sample_rate}]
+            max_length: drop utterance which is greater than max_length(10ms)
+            min_length: drop utterance which is less than min_length(10ms)
+            token_max_length: drop utterance which is greater than
+                token_max_length, especially when use char unit for
+                english modeling
+            token_min_length: drop utterance which is
+                less than token_max_length
+            min_output_input_ratio: minimal ration of
+                token_length / feats_length(10ms)
+            max_output_input_ratio: maximum ration of
+                token_length / feats_length(10ms)
+
+        Returns:
+            Iterable[{key, wav, label, sample_rate}]
+    """
+    for sample in data:
+        sample['speech'], sample['sample_rate'] = torchaudio.load(BytesIO(sample['audio_data']))
+        sample['speech'] = sample['speech'].mean(dim=0, keepdim=True)
+        del sample['audio_data']
+        # sample['wav'] is torch.Tensor, we have 100 frames every second
+        num_frames = sample['speech'].size(1) / sample['sample_rate'] * 100
+        if num_frames < min_length:
+            continue
+        if num_frames > max_length:
+            continue
+        if len(sample['text_token']) < token_min_length:
+            continue
+        if len(sample['text_token']) > token_max_length:
+            continue
+        if len(sample['speech_token']) == 0:
+            continue
+        if num_frames != 0:
+            if len(sample['text_token']) / num_frames < min_output_input_ratio:
+                continue
+            if len(sample['text_token']) / num_frames > max_output_input_ratio:
+                continue
+        yield sample
+
+
+def resample(data, resample_rate=22050, min_sample_rate=16000, mode='train'):
+    """ Resample data.
+        Inplace operation.
+
+        Args:
+            data: Iterable[{key, wav, label, sample_rate}]
+            resample_rate: target resample rate
+
+        Returns:
+            Iterable[{key, wav, label, sample_rate}]
+    """
+    for sample in data:
+        assert 'sample_rate' in sample
+        assert 'speech' in sample
+        sample_rate = sample['sample_rate']
+        waveform = sample['speech']
+        if sample_rate != resample_rate:
+            if sample_rate < min_sample_rate:
+                continue
+            sample['sample_rate'] = resample_rate
+            sample['speech'] = torchaudio.transforms.Resample(
+                orig_freq=sample_rate, new_freq=resample_rate)(waveform)
+        max_val = sample['speech'].abs().max()
+        if max_val > 1:
+            sample['speech'] /= max_val
+        yield sample
+
+
+def truncate(data, truncate_length=24576, mode='train'):
+    """ Truncate data.
+
+        Args:
+            data: Iterable[{key, wav, label, sample_rate}]
+            truncate_length: truncate length
+
+        Returns:
+            Iterable[{key, wav, label, sample_rate}]
+    """
+    for sample in data:
+        waveform = sample['speech']
+        if waveform.shape[1] > truncate_length:
+            start = random.randint(0, waveform.shape[1] - truncate_length)
+            waveform = waveform[:, start: start + truncate_length]
+        else:
+            waveform = torch.concat([waveform, torch.zeros(1, truncate_length - waveform.shape[1])], dim=1)
+        sample['speech'] = waveform
+        yield sample
+
+
+def compute_fbank(data,
+                  feat_extractor,
+                  mode='train'):
+    """ Extract fbank
+
+        Args:
+            data: Iterable[{key, wav, label, sample_rate}]
+
+        Returns:
+            Iterable[{key, feat, label}]
+    """
+    for sample in data:
+        assert 'sample_rate' in sample
+        assert 'speech' in sample
+        assert 'utt' in sample
+        assert 'text_token' in sample
+        waveform = sample['speech']
+        mat = feat_extractor(waveform).squeeze(dim=0).transpose(0, 1)
+        sample['speech_feat'] = mat
+        yield sample
+
+
+def compute_f0(data, sample_rate, hop_size, mode='train'):
+    """ Extract f0
+
+        Args:
+            data: Iterable[{key, wav, label, sample_rate}]
+
+        Returns:
+            Iterable[{key, feat, label}]
+    """
+    frame_period = hop_size * 1000 / sample_rate
+    for sample in data:
+        assert 'sample_rate' in sample
+        assert 'speech' in sample
+        assert 'utt' in sample
+        assert 'text_token' in sample
+        waveform = sample['speech']
+        _f0, t = pw.harvest(waveform.squeeze(dim=0).numpy().astype('double'), sample_rate, frame_period=frame_period)
+        if sum(_f0 != 0) < 5: # this happens when the algorithm fails
+            _f0, t = pw.dio(waveform.squeeze(dim=0).numpy().astype('double'), sample_rate, frame_period=frame_period) # if harvest fails, try dio
+        f0 = pw.stonemask(waveform.squeeze(dim=0).numpy().astype('double'), _f0, t, sample_rate)
+        f0 = F.interpolate(torch.from_numpy(f0).view(1, 1, -1), size=sample['speech_feat'].shape[0], mode='linear').view(-1)
+        sample['pitch_feat'] = f0
+        yield sample
+
+
+def parse_embedding(data, normalize, mode='train'):
+    """ Parse utt_embedding/spk_embedding
+
+        Args:
+            data: Iterable[{key, wav, label, sample_rate}]
+
+        Returns:
+            Iterable[{key, feat, label}]
+    """
+    for sample in data:
+        sample['utt_embedding'] = torch.tensor(sample['utt_embedding'], dtype=torch.float32)
+        sample['spk_embedding'] = torch.tensor(sample['spk_embedding'], dtype=torch.float32)
+        if normalize:
+            sample['utt_embedding'] = F.normalize(sample['utt_embedding'], dim=0)
+            sample['spk_embedding'] = F.normalize(sample['spk_embedding'], dim=0)
+        yield sample
+
+
+def tokenize(data, get_tokenizer, allowed_special, mode='train'):
+    """ Decode text to chars or BPE
+        Inplace operation
+
+        Args:
+            data: Iterable[{key, wav, txt, sample_rate}]
+
+        Returns:
+            Iterable[{key, wav, txt, tokens, label, sample_rate}]
+    """
+    tokenizer = get_tokenizer()
+    for sample in data:
+        assert 'text' in sample
+        sample['text_token'] = tokenizer.encode(sample['text'], allowed_special=allowed_special)
+        if mode == 'inference':
+            sample['tts_text_token'] = tokenizer.encode(sample['tts_text'], allowed_special=allowed_special)
+        yield sample
+
+
+def shuffle(data, shuffle_size=10000, mode='train'):
+    """ Local shuffle the data
+
+        Args:
+            data: Iterable[{key, feat, label}]
+            shuffle_size: buffer size for shuffle
+
+        Returns:
+            Iterable[{key, feat, label}]
+    """
+    buf = []
+    for sample in data:
+        buf.append(sample)
+        if len(buf) >= shuffle_size:
+            random.shuffle(buf)
+            for x in buf:
+                yield x
+            buf = []
+    # The sample left over
+    random.shuffle(buf)
+    for x in buf:
+        yield x
+
+
+def sort(data, sort_size=500, mode='train'):
+    """ Sort the data by feature length.
+        Sort is used after shuffle and before batch, so we can group
+        utts with similar lengths into a batch, and `sort_size` should
+        be less than `shuffle_size`
+
+        Args:
+            data: Iterable[{key, feat, label}]
+            sort_size: buffer size for sort
+
+        Returns:
+            Iterable[{key, feat, label}]
+    """
+
+    buf = []
+    for sample in data:
+        buf.append(sample)
+        if len(buf) >= sort_size:
+            buf.sort(key=lambda x: x['speech_feat'].size(0))
+            for x in buf:
+                yield x
+            buf = []
+    # The sample left over
+    buf.sort(key=lambda x: x['speech_feat'].size(0))
+    for x in buf:
+        yield x
+
+
+def static_batch(data, batch_size=16):
+    """ Static batch the data by `batch_size`
+
+        Args:
+            data: Iterable[{key, feat, label}]
+            batch_size: batch size
+
+        Returns:
+            Iterable[List[{key, feat, label}]]
+    """
+    buf = []
+    for sample in data:
+        buf.append(sample)
+        if len(buf) >= batch_size:
+            yield buf
+            buf = []
+    if len(buf) > 0:
+        yield buf
+
+
+def dynamic_batch(data, max_frames_in_batch=12000, mode='train'):
+    """ Dynamic batch the data until the total frames in batch
+        reach `max_frames_in_batch`
+
+        Args:
+            data: Iterable[{key, feat, label}]
+            max_frames_in_batch: max_frames in one batch
+
+        Returns:
+            Iterable[List[{key, feat, label}]]
+    """
+    buf = []
+    longest_frames = 0
+    for sample in data:
+        assert 'speech_feat' in sample
+        assert isinstance(sample['speech_feat'], torch.Tensor)
+        new_sample_frames = sample['speech_feat'].size(0)
+        longest_frames = max(longest_frames, new_sample_frames)
+        frames_after_padding = longest_frames * (len(buf) + 1)
+        if frames_after_padding > max_frames_in_batch:
+            yield buf
+            buf = [sample]
+            longest_frames = new_sample_frames
+        else:
+            buf.append(sample)
+    if len(buf) > 0:
+        yield buf
+
+
+def batch(data, batch_type='static', batch_size=16, max_frames_in_batch=12000, mode='train'):
+    """ Wrapper for static/dynamic batch
+    """
+    if mode == 'inference':
+        return static_batch(data, 1)
+    else:
+        if batch_type == 'static':
+            return static_batch(data, batch_size)
+        elif batch_type == 'dynamic':
+            return dynamic_batch(data, max_frames_in_batch)
+        else:
+            logging.fatal('Unsupported batch type {}'.format(batch_type))
+
+
+def padding(data, use_spk_embedding, mode='train', gan=False, dpo=False):
+    """ Padding the data into training data
+
+        Args:
+            data: Iterable[List[{key, feat, label}]]
+
+        Returns:
+            Iterable[Tuple(keys, feats, labels, feats lengths, label lengths)]
+    """
+    for sample in data:
+        assert isinstance(sample, list)
+        speech_feat_len = torch.tensor([x['speech_feat'].size(1) for x in sample],
+                                       dtype=torch.int32)
+        order = torch.argsort(speech_feat_len, descending=True)
+
+        utts = [sample[i]['utt'] for i in order]
+        speech = [sample[i]['speech'].squeeze(dim=0) for i in order]
+        speech_len = torch.tensor([i.size(0) for i in speech], dtype=torch.int32)
+        speech = pad_sequence(speech, batch_first=True, padding_value=0)
+        speech_token = [torch.tensor(sample[i]['speech_token']) for i in order]
+        speech_token_len = torch.tensor([i.size(0) for i in speech_token], dtype=torch.int32)
+        speech_token = pad_sequence(speech_token,
+                                    batch_first=True,
+                                    padding_value=0)
+        speech_feat = [sample[i]['speech_feat'] for i in order]
+        speech_feat_len = torch.tensor([i.size(0) for i in speech_feat], dtype=torch.int32)
+        speech_feat = pad_sequence(speech_feat,
+                                   batch_first=True,
+                                   padding_value=0)
+        text = [sample[i]['text'] for i in order]
+        text_token = [torch.tensor(sample[i]['text_token']) for i in order]
+        text_token_len = torch.tensor([i.size(0) for i in text_token], dtype=torch.int32)
+        text_token = pad_sequence(text_token, batch_first=True, padding_value=0)
+        utt_embedding = torch.stack([sample[i]['utt_embedding'] for i in order], dim=0)
+        spk_embedding = torch.stack([sample[i]['spk_embedding'] for i in order], dim=0)
+        batch = {
+            "utts": utts,
+            "speech": speech,
+            "speech_len": speech_len,
+            "speech_token": speech_token,
+            "speech_token_len": speech_token_len,
+            "speech_feat": speech_feat,
+            "speech_feat_len": speech_feat_len,
+            "text": text,
+            "text_token": text_token,
+            "text_token_len": text_token_len,
+            "utt_embedding": utt_embedding,
+            "spk_embedding": spk_embedding,
+        }
+        if dpo:
+            reject_speech_token = [torch.tensor(sample[i]['reject_speech_token']) for i in order]
+            reject_speech_token_len = torch.tensor([i.size(0) for i in reject_speech_token], dtype=torch.int32)
+            reject_speech_token = pad_sequence(reject_speech_token,
+                                                batch_first=True,
+                                                padding_value=0)
+            batch['reject_speech_token'] = reject_speech_token
+            batch['reject_speech_token_len'] = reject_speech_token_len
+        if gan is True:
+            # in gan train, we need pitch_feat
+            pitch_feat = [sample[i]['pitch_feat'] for i in order]
+            pitch_feat_len = torch.tensor([i.size(0) for i in pitch_feat], dtype=torch.int32)
+            pitch_feat = pad_sequence(pitch_feat,
+                                      batch_first=True,
+                                      padding_value=0)
+            batch["pitch_feat"] = pitch_feat
+            batch["pitch_feat_len"] = pitch_feat_len
+        else:
+            # only gan train needs speech, delete it to save memory
+            del batch["speech"]
+            del batch["speech_len"]
+        if mode == 'inference':
+            tts_text = [sample[i]['tts_text'] for i in order]
+            tts_index = [sample[i]['tts_index'] for i in order]
+            tts_text_token = [torch.tensor(sample[i]['tts_text_token']) for i in order]
+            tts_text_token_len = torch.tensor([i.size(0) for i in tts_text_token], dtype=torch.int32)
+            tts_text_token = pad_sequence(tts_text_token, batch_first=True, padding_value=-1)
+            batch.update({'tts_text': tts_text,
+                          'tts_index': tts_index,
+                          'tts_text_token': tts_text_token,
+                          'tts_text_token_len': tts_text_token_len})
+        if use_spk_embedding is True:
+            batch["embedding"] = batch["spk_embedding"]
+        else:
+            batch["embedding"] = batch["utt_embedding"]
+        yield batch

cosyvoice/flow/decoder.py

@@ -11,16 +11,15 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-from typing import Tuple, Optional, Dict, Any
+from typing import Tuple
 import torch
 import torch.nn as nn
 import torch.nn.functional as F
 from einops import pack, rearrange, repeat
-from diffusers.models.attention_processor import Attention, AttnProcessor2_0, inspect, logger, deprecate
 from cosyvoice.utils.common import mask_to_bias
 from cosyvoice.utils.mask import add_optional_chunk_mask
 from matcha.models.components.decoder import SinusoidalPosEmb, Block1D, ResnetBlock1D, Downsample1D, TimestepEmbedding, Upsample1D
-from matcha.models.components.transformer import BasicTransformerBlock, maybe_allow_in_graph
+from matcha.models.components.transformer import BasicTransformerBlock
 
 
 class Transpose(torch.nn.Module):
@@ -29,7 +28,7 @@ class Transpose(torch.nn.Module):
         self.dim0 = dim0
         self.dim1 = dim1
 
-    def forward(self, x: torch.Tensor) -> Tuple[torch.Tensor]:
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
         x = torch.transpose(x, self.dim0, self.dim1)
         return x
 
@@ -57,15 +56,10 @@ class CausalConv1d(torch.nn.Conv1d):
         assert stride == 1
         self.causal_padding = kernel_size - 1
 
-    def forward(self, x: torch.Tensor, cache: torch.Tensor = torch.zeros(0, 0, 0)) -> Tuple[torch.Tensor, torch.Tensor]:
-        if cache.size(2) == 0:
-            x = F.pad(x, (self.causal_padding, 0), value=0.0)
-        else:
-            assert cache.size(2) == self.causal_padding
-            x = torch.concat([cache, x], dim=2)
-        cache = x[:, :, -self.causal_padding:]
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        x = F.pad(x, (self.causal_padding, 0), value=0.0)
         x = super(CausalConv1d, self).forward(x)
-        return x, cache
+        return x
 
 
 class CausalBlock1D(Block1D):
@@ -79,11 +73,9 @@ class CausalBlock1D(Block1D):
             nn.Mish(),
         )
 
-    def forward(self, x: torch.Tensor, mask: torch.Tensor, cache: torch.Tensor = torch.zeros(0, 0, 0)) -> Tuple[torch.Tensor, torch.Tensor]:
-        output, cache = self.block[0](x * mask, cache)
-        for i in range(1, len(self.block)):
-            output = self.block[i](output)
-        return output * mask, cache
+    def forward(self, x: torch.Tensor, mask: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor]:
+        output = self.block(x * mask)
+        return output * mask
 
 
 class CausalResnetBlock1D(ResnetBlock1D):
@@ -92,303 +84,6 @@ class CausalResnetBlock1D(ResnetBlock1D):
         self.block1 = CausalBlock1D(dim, dim_out)
         self.block2 = CausalBlock1D(dim_out, dim_out)
 
-    def forward(self, x: torch.Tensor, mask: torch.Tensor, time_emb: torch.Tensor,
-                block1_cache: torch.Tensor = torch.zeros(0, 0, 0), block2_cache: torch.Tensor = torch.zeros(0, 0, 0)
-                ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
-        h, block1_cache = self.block1(x, mask, block1_cache)
-        h += self.mlp(time_emb).unsqueeze(-1)
-        h, block2_cache = self.block2(h, mask, block2_cache)
-        output = h + self.res_conv(x * mask)
-        return output, block1_cache, block2_cache
-
-
-class CausalAttnProcessor2_0(AttnProcessor2_0):
-    r"""
-    Processor for implementing scaled dot-product attention (enabled by default if you're using PyTorch 2.0).
-    """
-
-    def __init__(self):
-        super(CausalAttnProcessor2_0, self).__init__()
-
-    def __call__(
-        self,
-        attn: Attention,
-        hidden_states: torch.FloatTensor,
-        encoder_hidden_states: Optional[torch.FloatTensor] = None,
-        attention_mask: Optional[torch.FloatTensor] = None,
-        temb: Optional[torch.FloatTensor] = None,
-        cache: torch.Tensor = torch.zeros(0, 0, 0, 0),
-        *args,
-        **kwargs,
-    ) -> Tuple[torch.FloatTensor, torch.Tensor]:
-        if len(args) > 0 or kwargs.get("scale", None) is not None:
-            deprecation_message = "The `scale` argument is deprecated and will be ignored. Please remove it, as passing it will raise an error in the future. \
-                `scale` should directly be passed while calling the underlying pipeline component i.e., via `cross_attention_kwargs`."
-            deprecate("scale", "1.0.0", deprecation_message)
-
-        residual = hidden_states
-        if attn.spatial_norm is not None:
-            hidden_states = attn.spatial_norm(hidden_states, temb)
-
-        input_ndim = hidden_states.ndim
-
-        if input_ndim == 4:
-            batch_size, channel, height, width = hidden_states.shape
-            hidden_states = hidden_states.view(batch_size, channel, height * width).transpose(1, 2)
-
-        batch_size, sequence_length, _ = (
-            hidden_states.shape if encoder_hidden_states is None else encoder_hidden_states.shape
-        )
-
-        if attention_mask is not None:
-            # NOTE do not use attn.prepare_attention_mask as we have already provided the correct attention_mask
-            # scaled_dot_product_attention expects attention_mask shape to be
-            # (batch, heads, source_length, target_length)
-            attention_mask = attention_mask.unsqueeze(dim=1).repeat(1, attn.heads, 1, 1)
-
-        if attn.group_norm is not None:
-            hidden_states = attn.group_norm(hidden_states.transpose(1, 2)).transpose(1, 2)
-
-        query = attn.to_q(hidden_states)
-
-        if encoder_hidden_states is None:
-            encoder_hidden_states = hidden_states
-        elif attn.norm_cross:
-            encoder_hidden_states = attn.norm_encoder_hidden_states(encoder_hidden_states)
-
-        key_cache = attn.to_k(encoder_hidden_states)
-        value_cache = attn.to_v(encoder_hidden_states)
-        # NOTE here we judge cache.size(0) instead of cache.size(1), because init_cache has size (2, 0, 512, 2)
-        if cache.size(0) != 0:
-            key = torch.concat([cache[:, :, :, 0], key_cache], dim=1)
-            value = torch.concat([cache[:, :, :, 1], value_cache], dim=1)
-        else:
-            key, value = key_cache, value_cache
-        cache = torch.stack([key_cache, value_cache], dim=3)
-
-        inner_dim = key.shape[-1]
-        head_dim = inner_dim // attn.heads
-
-        query = query.view(batch_size, -1, attn.heads, head_dim).transpose(1, 2)
-
-        key = key.view(batch_size, -1, attn.heads, head_dim).transpose(1, 2)
-        value = value.view(batch_size, -1, attn.heads, head_dim).transpose(1, 2)
-
-        # the output of sdp = (batch, num_heads, seq_len, head_dim)
-        # TODO: add support for attn.scale when we move to Torch 2.1
-        hidden_states = F.scaled_dot_product_attention(
-            query, key, value, attn_mask=attention_mask, dropout_p=0.0, is_causal=False
-        )
-
-        hidden_states = hidden_states.transpose(1, 2).reshape(batch_size, -1, attn.heads * head_dim)
-        hidden_states = hidden_states.to(query.dtype)
-
-        # linear proj
-        hidden_states = attn.to_out[0](hidden_states)
-        # dropout
-        hidden_states = attn.to_out[1](hidden_states)
-
-        if input_ndim == 4:
-            hidden_states = hidden_states.transpose(-1, -2).reshape(batch_size, channel, height, width)
-
-        if attn.residual_connection:
-            hidden_states = hidden_states + residual
-
-        hidden_states = hidden_states / attn.rescale_output_factor
-
-        return hidden_states, cache
-
-
-@maybe_allow_in_graph
-class CausalAttention(Attention):
-    def __init__(
-        self,
-        query_dim: int,
-        cross_attention_dim: Optional[int] = None,
-        heads: int = 8,
-        dim_head: int = 64,
-        dropout: float = 0.0,
-        bias: bool = False,
-        upcast_attention: bool = False,
-        upcast_softmax: bool = False,
-        cross_attention_norm: Optional[str] = None,
-        cross_attention_norm_num_groups: int = 32,
-        qk_norm: Optional[str] = None,
-        added_kv_proj_dim: Optional[int] = None,
-        norm_num_groups: Optional[int] = None,
-        spatial_norm_dim: Optional[int] = None,
-        out_bias: bool = True,
-        scale_qk: bool = True,
-        only_cross_attention: bool = False,
-        eps: float = 1e-5,
-        rescale_output_factor: float = 1.0,
-        residual_connection: bool = False,
-        _from_deprecated_attn_block: bool = False,
-        processor: Optional["AttnProcessor2_0"] = None,
-        out_dim: int = None,
-    ):
-        super(CausalAttention, self).__init__(query_dim, cross_attention_dim, heads, dim_head, dropout, bias, upcast_attention, upcast_softmax,
-                                              cross_attention_norm, cross_attention_norm_num_groups, qk_norm, added_kv_proj_dim, norm_num_groups,
-                                              spatial_norm_dim, out_bias, scale_qk, only_cross_attention, eps, rescale_output_factor, residual_connection,
-                                              _from_deprecated_attn_block, processor, out_dim)
-        processor = CausalAttnProcessor2_0()
-        self.set_processor(processor)
-
-    def forward(
-        self,
-        hidden_states: torch.FloatTensor,
-        encoder_hidden_states: Optional[torch.FloatTensor] = None,
-        attention_mask: Optional[torch.FloatTensor] = None,
-        cache: torch.Tensor = torch.zeros(0, 0, 0, 0),
-        **cross_attention_kwargs,
-    ) -> Tuple[torch.Tensor, torch.Tensor]:
-        r"""
-        The forward method of the `Attention` class.
-
-        Args:
-            hidden_states (`torch.Tensor`):
-                The hidden states of the query.
-            encoder_hidden_states (`torch.Tensor`, *optional*):
-                The hidden states of the encoder.
-            attention_mask (`torch.Tensor`, *optional*):
-                The attention mask to use. If `None`, no mask is applied.
-            **cross_attention_kwargs:
-                Additional keyword arguments to pass along to the cross attention.
-
-        Returns:
-            `torch.Tensor`: The output of the attention layer.
-        """
-        # The `Attention` class can call different attention processors / attention functions
-        # here we simply pass along all tensors to the selected processor class
-        # For standard processors that are defined here, `**cross_attention_kwargs` is empty
-
-        attn_parameters = set(inspect.signature(self.processor.__call__).parameters.keys())
-        unused_kwargs = [k for k, _ in cross_attention_kwargs.items() if k not in attn_parameters]
-        if len(unused_kwargs) > 0:
-            logger.warning(
-                f"cross_attention_kwargs {unused_kwargs} are not expected by {self.processor.__class__.__name__} and will be ignored."
-            )
-        cross_attention_kwargs = {k: w for k, w in cross_attention_kwargs.items() if k in attn_parameters}
-
-        return self.processor(
-            self,
-            hidden_states,
-            encoder_hidden_states=encoder_hidden_states,
-            attention_mask=attention_mask,
-            cache=cache,
-            **cross_attention_kwargs,
-        )
-
-
-@maybe_allow_in_graph
-class CausalBasicTransformerBlock(BasicTransformerBlock):
-    def __init__(
-        self,
-        dim: int,
-        num_attention_heads: int,
-        attention_head_dim: int,
-        dropout=0.0,
-        cross_attention_dim: Optional[int] = None,
-        activation_fn: str = "geglu",
-        num_embeds_ada_norm: Optional[int] = None,
-        attention_bias: bool = False,
-        only_cross_attention: bool = False,
-        double_self_attention: bool = False,
-        upcast_attention: bool = False,
-        norm_elementwise_affine: bool = True,
-        norm_type: str = "layer_norm",
-        final_dropout: bool = False,
-    ):
-        super(CausalBasicTransformerBlock, self).__init__(dim, num_attention_heads, attention_head_dim, dropout,
-                                                          cross_attention_dim, activation_fn, num_embeds_ada_norm,
-                                                          attention_bias, only_cross_attention, double_self_attention,
-                                                          upcast_attention, norm_elementwise_affine, norm_type, final_dropout)
-        self.attn1 = CausalAttention(
-            query_dim=dim,
-            heads=num_attention_heads,
-            dim_head=attention_head_dim,
-            dropout=dropout,
-            bias=attention_bias,
-            cross_attention_dim=cross_attention_dim if only_cross_attention else None,
-            upcast_attention=upcast_attention,
-        )
-
-    def forward(
-        self,
-        hidden_states: torch.FloatTensor,
-        attention_mask: Optional[torch.FloatTensor] = None,
-        encoder_hidden_states: Optional[torch.FloatTensor] = None,
-        encoder_attention_mask: Optional[torch.FloatTensor] = None,
-        timestep: Optional[torch.LongTensor] = None,
-        cross_attention_kwargs: Dict[str, Any] = None,
-        class_labels: Optional[torch.LongTensor] = None,
-        cache: torch.Tensor = torch.zeros(0, 0, 0, 0),
-    ) -> Tuple[torch.Tensor, torch.Tensor]:
-        # Notice that normalization is always applied before the real computation in the following blocks.
-        # 1. Self-Attention
-        if self.use_ada_layer_norm:
-            norm_hidden_states = self.norm1(hidden_states, timestep)
-        elif self.use_ada_layer_norm_zero:
-            norm_hidden_states, gate_msa, shift_mlp, scale_mlp, gate_mlp = self.norm1(
-                hidden_states, timestep, class_labels, hidden_dtype=hidden_states.dtype
-            )
-        else:
-            norm_hidden_states = self.norm1(hidden_states)
-
-        cross_attention_kwargs = cross_attention_kwargs if cross_attention_kwargs is not None else {}
-
-        attn_output, cache = self.attn1(
-            norm_hidden_states,
-            encoder_hidden_states=encoder_hidden_states if self.only_cross_attention else None,
-            attention_mask=encoder_attention_mask if self.only_cross_attention else attention_mask,
-            cache=cache,
-            **cross_attention_kwargs,
-        )
-        if self.use_ada_layer_norm_zero:
-            attn_output = gate_msa.unsqueeze(1) * attn_output
-        hidden_states = attn_output + hidden_states
-
-        # 2. Cross-Attention
-        if self.attn2 is not None:
-            norm_hidden_states = (
-                self.norm2(hidden_states, timestep) if self.use_ada_layer_norm else self.norm2(hidden_states)
-            )
-
-            attn_output = self.attn2(
-                norm_hidden_states,
-                encoder_hidden_states=encoder_hidden_states,
-                attention_mask=encoder_attention_mask,
-                **cross_attention_kwargs,
-            )
-            hidden_states = attn_output + hidden_states
-
-        # 3. Feed-forward
-        norm_hidden_states = self.norm3(hidden_states)
-
-        if self.use_ada_layer_norm_zero:
-            norm_hidden_states = norm_hidden_states * (1 + scale_mlp[:, None]) + shift_mlp[:, None]
-
-        if self._chunk_size is not None:
-            # "feed_forward_chunk_size" can be used to save memory
-            if norm_hidden_states.shape[self._chunk_dim] % self._chunk_size != 0:
-                raise ValueError(f"`hidden_states` dimension to be chunked: {norm_hidden_states.shape[self._chunk_dim]} has to be divisible by chunk size: \
-                                 {self._chunk_size}. Make sure to set an appropriate `chunk_size` when calling `unet.enable_forward_chunking`.")
-
-            num_chunks = norm_hidden_states.shape[self._chunk_dim] // self._chunk_size
-            ff_output = torch.cat(
-                [self.ff(hid_slice) for hid_slice in norm_hidden_states.chunk(num_chunks, dim=self._chunk_dim)],
-                dim=self._chunk_dim,
-            )
-        else:
-            ff_output = self.ff(norm_hidden_states)
-
-        if self.use_ada_layer_norm_zero:
-            ff_output = gate_mlp.unsqueeze(1) * ff_output
-
-        hidden_states = ff_output + hidden_states
-
-        return hidden_states, cache
-
 
 class ConditionalDecoder(nn.Module):
     def __init__(
@@ -640,7 +335,7 @@ class CausalConditionalDecoder(ConditionalDecoder):
             resnet = CausalResnetBlock1D(dim=input_channel, dim_out=output_channel, time_emb_dim=time_embed_dim)
             transformer_blocks = nn.ModuleList(
                 [
-                    CausalBasicTransformerBlock(
+                    BasicTransformerBlock(
                         dim=output_channel,
                         num_attention_heads=num_heads,
                         attention_head_dim=attention_head_dim,
@@ -662,7 +357,7 @@ class CausalConditionalDecoder(ConditionalDecoder):
 
             transformer_blocks = nn.ModuleList(
                 [
-                    CausalBasicTransformerBlock(
+                    BasicTransformerBlock(
                         dim=output_channel,
                         num_attention_heads=num_heads,
                         attention_head_dim=attention_head_dim,
@@ -687,7 +382,7 @@ class CausalConditionalDecoder(ConditionalDecoder):
             )
             transformer_blocks = nn.ModuleList(
                 [
-                    CausalBasicTransformerBlock(
+                    BasicTransformerBlock(
                         dim=output_channel,
                         num_attention_heads=num_heads,
                         attention_head_dim=attention_head_dim,
@@ -724,6 +419,9 @@ class CausalConditionalDecoder(ConditionalDecoder):
         Returns:
             _type_: _description_
         """
+        if hasattr(self, 'streaming'):
+            assert self.training is False, 'you have self.streaming attr, make sure that you are running inference mode'
+            streaming = self.streaming
 
         t = self.time_embeddings(t).to(t.dtype)
         t = self.time_mlp(t)
@@ -740,36 +438,36 @@ class CausalConditionalDecoder(ConditionalDecoder):
         masks = [mask]
         for resnet, transformer_blocks, downsample in self.down_blocks:
             mask_down = masks[-1]
-            x, _, _ = resnet(x, mask_down, t)
+            x = resnet(x, mask_down, t)
             x = rearrange(x, "b c t -> b t c").contiguous()
             if streaming is True:
-                attn_mask = add_optional_chunk_mask(x, mask_down.bool(), False, False, 0, self.static_chunk_size, self.num_decoding_left_chunks)
+                attn_mask = add_optional_chunk_mask(x, mask_down.bool(), False, False, 0, self.static_chunk_size, -1)
             else:
                 attn_mask = add_optional_chunk_mask(x, mask_down.bool(), False, False, 0, 0, -1).repeat(1, x.size(1), 1)
             attn_mask = mask_to_bias(attn_mask, x.dtype)
             for transformer_block in transformer_blocks:
-                x, _ = transformer_block(
+                x = transformer_block(
                     hidden_states=x,
                     attention_mask=attn_mask,
                     timestep=t,
                 )
             x = rearrange(x, "b t c -> b c t").contiguous()
             hiddens.append(x)  # Save hidden states for skip connections
-            x, _ = downsample(x * mask_down)
+            x = downsample(x * mask_down)
             masks.append(mask_down[:, :, ::2])
         masks = masks[:-1]
         mask_mid = masks[-1]
 
         for resnet, transformer_blocks in self.mid_blocks:
-            x, _, _ = resnet(x, mask_mid, t)
+            x = resnet(x, mask_mid, t)
             x = rearrange(x, "b c t -> b t c").contiguous()
             if streaming is True:
-                attn_mask = add_optional_chunk_mask(x, mask_mid.bool(), False, False, 0, self.static_chunk_size, self.num_decoding_left_chunks)
+                attn_mask = add_optional_chunk_mask(x, mask_mid.bool(), False, False, 0, self.static_chunk_size, -1)
             else:
                 attn_mask = add_optional_chunk_mask(x, mask_mid.bool(), False, False, 0, 0, -1).repeat(1, x.size(1), 1)
             attn_mask = mask_to_bias(attn_mask, x.dtype)
             for transformer_block in transformer_blocks:
-                x, _ = transformer_block(
+                x = transformer_block(
                     hidden_states=x,
                     attention_mask=attn_mask,
                     timestep=t,
@@ -780,124 +478,21 @@ class CausalConditionalDecoder(ConditionalDecoder):
             mask_up = masks.pop()
             skip = hiddens.pop()
             x = pack([x[:, :, :skip.shape[-1]], skip], "b * t")[0]
-            x, _, _ = resnet(x, mask_up, t)
+            x = resnet(x, mask_up, t)
             x = rearrange(x, "b c t -> b t c").contiguous()
             if streaming is True:
-                attn_mask = add_optional_chunk_mask(x, mask_up.bool(), False, False, 0, self.static_chunk_size, self.num_decoding_left_chunks)
+                attn_mask = add_optional_chunk_mask(x, mask_up.bool(), False, False, 0, self.static_chunk_size, -1)
             else:
                 attn_mask = add_optional_chunk_mask(x, mask_up.bool(), False, False, 0, 0, -1).repeat(1, x.size(1), 1)
             attn_mask = mask_to_bias(attn_mask, x.dtype)
             for transformer_block in transformer_blocks:
-                x, _ = transformer_block(
+                x = transformer_block(
                     hidden_states=x,
                     attention_mask=attn_mask,
                     timestep=t,
                 )
             x = rearrange(x, "b t c -> b c t").contiguous()
-            x, _ = upsample(x * mask_up)
-        x, _ = self.final_block(x, mask_up)
+            x = upsample(x * mask_up)
+        x = self.final_block(x, mask_up)
         output = self.final_proj(x * mask_up)
         return output * mask
-
-    @torch.inference_mode()
-    def forward_chunk(self, x, mask, mu, t, spks=None, cond=None,
-                      down_blocks_conv_cache: torch.Tensor = torch.zeros(0, 0, 0, 0),
-                      down_blocks_kv_cache: torch.Tensor = torch.zeros(0, 0, 0, 0, 0, 0),
-                      mid_blocks_conv_cache: torch.Tensor = torch.zeros(0, 0, 0, 0),
-                      mid_blocks_kv_cache: torch.Tensor = torch.zeros(0, 0, 0, 0, 0, 0),
-                      up_blocks_conv_cache: torch.Tensor = torch.zeros(0, 0, 0, 0),
-                      up_blocks_kv_cache: torch.Tensor = torch.zeros(0, 0, 0, 0, 0, 0),
-                      final_blocks_conv_cache: torch.Tensor = torch.zeros(0, 0, 0)
-                      ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]:
-        """Forward pass of the UNet1DConditional model.
-
-        Args:
-            x (torch.Tensor): shape (batch_size, in_channels, time)
-            mask (_type_): shape (batch_size, 1, time)
-            t (_type_): shape (batch_size)
-            spks (_type_, optional): shape: (batch_size, condition_channels). Defaults to None.
-            cond (_type_, optional): placeholder for future use. Defaults to None.
-
-        Raises:
-            ValueError: _description_
-            ValueError: _description_
-
-        Returns:
-            _type_: _description_
-        """
-
-        t = self.time_embeddings(t).to(t.dtype)
-        t = self.time_mlp(t)
-
-        x = pack([x, mu], "b * t")[0]
-
-        if spks is not None:
-            spks = repeat(spks, "b c -> b c t", t=x.shape[-1])
-            x = pack([x, spks], "b * t")[0]
-        if cond is not None:
-            x = pack([x, cond], "b * t")[0]
-
-        hiddens = []
-        masks = [mask]
-
-        down_blocks_kv_cache_new = torch.zeros(1, 4, 2, x.size(2), 512, 2).to(x.device)
-        mid_blocks_kv_cache_new = torch.zeros(12, 4, 2, x.size(2), 512, 2).to(x.device)
-        up_blocks_kv_cache_new = torch.zeros(1, 4, 2, x.size(2), 512, 2).to(x.device)
-        for index, (resnet, transformer_blocks, downsample) in enumerate(self.down_blocks):
-            mask_down = masks[-1]
-            x, down_blocks_conv_cache[index][:, :320], down_blocks_conv_cache[index][:, 320: 576] = \
-                resnet(x, mask_down, t, down_blocks_conv_cache[index][:, :320], down_blocks_conv_cache[index][:, 320: 576])
-            x = rearrange(x, "b c t -> b t c").contiguous()
-            attn_mask = torch.ones(x.size(0), x.size(1), x.size(1) + down_blocks_kv_cache.size(3), device=x.device).bool()
-            attn_mask = mask_to_bias(attn_mask, x.dtype)
-            for i, transformer_block in enumerate(transformer_blocks):
-                x, down_blocks_kv_cache_new[index, i] = transformer_block(
-                    hidden_states=x,
-                    attention_mask=attn_mask,
-                    timestep=t,
-                    cache=down_blocks_kv_cache[index, i],
-                )
-            x = rearrange(x, "b t c -> b c t").contiguous()
-            hiddens.append(x)  # Save hidden states for skip connections
-            x, down_blocks_conv_cache[index][:, 576:] = downsample(x * mask_down, down_blocks_conv_cache[index][:, 576:])
-            masks.append(mask_down[:, :, ::2])
-        masks = masks[:-1]
-        mask_mid = masks[-1]
-
-        for index, (resnet, transformer_blocks) in enumerate(self.mid_blocks):
-            x, mid_blocks_conv_cache[index][:, :256], mid_blocks_conv_cache[index][:, 256:] = \
-                resnet(x, mask_mid, t, mid_blocks_conv_cache[index][:, :256], mid_blocks_conv_cache[index][:, 256:])
-            x = rearrange(x, "b c t -> b t c").contiguous()
-            attn_mask = torch.ones(x.size(0), x.size(1), x.size(1) + mid_blocks_kv_cache.size(3), device=x.device).bool()
-            attn_mask = mask_to_bias(attn_mask, x.dtype)
-            for i, transformer_block in enumerate(transformer_blocks):
-                x, mid_blocks_kv_cache_new[index, i] = transformer_block(
-                    hidden_states=x,
-                    attention_mask=attn_mask,
-                    timestep=t,
-                    cache=mid_blocks_kv_cache[index, i]
-                )
-            x = rearrange(x, "b t c -> b c t").contiguous()
-
-        for index, (resnet, transformer_blocks, upsample) in enumerate(self.up_blocks):
-            mask_up = masks.pop()
-            skip = hiddens.pop()
-            x = pack([x[:, :, :skip.shape[-1]], skip], "b * t")[0]
-            x, up_blocks_conv_cache[index][:, :512], up_blocks_conv_cache[index][:, 512: 768] = \
-                resnet(x, mask_up, t, up_blocks_conv_cache[index][:, :512], up_blocks_conv_cache[index][:, 512: 768])
-            x = rearrange(x, "b c t -> b t c").contiguous()
-            attn_mask = torch.ones(x.size(0), x.size(1), x.size(1) + up_blocks_kv_cache.size(3), device=x.device).bool()
-            attn_mask = mask_to_bias(attn_mask, x.dtype)
-            for i, transformer_block in enumerate(transformer_blocks):
-                x, up_blocks_kv_cache_new[index, i] = transformer_block(
-                    hidden_states=x,
-                    attention_mask=attn_mask,
-                    timestep=t,
-                    cache=up_blocks_kv_cache[index, i]
-                )
-            x = rearrange(x, "b t c -> b c t").contiguous()
-            x, up_blocks_conv_cache[index][:, 768:] = upsample(x * mask_up, up_blocks_conv_cache[index][:, 768:])
-        x, final_blocks_conv_cache = self.final_block(x, mask_up, final_blocks_conv_cache)
-        output = self.final_proj(x * mask_up)
-        return output * mask, down_blocks_conv_cache, down_blocks_kv_cache_new, mid_blocks_conv_cache, mid_blocks_kv_cache_new, \
-            up_blocks_conv_cache, up_blocks_kv_cache_new, final_blocks_conv_cache

cosyvoice/flow/flow.py

@@ -92,7 +92,6 @@ class MaskedDiffWithXvec(torch.nn.Module):
 
         mask = (~make_pad_mask(feat_len)).to(h)
         # NOTE this is unnecessary, feat/h already same shape
-        feat = F.interpolate(feat.unsqueeze(dim=1), size=h.shape[1:], mode="nearest").squeeze(dim=1)
         loss, _ = self.decoder.compute_loss(
             feat.transpose(1, 2).contiguous(),
             mask.unsqueeze(1),
@@ -214,7 +213,6 @@ class CausalMaskedDiffWithXvec(torch.nn.Module):
         h = self.encoder_proj(h)
 
         # get conditions
-        feat = F.interpolate(feat.unsqueeze(dim=1), size=h.shape[1:], mode="nearest").squeeze(dim=1)
         conds = torch.zeros(feat.shape, device=token.device)
         for i, j in enumerate(feat_len):
             if random.random() < 0.5:
@@ -243,7 +241,6 @@ class CausalMaskedDiffWithXvec(torch.nn.Module):
                   prompt_feat,
                   prompt_feat_len,
                   embedding,
-                  cache,
                   finalize):
         assert token.shape[0] == 1
         # xvec projection
@@ -257,16 +254,10 @@ class CausalMaskedDiffWithXvec(torch.nn.Module):
 
         # text encode
         if finalize is True:
-            h, h_lengths, encoder_cache = self.encoder.forward_chunk(token, token_len, **cache['encoder_cache'])
+            h, h_lengths = self.encoder(token, token_len)
         else:
             token, context = token[:, :-self.pre_lookahead_len], token[:, -self.pre_lookahead_len:]
-            h, h_lengths, encoder_cache = self.encoder.forward_chunk(token, token_len, context=context, **cache['encoder_cache'])
-        cache['encoder_cache']['offset'] = encoder_cache[0]
-        cache['encoder_cache']['pre_lookahead_layer_conv2_cache'] = encoder_cache[1]
-        cache['encoder_cache']['encoders_kv_cache'] = encoder_cache[2]
-        cache['encoder_cache']['upsample_offset'] = encoder_cache[3]
-        cache['encoder_cache']['upsample_conv_cache'] = encoder_cache[4]
-        cache['encoder_cache']['upsample_kv_cache'] = encoder_cache[5]
+            h, h_lengths = self.encoder(token, token_len, context=context)
         mel_len1, mel_len2 = prompt_feat.shape[1], h.shape[1] - prompt_feat.shape[1]
         h = self.encoder_proj(h)
 
@@ -276,14 +267,13 @@ class CausalMaskedDiffWithXvec(torch.nn.Module):
         conds = conds.transpose(1, 2)
 
         mask = (~make_pad_mask(torch.tensor([mel_len1 + mel_len2]))).to(h)
-        feat, cache['decoder_cache'] = self.decoder(
+        feat, _ = self.decoder(
             mu=h.transpose(1, 2).contiguous(),
             mask=mask.unsqueeze(1),
             spks=embedding,
             cond=conds,
             n_timesteps=10,
-            cache=cache['decoder_cache']
         )
         feat = feat[:, :, mel_len1:]
         assert feat.shape[2] == mel_len2
-        return feat.float(), cache
+        return feat.float(), None

cosyvoice/flow/flow_matching.py

@@ -1,4 +1,5 @@
 # Copyright (c) 2024 Alibaba Inc (authors: Xiang Lyu, Zhihao Du)
+#               2025 Alibaba Inc (authors: Xiang Lyu, Bofan Zhou)
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -125,21 +126,26 @@ class ConditionalCFM(BASECFM):
         if isinstance(self.estimator, torch.nn.Module):
             return self.estimator(x, mask, mu, t, spks, cond)
         else:
-            with self.lock:
-                self.estimator.set_input_shape('x', (2, 80, x.size(2)))
-                self.estimator.set_input_shape('mask', (2, 1, x.size(2)))
-                self.estimator.set_input_shape('mu', (2, 80, x.size(2)))
-                self.estimator.set_input_shape('t', (2,))
-                self.estimator.set_input_shape('spks', (2, 80))
-                self.estimator.set_input_shape('cond', (2, 80, x.size(2)))
-                # run trt engine
-                assert self.estimator.execute_v2([x.contiguous().data_ptr(),
-                                                  mask.contiguous().data_ptr(),
-                                                  mu.contiguous().data_ptr(),
-                                                  t.contiguous().data_ptr(),
-                                                  spks.contiguous().data_ptr(),
-                                                  cond.contiguous().data_ptr(),
-                                                  x.data_ptr()]) is True
+            estimator, trt_engine = self.estimator.acquire_estimator()
+            estimator.set_input_shape('x', (2, 80, x.size(2)))
+            estimator.set_input_shape('mask', (2, 1, x.size(2)))
+            estimator.set_input_shape('mu', (2, 80, x.size(2)))
+            estimator.set_input_shape('t', (2,))
+            estimator.set_input_shape('spks', (2, 80))
+            estimator.set_input_shape('cond', (2, 80, x.size(2)))
+            data_ptrs = [x.contiguous().data_ptr(),
+                         mask.contiguous().data_ptr(),
+                         mu.contiguous().data_ptr(),
+                         t.contiguous().data_ptr(),
+                         spks.contiguous().data_ptr(),
+                         cond.contiguous().data_ptr(),
+                         x.data_ptr()]
+            for i, j in enumerate(data_ptrs):
+                estimator.set_tensor_address(trt_engine.get_tensor_name(i), j)
+            # run trt engine
+            assert estimator.execute_async_v3(torch.cuda.current_stream().cuda_stream) is True
+            torch.cuda.current_stream().synchronize()
+            self.estimator.release_estimator(estimator)
             return x
 
     def compute_loss(self, x1, mask, mu, spks=None, cond=None, streaming=False):
@@ -190,7 +196,7 @@ class CausalConditionalCFM(ConditionalCFM):
         self.rand_noise = torch.randn([1, 80, 50 * 300])
 
     @torch.inference_mode()
-    def forward(self, mu, mask, n_timesteps, temperature=1.0, spks=None, cond=None, cache={}):
+    def forward(self, mu, mask, n_timesteps, temperature=1.0, spks=None, cond=None):
         """Forward diffusion
 
         Args:
@@ -209,131 +215,9 @@ class CausalConditionalCFM(ConditionalCFM):
                 shape: (batch_size, n_feats, mel_timesteps)
         """
 
-        offset = cache.pop('offset')
-        z = self.rand_noise[:, :, :mu.size(2) + offset].to(mu.device).to(mu.dtype) * temperature
-        z = z[:, :, offset:]
-        offset += mu.size(2)
+        z = self.rand_noise[:, :, :mu.size(2)].to(mu.device).to(mu.dtype) * temperature
         # fix prompt and overlap part mu and z
         t_span = torch.linspace(0, 1, n_timesteps + 1, device=mu.device, dtype=mu.dtype)
         if self.t_scheduler == 'cosine':
             t_span = 1 - torch.cos(t_span * 0.5 * torch.pi)
-        mel, cache = self.solve_euler(z, t_span=t_span, mu=mu, mask=mask, spks=spks, cond=cond, cache=cache)
-        cache['offset'] = offset
-        return mel, cache
-
-    def solve_euler(self, x, t_span, mu, mask, spks, cond, cache):
-        """
-        Fixed euler solver for ODEs.
-        Args:
-            x (torch.Tensor): random noise
-            t_span (torch.Tensor): n_timesteps interpolated
-                shape: (n_timesteps + 1,)
-            mu (torch.Tensor): output of encoder
-                shape: (batch_size, n_feats, mel_timesteps)
-            mask (torch.Tensor): output_mask
-                shape: (batch_size, 1, mel_timesteps)
-            spks (torch.Tensor, optional): speaker ids. Defaults to None.
-                shape: (batch_size, spk_emb_dim)
-            cond: Not used but kept for future purposes
-        """
-        t, _, dt = t_span[0], t_span[-1], t_span[1] - t_span[0]
-        t = t.unsqueeze(dim=0)
-
-        # I am storing this because I can later plot it by putting a debugger here and saving it to a file
-        # Or in future might add like a return_all_steps flag
-        sol = []
-
-        # Do not use concat, it may cause memory format changed and trt infer with wrong results!
-        x_in = torch.zeros([2, 80, x.size(2)], device=x.device, dtype=x.dtype)
-        mask_in = torch.zeros([2, 1, x.size(2)], device=x.device, dtype=x.dtype)
-        mu_in = torch.zeros([2, 80, x.size(2)], device=x.device, dtype=x.dtype)
-        t_in = torch.zeros([2], device=x.device, dtype=x.dtype)
-        spks_in = torch.zeros([2, 80], device=x.device, dtype=x.dtype)
-        cond_in = torch.zeros([2, 80, x.size(2)], device=x.device, dtype=x.dtype)
-        flow_cache_size = cache['down_blocks_kv_cache'].shape[4]
-        for step in range(1, len(t_span)):
-            # Classifier-Free Guidance inference introduced in VoiceBox
-            x_in[:] = x
-            mask_in[:] = mask
-            mu_in[0] = mu
-            t_in[:] = t.unsqueeze(0)
-            spks_in[0] = spks
-            cond_in[0] = cond
-            cache_step = {k: v[step - 1] for k, v in cache.items()}
-            dphi_dt, cache_step = self.forward_estimator(
-                x_in, mask_in,
-                mu_in, t_in,
-                spks_in,
-                cond_in,
-                cache_step
-            )
-            # NOTE if smaller than flow_cache_size, means last chunk, no need to cache
-            if flow_cache_size != 0 and x_in.shape[2] >= flow_cache_size:
-                cache['down_blocks_conv_cache'][step - 1] = cache_step[0]
-                cache['down_blocks_kv_cache'][step - 1] = cache_step[1][:, :, :, -flow_cache_size:]
-                cache['mid_blocks_conv_cache'][step - 1] = cache_step[2]
-                cache['mid_blocks_kv_cache'][step - 1] = cache_step[3][:, :, :, -flow_cache_size:]
-                cache['up_blocks_conv_cache'][step - 1] = cache_step[4]
-                cache['up_blocks_kv_cache'][step - 1] = cache_step[5][:, :, :, -flow_cache_size:]
-                cache['final_blocks_conv_cache'][step - 1] = cache_step[6]
-            dphi_dt, cfg_dphi_dt = torch.split(dphi_dt, [x.size(0), x.size(0)], dim=0)
-            dphi_dt = ((1.0 + self.inference_cfg_rate) * dphi_dt - self.inference_cfg_rate * cfg_dphi_dt)
-            x = x + dt * dphi_dt
-            t = t + dt
-            sol.append(x)
-            if step < len(t_span) - 1:
-                dt = t_span[step + 1] - t
-        return sol[-1].float(), cache
-
-    def forward_estimator(self, x, mask, mu, t, spks, cond, cache):
-        if isinstance(self.estimator, torch.nn.Module):
-            x, cache1, cache2, cache3, cache4, cache5, cache6, cache7 = self.estimator.forward_chunk(x, mask, mu, t, spks, cond, **cache)
-            cache = (cache1, cache2, cache3, cache4, cache5, cache6, cache7)
-        else:
-            with self.lock:
-                self.estimator.set_input_shape('x', (2, 80, x.size(2)))
-                self.estimator.set_input_shape('mask', (2, 1, x.size(2)))
-                self.estimator.set_input_shape('mu', (2, 80, x.size(2)))
-                self.estimator.set_input_shape('t', (2,))
-                self.estimator.set_input_shape('spks', (2, 80))
-                self.estimator.set_input_shape('cond', (2, 80, x.size(2)))
-                self.estimator.set_input_shape('down_blocks_conv_cache', cache['down_blocks_conv_cache'].shape)
-                self.estimator.set_input_shape('down_blocks_kv_cache', cache['down_blocks_kv_cache'].shape)
-                self.estimator.set_input_shape('mid_blocks_conv_cache', cache['mid_blocks_conv_cache'].shape)
-                self.estimator.set_input_shape('mid_blocks_kv_cache', cache['mid_blocks_kv_cache'].shape)
-                self.estimator.set_input_shape('up_blocks_conv_cache', cache['up_blocks_conv_cache'].shape)
-                self.estimator.set_input_shape('up_blocks_kv_cache', cache['up_blocks_kv_cache'].shape)
-                self.estimator.set_input_shape('final_blocks_conv_cache', cache['final_blocks_conv_cache'].shape)
-                # run trt engine
-                down_blocks_kv_cache_out = torch.zeros(1, 4, 2, x.size(2), 512, 2).to(x)
-                mid_blocks_kv_cache_out = torch.zeros(12, 4, 2, x.size(2), 512, 2).to(x)
-                up_blocks_kv_cache_out = torch.zeros(1, 4, 2, x.size(2), 512, 2).to(x)
-                assert self.estimator.execute_v2([x.contiguous().data_ptr(),
-                                                  mask.contiguous().data_ptr(),
-                                                  mu.contiguous().data_ptr(),
-                                                  t.contiguous().data_ptr(),
-                                                  spks.contiguous().data_ptr(),
-                                                  cond.contiguous().data_ptr(),
-                                                  cache['down_blocks_conv_cache'].contiguous().data_ptr(),
-                                                  cache['down_blocks_kv_cache'].contiguous().data_ptr(),
-                                                  cache['mid_blocks_conv_cache'].contiguous().data_ptr(),
-                                                  cache['mid_blocks_kv_cache'].contiguous().data_ptr(),
-                                                  cache['up_blocks_conv_cache'].contiguous().data_ptr(),
-                                                  cache['up_blocks_kv_cache'].contiguous().data_ptr(),
-                                                  cache['final_blocks_conv_cache'].contiguous().data_ptr(),
-                                                  x.data_ptr(),
-                                                  cache['down_blocks_conv_cache'].data_ptr(),
-                                                  down_blocks_kv_cache_out.data_ptr(),
-                                                  cache['mid_blocks_conv_cache'].data_ptr(),
-                                                  mid_blocks_kv_cache_out.data_ptr(),
-                                                  cache['up_blocks_conv_cache'].data_ptr(),
-                                                  up_blocks_kv_cache_out.data_ptr(),
-                                                  cache['final_blocks_conv_cache'].data_ptr()]) is True
-                cache = (cache['down_blocks_conv_cache'],
-                         down_blocks_kv_cache_out,
-                         cache['mid_blocks_conv_cache'],
-                         mid_blocks_kv_cache_out,
-                         cache['up_blocks_conv_cache'],
-                         up_blocks_kv_cache_out,
-                         cache['final_blocks_conv_cache'])
-        return x, cache
+        return self.solve_euler(z, t_span=t_span, mu=mu, mask=mask, spks=spks, cond=cond), None

cosyvoice/hifigan/generator.py

@@ -223,6 +223,172 @@ class SourceModuleHnNSF(torch.nn.Module):
         return sine_merge, noise, uv
 
 
+class SineGen2(torch.nn.Module):
+    """ Definition of sine generator
+    SineGen(samp_rate, harmonic_num = 0,
+            sine_amp = 0.1, noise_std = 0.003,
+            voiced_threshold = 0,
+            flag_for_pulse=False)
+    samp_rate: sampling rate in Hz
+    harmonic_num: number of harmonic overtones (default 0)
+    sine_amp: amplitude of sine-wavefrom (default 0.1)
+    noise_std: std of Gaussian noise (default 0.003)
+    voiced_thoreshold: F0 threshold for U/V classification (default 0)
+    flag_for_pulse: this SinGen is used inside PulseGen (default False)
+    Note: when flag_for_pulse is True, the first time step of a voiced
+        segment is always sin(np.pi) or cos(0)
+    """
+
+    def __init__(self, samp_rate, upsample_scale, harmonic_num=0,
+                 sine_amp=0.1, noise_std=0.003,
+                 voiced_threshold=0,
+                 flag_for_pulse=False):
+        super(SineGen2, self).__init__()
+        self.sine_amp = sine_amp
+        self.noise_std = noise_std
+        self.harmonic_num = harmonic_num
+        self.dim = self.harmonic_num + 1
+        self.sampling_rate = samp_rate
+        self.voiced_threshold = voiced_threshold
+        self.flag_for_pulse = flag_for_pulse
+        self.upsample_scale = upsample_scale
+
+    def _f02uv(self, f0):
+        # generate uv signal
+        uv = (f0 > self.voiced_threshold).type(torch.float32)
+        return uv
+
+    def _f02sine(self, f0_values):
+        """ f0_values: (batchsize, length, dim)
+            where dim indicates fundamental tone and overtones
+        """
+        # convert to F0 in rad. The interger part n can be ignored
+        # because 2 * np.pi * n doesn't affect phase
+        rad_values = (f0_values / self.sampling_rate) % 1
+
+        # initial phase noise (no noise for fundamental component)
+        rand_ini = torch.rand(f0_values.shape[0], f0_values.shape[2], device=f0_values.device)
+        rand_ini[:, 0] = 0
+        rad_values[:, 0, :] = rad_values[:, 0, :] + rand_ini
+
+        # instantanouse phase sine[t] = sin(2*pi \sum_i=1 ^{t} rad)
+        if not self.flag_for_pulse:
+            rad_values = torch.nn.functional.interpolate(rad_values.transpose(1, 2),
+                                                         scale_factor=1 / self.upsample_scale,
+                                                         mode="linear").transpose(1, 2)
+
+            phase = torch.cumsum(rad_values, dim=1) * 2 * np.pi
+            phase = torch.nn.functional.interpolate(phase.transpose(1, 2) * self.upsample_scale,
+                                                    scale_factor=self.upsample_scale, mode="linear").transpose(1, 2)
+            sines = torch.sin(phase)
+        else:
+            # If necessary, make sure that the first time step of every
+            # voiced segments is sin(pi) or cos(0)
+            # This is used for pulse-train generation
+
+            # identify the last time step in unvoiced segments
+            uv = self._f02uv(f0_values)
+            uv_1 = torch.roll(uv, shifts=-1, dims=1)
+            uv_1[:, -1, :] = 1
+            u_loc = (uv < 1) * (uv_1 > 0)
+
+            # get the instantanouse phase
+            tmp_cumsum = torch.cumsum(rad_values, dim=1)
+            # different batch needs to be processed differently
+            for idx in range(f0_values.shape[0]):
+                temp_sum = tmp_cumsum[idx, u_loc[idx, :, 0], :]
+                temp_sum[1:, :] = temp_sum[1:, :] - temp_sum[0:-1, :]
+                # stores the accumulation of i.phase within
+                # each voiced segments
+                tmp_cumsum[idx, :, :] = 0
+                tmp_cumsum[idx, u_loc[idx, :, 0], :] = temp_sum
+
+            # rad_values - tmp_cumsum: remove the accumulation of i.phase
+            # within the previous voiced segment.
+            i_phase = torch.cumsum(rad_values - tmp_cumsum, dim=1)
+
+            # get the sines
+            sines = torch.cos(i_phase * 2 * np.pi)
+        return sines
+
+    def forward(self, f0):
+        """ sine_tensor, uv = forward(f0)
+        input F0: tensor(batchsize=1, length, dim=1)
+                  f0 for unvoiced steps should be 0
+        output sine_tensor: tensor(batchsize=1, length, dim)
+        output uv: tensor(batchsize=1, length, 1)
+        """
+        # fundamental component
+        fn = torch.multiply(f0, torch.FloatTensor([[range(1, self.harmonic_num + 2)]]).to(f0.device))
+
+        # generate sine waveforms
+        sine_waves = self._f02sine(fn) * self.sine_amp
+
+        # generate uv signal
+        uv = self._f02uv(f0)
+
+        # noise: for unvoiced should be similar to sine_amp
+        #        std = self.sine_amp/3 -> max value ~ self.sine_amp
+        # .       for voiced regions is self.noise_std
+        noise_amp = uv * self.noise_std + (1 - uv) * self.sine_amp / 3
+        noise = noise_amp * torch.randn_like(sine_waves)
+
+        # first: set the unvoiced part to 0 by uv
+        # then: additive noise
+        sine_waves = sine_waves * uv + noise
+        return sine_waves, uv, noise
+
+
+class SourceModuleHnNSF2(torch.nn.Module):
+    """ SourceModule for hn-nsf
+    SourceModule(sampling_rate, harmonic_num=0, sine_amp=0.1,
+                 add_noise_std=0.003, voiced_threshod=0)
+    sampling_rate: sampling_rate in Hz
+    harmonic_num: number of harmonic above F0 (default: 0)
+    sine_amp: amplitude of sine source signal (default: 0.1)
+    add_noise_std: std of additive Gaussian noise (default: 0.003)
+        note that amplitude of noise in unvoiced is decided
+        by sine_amp
+    voiced_threshold: threhold to set U/V given F0 (default: 0)
+    Sine_source, noise_source = SourceModuleHnNSF(F0_sampled)
+    F0_sampled (batchsize, length, 1)
+    Sine_source (batchsize, length, 1)
+    noise_source (batchsize, length 1)
+    uv (batchsize, length, 1)
+    """
+
+    def __init__(self, sampling_rate, upsample_scale, harmonic_num=0, sine_amp=0.1,
+                 add_noise_std=0.003, voiced_threshod=0):
+        super(SourceModuleHnNSF2, self).__init__()
+
+        self.sine_amp = sine_amp
+        self.noise_std = add_noise_std
+
+        # to produce sine waveforms
+        self.l_sin_gen = SineGen2(sampling_rate, upsample_scale, harmonic_num,
+                                  sine_amp, add_noise_std, voiced_threshod)
+
+        # to merge source harmonics into a single excitation
+        self.l_linear = torch.nn.Linear(harmonic_num + 1, 1)
+        self.l_tanh = torch.nn.Tanh()
+
+    def forward(self, x):
+        """
+        Sine_source, noise_source = SourceModuleHnNSF(F0_sampled)
+        F0_sampled (batchsize, length, 1)
+        Sine_source (batchsize, length, 1)
+        noise_source (batchsize, length 1)
+        """
+        # source for harmonic branch
+        with torch.no_grad():
+            sine_wavs, uv, _ = self.l_sin_gen(x)
+        sine_merge = self.l_tanh(self.l_linear(sine_wavs))
+
+        # source for noise branch, in the same shape as uv
+        noise = torch.randn_like(uv) * self.sine_amp / 3
+        return sine_merge, noise, uv
+
+
 class HiFTGenerator(nn.Module):
     """
     HiFTNet Generator: Neural Source Filter + ISTFTNet
@@ -259,7 +425,9 @@ class HiFTGenerator(nn.Module):
 
         self.num_kernels = len(resblock_kernel_sizes)
         self.num_upsamples = len(upsample_rates)
-        self.m_source = SourceModuleHnNSF(
+        # NOTE in CosyVoice2, we use the original SourceModuleHnNSF implementation
+        this_SourceModuleHnNSF = SourceModuleHnNSF if self.sampling_rate == 22050 else SourceModuleHnNSF2
+        self.m_source = this_SourceModuleHnNSF(
             sampling_rate=sampling_rate,
             upsample_scale=np.prod(upsample_rates) * istft_params["hop_len"],
             harmonic_num=nb_harmonics,

cosyvoice/llm/llm_dpo.py

@@ -0,0 +1,556 @@
+# Copyright (c) 2024 Alibaba Inc (authors: Xiang Lyu, Zhihao Du)
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+from typing import Dict, Optional, Callable, List, Generator
+import torch
+from torch import nn
+import torch.nn.functional as F
+from transformers import Qwen2ForCausalLM
+from torch.nn.utils.rnn import pad_sequence, unpad_sequence
+from cosyvoice.utils.common import IGNORE_ID
+from cosyvoice.transformer.label_smoothing_loss import LabelSmoothingLoss
+from cosyvoice.utils.common import th_accuracy
+from cosyvoice.utils.file_utils import logging
+from cosyvoice.utils.mask import make_pad_mask
+
+
+class TransformerLM(torch.nn.Module):
+    def __init__(
+            self,
+            text_encoder_input_size: int,
+            llm_input_size: int,
+            llm_output_size: int,
+            text_token_size: int,
+            speech_token_size: int,
+            text_encoder: torch.nn.Module,
+            llm: torch.nn.Module,
+            sampling: Callable,
+            length_normalized_loss: bool = True,
+            lsm_weight: float = 0.0,
+            spk_embed_dim: int = 192,
+    ):
+        super().__init__()
+        self.llm_input_size = llm_input_size
+        self.speech_token_size = speech_token_size
+        # 1. build text token inputs related modules
+        self.text_embedding = torch.nn.Embedding(text_token_size, text_encoder_input_size)
+        self.text_encoder = text_encoder
+        self.text_encoder_affine_layer = nn.Linear(
+            self.text_encoder.output_size(),
+            llm_input_size
+        )
+
+        # 2. build speech token language model related modules
+        self.sos_eos = 0
+        self.task_id = 1
+        self.llm_embedding = torch.nn.Embedding(2, llm_input_size)
+        self.llm = llm
+        self.llm_decoder = nn.Linear(llm_output_size, speech_token_size + 1)
+        self.criterion_ce = LabelSmoothingLoss(
+            size=speech_token_size + 1,
+            padding_idx=IGNORE_ID,
+            smoothing=lsm_weight,
+            normalize_length=length_normalized_loss,
+        )
+
+        # 3. [Optional] build speech token related modules
+        self.speech_embedding = torch.nn.Embedding(speech_token_size, llm_input_size)
+        self.spk_embed_affine_layer = torch.nn.Linear(spk_embed_dim, llm_input_size)
+
+        # 4. sampling method
+        self.sampling = sampling
+
+    def encode(
+            self,
+            text: torch.Tensor,
+            text_lengths: torch.Tensor,
+    ):
+        encoder_out, encoder_mask = self.text_encoder(text, text_lengths, decoding_chunk_size=1, num_decoding_left_chunks=-1)
+        encoder_out_lens = encoder_mask.squeeze(1).sum(1)
+        encoder_out = self.text_encoder_affine_layer(encoder_out)
+        return encoder_out, encoder_out_lens
+
+    def pad_unpad_sequence(self, sos_eos_emb, embedding, text_token, text_token_len, task_id_emb, speech_token, speech_token_len):
+        text_token = unpad_sequence(text_token, text_token_len.cpu(), batch_first=True)
+        speech_token = unpad_sequence(speech_token, speech_token_len.cpu(), batch_first=True)
+        lm_input = [torch.concat([sos_eos_emb.squeeze(dim=0), embedding[i], text_token[i], task_id_emb.squeeze(dim=0), speech_token[i]], dim=0)
+                    for i in range(len(text_token))]
+        lm_input_len = torch.tensor([i.size(0) for i in lm_input], dtype=torch.int32)
+        lm_input = pad_sequence(lm_input, batch_first=True, padding_value=IGNORE_ID)
+        return lm_input, lm_input_len
+
+    def forward(
+            self,
+            batch: dict,
+            device: torch.device,
+    ) -> Dict[str, Optional[torch.Tensor]]:
+        """
+        Args:
+            text: (B, L, D)
+            text_lengths: (B,)
+            audio: (B, T, N) or (B, T)
+            audio_lengths: (B,)
+        """
+        text_token = batch['text_token'].to(device)
+        text_token_len = batch['text_token_len'].to(device)
+        speech_token = batch['speech_token'].to(device)
+        speech_token_len = batch['speech_token_len'].to(device)
+        embedding = batch['embedding'].to(device)
+
+        # 1. prepare llm_target
+        lm_target = [torch.tensor([IGNORE_ID] * (2 + text_token_len[i]) + speech_token[i, :speech_token_len[i]].tolist() +
+                                  [self.speech_token_size]) for i in range(text_token.size(0))]
+        lm_target = pad_sequence(lm_target, batch_first=True, padding_value=IGNORE_ID).to(device)
+
+        # 1. encode text_token
+        text_token = self.text_embedding(text_token)
+        text_token, text_token_len = self.encode(text_token, text_token_len)
+
+        # 2. embedding projection
+        embedding = F.normalize(embedding, dim=1)
+        embedding = self.spk_embed_affine_layer(embedding)
+        embedding = embedding.unsqueeze(1)
+
+        # 3. eos and task_id
+        sos_eos_emb = self.llm_embedding.weight[self.sos_eos].reshape(1, 1, -1)
+        task_id_emb = self.llm_embedding.weight[self.task_id].reshape(1, 1, -1)
+
+        # 4. encode speech_token
+        speech_token = self.speech_embedding(speech_token)
+
+        # 5. unpad and pad
+        lm_input, lm_input_len = self.pad_unpad_sequence(sos_eos_emb, embedding, text_token, text_token_len,
+                                                         task_id_emb, speech_token, speech_token_len)
+
+        # 6. run lm forward
+        lm_output, lm_output_mask = self.llm(lm_input, lm_input_len.to(device))
+        logits = self.llm_decoder(lm_output)
+        loss = self.criterion_ce(logits, lm_target)
+        acc = th_accuracy(logits.view(-1, self.speech_token_size + 1), lm_target, ignore_label=IGNORE_ID)
+        return {'loss': loss, 'acc': acc}
+
+    def sampling_ids(
+            self,
+            weighted_scores: torch.Tensor,
+            decoded_tokens: List,
+            sampling: int,
+            ignore_eos: bool = True,
+    ):
+        num_trials, max_trials = 0, 100
+        while True:
+            top_ids = self.sampling(weighted_scores, decoded_tokens, sampling)
+            if (not ignore_eos) or (self.speech_token_size not in top_ids):
+                break
+            num_trials += 1
+            if num_trials > max_trials:
+                raise RuntimeError('sampling reaches max_trials {} and still get eos when ignore_eos is True, check your input!'.format(max_trials))
+        return top_ids
+
+    @torch.inference_mode()
+    def inference(
+            self,
+            text: torch.Tensor,
+            text_len: torch.Tensor,
+            prompt_text: torch.Tensor,
+            prompt_text_len: torch.Tensor,
+            prompt_speech_token: torch.Tensor,
+            prompt_speech_token_len: torch.Tensor,
+            embedding: torch.Tensor,
+            sampling: int = 25,
+            max_token_text_ratio: float = 20,
+            min_token_text_ratio: float = 2,
+    ) -> Generator[torch.Tensor, None, None]:
+        if self.fp16 is True:
+            embedding = embedding.half()
+
+        device = text.device
+        text = torch.concat([prompt_text, text], dim=1)
+        text_len += prompt_text_len
+        text = self.text_embedding(text)
+
+        # 1. encode text
+        text, text_len = self.encode(text, text_len)
+
+        # 2. encode embedding
+        if embedding.shape[0] != 0:
+            embedding = F.normalize(embedding, dim=1)
+            embedding = self.spk_embed_affine_layer(embedding)
+            embedding = embedding.unsqueeze(dim=1)
+        else:
+            embedding = torch.zeros(1, 0, self.llm_input_size, dtype=text.dtype).to(device).to(text.dtype)
+
+        # 3. concat llm_input
+        sos_eos_emb = self.llm_embedding.weight[self.sos_eos].reshape(1, 1, -1)
+        task_id_emb = self.llm_embedding.weight[self.task_id].reshape(1, 1, -1)
+        if prompt_speech_token_len != 0:
+            prompt_speech_token_emb = self.speech_embedding(prompt_speech_token)
+        else:
+            prompt_speech_token_emb = torch.zeros(1, 0, self.llm_input_size, dtype=text.dtype).to(device)
+        lm_input = torch.concat([sos_eos_emb, embedding, text, task_id_emb, prompt_speech_token_emb], dim=1)
+
+        # 4. cal min/max_length
+        min_len = int((text_len - prompt_text_len) * min_token_text_ratio)
+        max_len = int((text_len - prompt_text_len) * max_token_text_ratio)
+
+        # 5. step by step decode
+        out_tokens = []
+        offset = 0
+        att_cache, cnn_cache = torch.zeros((0, 0, 0, 0), device=lm_input.device), torch.zeros((0, 0, 0, 0), device=lm_input.device)
+        for i in range(max_len):
+            y_pred, att_cache, cnn_cache = self.llm.forward_chunk(lm_input, offset=offset, required_cache_size=-1,
+                                                                  att_cache=att_cache, cnn_cache=cnn_cache,
+                                                                  att_mask=torch.tril(torch.ones((1, lm_input.shape[1], lm_input.shape[1]),
+                                                                                                 device=lm_input.device)).to(torch.bool))
+            logp = self.llm_decoder(y_pred[:, -1]).log_softmax(dim=-1)
+            # force continue decode first token
+            if i == 0:
+                logp[:, self.speech_token_size] = -float('inf')
+            top_ids = self.sampling_ids(logp.squeeze(dim=0), out_tokens, sampling, ignore_eos=True if i < min_len else False).item()
+            if top_ids == self.speech_token_size:
+                break
+            # in stream mode, yield token one by one
+            yield top_ids
+            out_tokens.append(top_ids)
+            offset += lm_input.size(1)
+            lm_input = self.speech_embedding.weight[top_ids].reshape(1, 1, -1)
+
+
+class Qwen2Encoder(torch.nn.Module):
+    def __init__(self, pretrain_path):
+        super().__init__()
+        self.model = Qwen2ForCausalLM.from_pretrained(pretrain_path)
+
+    def forward_one_step(self, xs, masks, cache=None):
+        input_masks = masks[:, -1, :]
+        outs = self.model(
+            inputs_embeds=xs,
+            attention_mask=input_masks,
+            output_hidden_states=True,
+            return_dict=True,
+            use_cache=True,
+            past_key_values=cache,
+        )
+        xs = outs.hidden_states[-1]
+        new_cache = outs.past_key_values
+        return xs, new_cache
+
+
+class Qwen2LM(TransformerLM):
+    def __init__(
+            self,
+            llm_input_size: int,
+            llm_output_size: int,
+            speech_token_size: int,
+            llm: torch.nn.Module,
+            sampling: Callable,
+            length_normalized_loss: bool = True,
+            lsm_weight: float = 0.0,
+            mix_ratio: List[int] = [5, 15],
+            dpo: bool = False,
+    ):
+        torch.nn.Module.__init__(self)
+        self.llm_input_size = llm_input_size
+        self.llm_output_size = llm_output_size
+        self.speech_token_size = speech_token_size
+
+        # 2. build speech token language model related modules
+        self.sos_eos = 0
+        self.task_id = 1
+        self.fill_token = 2
+
+        self.llm_embedding = torch.nn.Embedding(2, llm_input_size)
+        self.llm = llm
+        self.llm_decoder = nn.Linear(llm_output_size, speech_token_size + 3)
+        self.criterion_ce = LabelSmoothingLoss(
+            size=speech_token_size + 3,
+            padding_idx=IGNORE_ID,
+            smoothing=lsm_weight,
+            normalize_length=length_normalized_loss,
+        )
+
+        # 3. [Optional] build speech token related modules
+        self.speech_embedding = torch.nn.Embedding(speech_token_size + 3, llm_input_size)
+
+        # 4. sampling method
+        self.sampling = sampling
+        self.mix_ratio = mix_ratio
+
+        # 5. [Optional] set dpo
+        self.dpo = dpo
+
+
+    def forward(
+            self,
+            batch: dict,
+            device: torch.device,
+        ) -> Dict[str, Optional[torch.Tensor]]:
+        text_token = batch['text_token'].to(device)
+        text_token_len = batch['text_token_len'].to(device)
+        speech_token = batch['speech_token'].to(device)
+        speech_token_len = batch['speech_token_len'].to(device)
+        if self.dpo:
+            reject_speech_token = batch['reject_speech_token'].to(device)
+            reject_speech_token_len = batch['reject_speech_token_len'].to(device)
+        # 1. prepare llm_target
+        sos_eos_emb = self.llm_embedding.weight[self.sos_eos].reshape(1, 1, -1)
+        task_id_emb = self.llm_embedding.weight[self.task_id].reshape(1, 1, -1)
+        target_ids = [torch.tensor([IGNORE_ID] * (1 + text_token_len[i]) + speech_token[i, :speech_token_len[i]].tolist() +
+                                        [self.speech_token_size]) for i in range(text_token.size(0))]
+        if self.dpo:
+            reject_target_ids = [torch.tensor([IGNORE_ID] * (1 + text_token_len[i]) + reject_speech_token[i, :reject_speech_token_len[i]].tolist() +
+                                            [self.speech_token_size]) for i in range(text_token.size(0))]
+            target_ids.extend(reject_target_ids)
+        target_ids = pad_sequence(target_ids, batch_first=True, padding_value=IGNORE_ID).to(device)
+
+        # 2. speech token projection
+        speech_emb = self.speech_embedding(speech_token)
+        if self.dpo:
+            reject_speech_emb = self.speech_embedding(reject_speech_token)
+
+        # 3. text token projection
+        text_token_lst = unpad_sequence(text_token, text_token_len, batch_first=True)
+        text_emb = [self.llm.model.model.embed_tokens(y) for y in text_token_lst]
+
+        # 4. prepare llm_input
+        speech_emb = unpad_sequence(speech_emb, speech_token_len.cpu(), batch_first=True)
+        input_emb = [torch.concat([sos_eos_emb.squeeze(dim=0), text_emb[i], task_id_emb.squeeze(dim=0), speech_emb[i]], dim=0)
+                     for i in range(len(text_emb))]
+        if self.dpo:
+            reject_speech_emb = unpad_sequence(reject_speech_emb, reject_speech_token_len.cpu(), batch_first=True)
+            reject_input_emb = [torch.concat([sos_eos_emb.squeeze(dim=0), text_emb[i], task_id_emb.squeeze(dim=0), reject_speech_emb[i]], dim=0)
+                                for i in range(len(text_emb))]
+            input_emb.extend(reject_input_emb)
+        input_emb_lengths = torch.tensor([i.size(0) for i in input_emb], dtype=torch.int32).to(device)
+        input_emb = pad_sequence(input_emb, batch_first=True, padding_value=IGNORE_ID).to(device)
+
+        attention_mask = ~make_pad_mask(input_emb_lengths)
+
+        result = self.llm.model(
+            inputs_embeds=input_emb,
+            attention_mask=attention_mask,
+            return_dict=True
+        )
+        hidden_states = result.hidden_states
+        logits = self.llm_decoder(hidden_states[-1])
+        loss = self.criterion_ce(logits[: speech_token.shape[0]], target_ids[: speech_token.shape[0]])
+        acc = th_accuracy(
+            logits[: speech_token.shape[0]].view(-1, self.speech_token_size + 3),
+            target_ids[: speech_token.shape[0]],
+            ignore_label=IGNORE_ID,
+        )
+        if not self.dpo:
+            return {
+                "loss": loss,
+                "acc": acc,
+            }
+        else:
+            all_logps_sum, all_logps_mean = self.get_batch_logps(
+                logits, target_ids, attention_mask, text_token_len, average_log_prob=False, ignore_id=IGNORE_ID
+            )
+            chosen_logps = all_logps_sum[: speech_token.shape[0]]
+            rejected_logps = all_logps_sum[speech_token.shape[0]:]
+            return {
+                "loss": loss,
+                "acc": acc,
+                "chosen_logps": chosen_logps,
+                "rejected_logps": rejected_logps
+            }
+
+
+    def get_batch_logps(
+        self,
+        logits: torch.FloatTensor,
+        labels: torch.LongTensor,
+        attention_mask,
+        prompt_token_lens,
+        average_log_prob: bool = False,
+        ignore_id: int = -1,
+    ) -> torch.FloatTensor:
+        """Compute the log probabilities of the given labels under the given logits.
+
+        Args:
+            logits: Logits of the model (unnormalized). Shape: (batch_size, sequence_length, vocab_size)
+            labels: Labels for which to compute the log probabilities. Label tokens with a value of -100 are ignored. Shape: (batch_size, sequence_length)
+            average_log_prob: If True, return the average log probability per (non-masked) token. Otherwise, return the sum of the log probabilities of the (non-masked) tokens.
+
+        Returns:
+            A tensor of shape (batch_size,) containing the average/sum log probabilities of the given labels under the given logits.
+        """
+        assert average_log_prob == False
+        assert logits.shape[:-1] == labels.shape
+        labels = labels[:, 1:].clone()
+        logits = logits[:, :-1, :]
+        loss_masks = attention_mask.clone().bool()
+        # mask prompts
+        for mask, text_token_len in zip(loss_masks, prompt_token_lens):
+            mask[:text_token_len + 1] = False
+        loss_masks = loss_masks[:, 1:]
+        labels[loss_masks == False] = 0
+        # dummy token; we'll ignore the losses on these tokens later
+        ignore = labels == ignore_id
+        labels = labels.masked_fill(ignore, 0)  # avoid -1 index
+        per_token_logps = torch.gather(logits.log_softmax(-1), dim=2, index=labels.unsqueeze(2)).squeeze(2)   # (bs, time,)
+        logprobs_sums = (per_token_logps * loss_masks).sum(-1)
+        logprobs_means = (per_token_logps * loss_masks).sum(-1) / loss_masks.sum(-1)
+        return logprobs_sums, logprobs_means
+
+
+    @torch.inference_mode()
+    def inference(
+            self,
+            text: torch.Tensor,
+            text_len: torch.Tensor,
+            prompt_text: torch.Tensor,
+            prompt_text_len: torch.Tensor,
+            prompt_speech_token: torch.Tensor,
+            prompt_speech_token_len: torch.Tensor,
+            embedding: torch.Tensor,
+            sampling: int = 25,
+            max_token_text_ratio: float = 20,
+            min_token_text_ratio: float = 2,
+    ) -> Generator[torch.Tensor, None, None]:
+        device = text.device
+        text = torch.concat([prompt_text, text], dim=1)
+        text_len += prompt_text_len
+        text = self.llm.model.model.embed_tokens(text)
+
+        # 3. concat llm_input
+        sos_eos_emb = self.llm_embedding.weight[self.sos_eos].reshape(1, 1, -1)
+        task_id_emb = self.llm_embedding.weight[self.task_id].reshape(1, 1, -1)
+        if prompt_speech_token_len != 0:
+            prompt_speech_token_emb = self.speech_embedding(prompt_speech_token)
+        else:
+            prompt_speech_token_emb = torch.zeros(1, 0, self.llm_input_size, dtype=text.dtype).to(device)
+        lm_input = torch.concat([sos_eos_emb, text, task_id_emb, prompt_speech_token_emb], dim=1)
+
+        # 4. cal min/max_length
+        min_len = int((text_len - prompt_text_len) * min_token_text_ratio)
+        max_len = int((text_len - prompt_text_len) * max_token_text_ratio)
+
+        # 5. step by step decode
+        out_tokens = []
+        cache = None
+        for i in range(max_len):
+            y_pred, cache = self.llm.forward_one_step(lm_input,
+                                                      masks=torch.tril(torch.ones((1, lm_input.shape[1], lm_input.shape[1]), device=lm_input.device)).to(torch.bool),
+                                                      cache=cache)
+            logp = self.llm_decoder(y_pred[:, -1]).log_softmax(dim=-1)
+            top_ids = self.sampling_ids(logp.squeeze(dim=0), out_tokens, sampling, ignore_eos=True if i < min_len else False).item()
+            if top_ids == self.speech_token_size:
+                break
+            if top_ids > self.speech_token_size:
+                continue
+            # in stream mode, yield token one by one
+            yield top_ids
+            out_tokens.append(top_ids)
+            lm_input = self.speech_embedding.weight[top_ids].reshape(1, 1, -1)
+
+    @torch.inference_mode()
+    def inference_bistream(
+            self,
+            text: Generator,
+            prompt_text: torch.Tensor,
+            prompt_text_len: torch.Tensor,
+            prompt_speech_token: torch.Tensor,
+            prompt_speech_token_len: torch.Tensor,
+            embedding: torch.Tensor,
+            sampling: int = 25,
+            max_token_text_ratio: float = 20,
+            min_token_text_ratio: float = 2,
+    ) -> Generator[torch.Tensor, None, None]:
+
+        device = prompt_text.device
+        # 1. prepare input
+        sos_eos_emb = self.llm_embedding.weight[self.sos_eos].reshape(1, 1, -1)
+        task_id_emb = self.llm_embedding.weight[self.task_id].reshape(1, 1, -1)
+        if prompt_speech_token_len != 0:
+            prompt_speech_token_emb = self.speech_embedding(prompt_speech_token)
+        else:
+            prompt_speech_token_emb = torch.zeros(1, 0, self.llm_input_size, dtype=prompt_text.dtype).to(device)
+        lm_input = torch.concat([sos_eos_emb], dim=1)
+
+        # 2. iterate text
+        out_tokens = []
+        cache = None
+        # NOTE init prompt_text as text_cache as it is basically impossible prompt_speech_token/prompt_text < 15/5
+        text_cache = self.llm.model.model.embed_tokens(prompt_text)
+        next_fill_index = -1
+        for this_text in text:
+            text_cache = torch.concat([text_cache, self.llm.model.model.embed_tokens(this_text)], dim=1)
+            # prompt_speech_token_emb not empty, try append to lm_input
+            while prompt_speech_token_emb.size(1) != 0:
+                if text_cache.size(1) >= self.mix_ratio[0]:
+                    lm_input_text, lm_input_speech = text_cache[:, :self.mix_ratio[0]], prompt_speech_token_emb[:, :self.mix_ratio[1]]
+                    logging.info('append {} text token {} speech token'.format(lm_input_text.size(1), lm_input_speech.size(1)))
+                    lm_input = torch.concat([lm_input, lm_input_text, lm_input_speech], dim=1)
+                    text_cache, prompt_speech_token_emb = text_cache[:, self.mix_ratio[0]:], prompt_speech_token_emb[:, self.mix_ratio[1]:]
+                else:
+                    logging.info('not enough text token to decode, wait for more')
+                    break
+            # no prompt_speech_token_emb remain, can decode some speech token
+            if prompt_speech_token_emb.size(1) == 0:
+                if (len(out_tokens) != 0 and out_tokens[-1] == self.speech_token_size + 2) or (len(out_tokens) == 0 and lm_input.size(1) == 1):
+                    logging.info('get fill token, need to append more text token')
+                    if text_cache.size(1) >= self.mix_ratio[0]:
+                        lm_input_text = text_cache[:, :self.mix_ratio[0]]
+                        logging.info('append {} text token'.format(lm_input_text.size(1)))
+                        if len(out_tokens) != 0 and out_tokens[-1] == self.speech_token_size + 2:
+                            lm_input = lm_input_text
+                        else:
+                            lm_input = torch.concat([lm_input, lm_input_text], dim=1)
+                        text_cache = text_cache[:, self.mix_ratio[0]:]
+                    else:
+                        logging.info('not enough text token to decode, wait for more')
+                        continue
+                while True:
+                    seq_len = lm_input.shape[1] if cache is None else lm_input.shape[1] + cache[0][0].size(2)
+                    y_pred, cache = self.llm.forward_one_step(lm_input,
+                                                masks=torch.tril(torch.ones((1, seq_len, seq_len), device=lm_input.device)).to(torch.bool),
+                                                cache=cache)
+                    logp = self.llm_decoder(y_pred[:, -1]).log_softmax(dim=-1)
+                    if next_fill_index != -1 and len(out_tokens) == next_fill_index:
+                        top_ids = self.speech_token_size + 2
+                        next_fill_index += (self.mix_ratio[1] + 1)
+                    else:
+                        top_ids = self.sampling_ids(logp.squeeze(dim=0), out_tokens, sampling, ignore_eos=True).item()
+                    if top_ids == self.speech_token_size + 2:
+                        next_fill_index = len(out_tokens) + self.mix_ratio[1] + 1
+                        logging.info('fill_token index {} next fill_token index {}'.format(len(out_tokens), next_fill_index))
+                    out_tokens.append(top_ids)
+                    if top_ids >= self.speech_token_size:
+                        if top_ids == self.speech_token_size + 2:
+                            break
+                        else:
+                            raise ValueError('should not get token {}'.format(top_ids))
+                    yield top_ids
+                    lm_input = self.speech_embedding.weight[top_ids].reshape(1, 1, -1)
+
+        # 3. final decode
+        lm_input = torch.concat([lm_input, text_cache, task_id_emb], dim=1)
+        logging.info('no more text token, decode until met eos')
+        while True:
+            seq_len = lm_input.shape[1] if cache is None else lm_input.shape[1] + cache[0][0].size(2)
+            y_pred, cache = self.llm.forward_one_step(lm_input,
+                                                      masks=torch.tril(torch.ones((1, seq_len, seq_len), device=lm_input.device)).to(torch.bool),
+                                                      cache=cache)
+            logp = self.llm_decoder(y_pred[:, -1]).log_softmax(dim=-1)
+            top_ids = self.sampling_ids(logp.squeeze(dim=0), out_tokens, sampling, ignore_eos=False).item()
+            out_tokens.append(top_ids)
+            if top_ids >= self.speech_token_size:
+                if top_ids == self.speech_token_size:
+                    break
+                else:
+                    raise ValueError('should not get token {}'.format(top_ids))
+            # in stream mode, yield token one by one
+            yield top_ids
+            lm_input = self.speech_embedding.weight[top_ids].reshape(1, 1, -1)

cosyvoice/llm/llm_vllm.py

@@ -0,0 +1,212 @@
+# Copyright (c) 2024 Alibaba Inc (authors: Xiang Lyu, Zhihao Du)
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+import time
+import queue
+import asyncio
+import threading
+from typing import List, Generator, AsyncGenerator
+import torch
+from cosyvoice.utils.file_utils import logging
+from cosyvoice.llm.llm import Qwen2LM
+
+# 启用vllm V1版本
+import os
+os.environ["VLLM_USE_V1"] = '1'
+from vllm import ModelRegistry
+from vllm import LLMEngine, AsyncLLMEngine, CompletionOutput
+from vllm.engine.arg_utils import EngineArgs, AsyncEngineArgs
+from vllm.sampling_params import SamplingParams
+
+from cosyvoice.llm.vllm_use_cosyvoice2_model import CosyVoice2Model as CosyVoice2LLM
+ModelRegistry.register_model("CosyVoice2Model", CosyVoice2LLM)
+
+# EngineArgs
+ENGINE_ARGS = {
+    "block_size": 16,
+    "swap_space": 0,
+    # "enforce_eager": True,
+    "gpu_memory_utilization": 0.4,
+    "max_num_batched_tokens": 1024,
+    "max_model_len": 1024,
+    "max_num_seqs": 256,
+    "disable_log_requests": True,
+    "disable_log_stats": True,
+    "dtype": "float16"
+}
+
+from vllm.sampling_params import RequestOutputKind
+# SamplingParams
+SAMPLING_PARAMS = {
+    "temperature": 1,  # 不能低于0.8, 否则会生成非常多的空音频，或者无法正常生成语音Token
+    "top_p": 1,       # 不能低于0.8, 否则会生成非常多的空音频，或者无法正常生成语音Token
+    "top_k": 25,
+    # "min_tokens": 80,       # 不支持设置最小的tokens数量设置，开启后vllm直接崩溃，无法启动
+    # "presence_penalty": 1.0,    # 不支持设置
+    # "frequency_penalty": 0.0,   # 不支持设置
+    "max_tokens": 1024,
+    "detokenize": False,          # 目前 vllm 0.7.3 v1版本中设置无效，待后续版本更新后减少计算
+    "ignore_eos": False,
+    "output_kind": RequestOutputKind.DELTA  # 设置为DELTA，如调整该参数，请同时调整llm_inference的处理代码
+}
+
+def tensor_to_list(tensor: torch.tensor):
+    return tensor.view(-1).cpu().numpy().tolist()
+
+class VllmQwen2LM(Qwen2LM):
+    def __init__(
+            self,
+            model_dir,
+            mix_ratio: List[int] = [5, 15],
+    ):
+        self.fp16 = False
+        self.half = lambda: None
+        self.mix_ratio = mix_ratio
+        # ---------------------------------------------
+        # vllm engine 的参数配置
+        engine_args = AsyncEngineArgs(
+            model=model_dir,
+            **ENGINE_ARGS,
+        )
+        self.llm_engine: AsyncLLMEngine = AsyncLLMEngine.from_engine_args(engine_args)
+
+        self.speech_token_size = 6564       # 6561 + 3
+        self.llm_token_size = 151936        # llm  vocab_size
+        self.sos_eos_token_id = self.speech_token_size + self.llm_token_size + 1
+        self.task_token_id = self.sos_eos_token_id + 1
+        self.zero_token_id = self.task_token_id + 1
+
+        # vllm 的推理任务需要在一个固定的事件循环中，因此启动一个后台线程运行转用于推理任务
+        self.loop = asyncio.new_event_loop()
+        self.loop_thread = threading.Thread(target=self._run_event_loop, daemon=True)
+        self.loop_thread.start()
+
+    def _run_event_loop(self):
+        asyncio.set_event_loop(self.loop)
+        self.loop.run_forever()
+
+    async def async_llm_inference(self, out_queue, prompt_token_ids, request_id, stop_token_ids, max_tokens):
+        sampling_params = SamplingParams(**SAMPLING_PARAMS)
+        sampling_params.stop_token_ids = stop_token_ids or [6561]
+        if max_tokens:
+            sampling_params.max_tokens = max_tokens
+        async for output in self.llm_engine.generate(
+                {
+                    "prompt_token_ids": prompt_token_ids,
+                },
+                sampling_params=sampling_params,
+                request_id=request_id or f"{time.time()}",
+        ):
+            out_queue.put((output.outputs[0], output.finished))
+
+    def llm_inference(self, prompt_token_ids: List[int], request_id: str=None, stop_token_ids=None, max_tokens=None):
+        out_queue = queue.Queue()
+        asyncio.run_coroutine_threadsafe(
+            self.async_llm_inference(out_queue, prompt_token_ids, request_id, stop_token_ids, max_tokens), self.loop
+        )
+        # 接收 out_queue 返回的结果
+        finished = False
+        while not finished:
+            (output, finished) = out_queue.get_nowait() if not out_queue.empty() else out_queue.get()
+            yield output
+
+    def inference(
+            self,
+            text: torch.Tensor,
+            text_len: torch.Tensor,
+            prompt_text: torch.Tensor,
+            prompt_text_len: torch.Tensor,
+            prompt_speech_token: torch.Tensor,
+            prompt_speech_token_len: torch.Tensor,
+            embedding: torch.Tensor,
+            sampling: int = 25,
+            max_token_text_ratio: float = 20,
+            min_token_text_ratio: float = 2,
+    ) -> Generator[torch.Tensor|int, None, None]:
+        prompt_text = tensor_to_list(prompt_text + torch.tensor(6564))
+        prompt_speech_token = tensor_to_list(prompt_speech_token)
+
+        text = tensor_to_list(text + torch.tensor(6564))
+        prompt_token_ids = [self.sos_eos_token_id] + prompt_text + text + \
+                           [self.task_token_id] + prompt_speech_token
+        max_tokens = len(text) * 20
+        for output in self.llm_inference(
+                prompt_token_ids,
+                stop_token_ids=[6561],
+                max_tokens=max_tokens,
+        ):
+            if output.token_ids[-1] == 6561:
+                need_add_tokens = output.token_ids[:-1]
+            else:
+                need_add_tokens = output.token_ids
+            for token in need_add_tokens:
+                yield token
+
+    def inference_bistream(
+            self,
+            text: Generator,
+            prompt_text: torch.Tensor,
+            prompt_text_len: torch.Tensor,
+            prompt_speech_token: torch.Tensor,
+            prompt_speech_token_len: torch.Tensor,
+            embedding: torch.Tensor,
+            sampling: int = 25,
+            max_token_text_ratio: float = 20,
+            min_token_text_ratio: float = 2,
+    ) -> Generator[torch.Tensor, None, None]:
+        prompt_text = tensor_to_list(prompt_text + torch.tensor(6564))
+        prompt_speech_token = tensor_to_list(prompt_speech_token)
+
+        last_tokens = []
+        prompt_token_ids = [self.sos_eos_token_id]
+        text_tokens_cache = prompt_text
+        for this_text in text:
+            this_text = tensor_to_list(this_text + torch.tensor(6564))
+            # text need tokens
+            assert isinstance(this_text, list), "text need token ids List[int]."
+            text_tokens_cache += this_text
+            while len(prompt_speech_token) != 0:
+                if len(text_tokens_cache) >= self.mix_ratio[0]:
+                    text_input_token = text_tokens_cache[:self.mix_ratio[0]]
+                    speech_input_token = prompt_speech_token[:self.mix_ratio[1]]
+                    prompt_token_ids += text_input_token + speech_input_token
+                    # reset the last cache
+                    text_tokens_cache = text_tokens_cache[self.mix_ratio[0]:]
+                    prompt_speech_token = prompt_speech_token[self.mix_ratio[1]:]
+                else:
+                    break
+            if len(prompt_speech_token) == 0:
+                if (len(last_tokens) > 0 and last_tokens[-1] == 6563) or len(prompt_token_ids) == 1:
+                    if len(text_tokens_cache) >= self.mix_ratio[0]:
+                        text_tokens_temp = text_tokens_cache[:self.mix_ratio[0]]
+                        prompt_token_ids += text_tokens_temp
+                        text_tokens_cache = text_tokens_cache[self.mix_ratio[0]:]
+                    else:
+                        continue
+                for output in self.llm_inference(prompt_token_ids, stop_token_ids=[6563]):
+                    last_tokens = output.token_ids
+                    if last_tokens[-1] == 6563:
+                        need_add_tokens = last_tokens[:-1]
+                    else:
+                        need_add_tokens = last_tokens
+                    for token in need_add_tokens:
+                        yield token
+                    prompt_token_ids.extend(need_add_tokens)
+        prompt_token_ids += text_tokens_cache + [self.task_token_id]
+        for output in self.llm_inference(prompt_token_ids, stop_token_ids=[6561]):
+            if output.token_ids[-1] == 6561:
+                need_add_tokens = output.token_ids[:-1]
+            else:
+                need_add_tokens = output.token_ids
+            for token in need_add_tokens:
+                yield token

cosyvoice/llm/vllm_use_cosyvoice2_model.py

@@ -0,0 +1,263 @@
+# SPDX-License-Identifier: Apache-2.0
+
+# Adapted from
+# https://github.com/huggingface/transformers/blob/v4.28.0/src/transformers/models/qwen2/modeling_qwen2.py
+# Copyright 2024 The Qwen team.
+# Copyright 2023 The vLLM team.
+# Copyright 2022 EleutherAI and the HuggingFace Inc. team. All rights reserved.
+#
+# This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX
+# and OPT implementations in this library. It has been modified from its
+# original forms to accommodate minor architectural differences compared
+# to GPT-NeoX and OPT used by the Meta AI team that trained the model.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""Inference-only Qwen2 model compatible with HuggingFace weights."""
+from typing import Iterable, List, Optional, Set, Tuple, Union, Iterator, overload, TypedDict, Mapping, Any
+from typing_extensions import TypeVar
+
+import torch
+from torch import nn
+
+from vllm.attention import AttentionMetadata
+from vllm.config import VllmConfig
+from vllm.logger import init_logger
+from vllm.model_executor.layers.logits_processor import LogitsProcessor
+from vllm.model_executor.layers.sampler import SamplerOutput, get_sampler
+from vllm.model_executor.layers.vocab_parallel_embedding import ParallelLMHead
+from vllm.model_executor.sampling_metadata import SamplingMetadata
+from vllm.sequence import IntermediateTensors
+
+from vllm.model_executor.models.interfaces import T
+from vllm.model_executor.models.qwen2 import Qwen2Model
+
+from vllm.model_executor.models.utils import AutoWeightsLoader, maybe_prefix, merge_multimodal_embeddings
+
+logger = init_logger(__name__)
+
+IGNORE_ID = -1
+
+
+class CosyVoice2Model(nn.Module):
+
+    packed_modules_mapping = {
+        "qkv_proj": [
+            "q_proj",
+            "k_proj",
+            "v_proj",
+        ],
+        "gate_up_proj": [
+            "gate_proj",
+            "up_proj",
+        ],
+    }
+
+    def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
+        super().__init__()
+        config = vllm_config.model_config.hf_config
+        quant_config = vllm_config.quant_config
+        lora_config = vllm_config.lora_config
+
+        self.config = config
+        self.lora_config = lora_config
+        self.quant_config = quant_config
+
+        self.llm_input_size = 896
+        self.llm_output_size = 896
+
+        self.speech_token_size = 6561+3
+        self.llm_token_size = config.vocab_size
+
+        # 2. build speech token language model related modules
+        self.sos_eos = 0
+        self.task_id = 1
+        self.fill_token = 2
+
+
+        self.allow_patterns_overrides = ["llm.*"]
+        self.llm_embedding = torch.nn.Embedding(2, self.llm_input_size)
+        self.model = Qwen2Model(vllm_config=vllm_config,
+                              prefix=maybe_prefix(prefix, "model"))
+
+        # self.llm_decoder = nn.Linear(self.llm_output_size, self.speech_token_size)
+        self.llm_decoder = ParallelLMHead(self.speech_token_size,
+                                      self.llm_output_size,
+                                      bias=True,
+                                      quant_config=quant_config,
+                                      prefix=maybe_prefix(
+                                          prefix, "llm_decoder"))
+        self.logits_processor = LogitsProcessor(self.speech_token_size)
+
+        # length_normalized_loss: bool = True,
+        # lsm_weight: float = 0.0,
+        # self.criterion_ce = LabelSmoothingLoss(
+        #     size=self.speech_token_size,
+        #     padding_idx=IGNORE_ID,
+        #     smoothing=lsm_weight,
+        #     normalize_length=length_normalized_loss,
+        # )
+
+        # 3. [Optional] build speech token related modules
+        self.speech_embedding = torch.nn.Embedding(self.speech_token_size, self.llm_input_size)
+
+        # 4. sampling method
+        ## use vllm sampling method
+        self.sampler = get_sampler()
+        self.make_empty_intermediate_tensors = (
+            self.model.make_empty_intermediate_tensors)
+
+        self.mix_ratio: List[int] = [5, 15]
+
+        # 定义特殊token常量
+        self.llm_token_id_delta = torch.tensor(self.speech_token_size, dtype=torch.int32)
+        self.sos_eos_token_id = torch.tensor((self.llm_token_id_delta + self.llm_token_size + 1), dtype=torch.int32)  # 163840 + 6564 = 170404
+        self.task_token_id = self.sos_eos_token_id + torch.tensor(1, dtype=torch.int32)  # 170405
+        self.zero_token_id = self.task_token_id + torch.tensor(1, dtype=torch.int32)
+
+        self.zero_embed_buffer = torch.zeros(
+            (vllm_config.scheduler_config.max_num_seqs, self.llm_input_size),
+            dtype=self.llm_embedding.weight.dtype,
+            device=self.llm_embedding.weight.device
+        )
+        self.inputs_embed_buffer = torch.zeros(
+            (vllm_config.scheduler_config.max_num_batched_tokens, self.llm_input_size),
+            dtype=self.llm_embedding.weight.dtype,
+            device=self.llm_embedding.weight.device,
+        )
+
+    def get_sos_eos_emb(self):
+        return self.llm_embedding.weight[self.sos_eos].reshape(1, 1, -1)
+
+    def get_task_id_emb(self):
+        return self.llm_embedding.weight[self.task_id].reshape(1, 1, -1)
+
+    def get_input_embeddings(
+        self,
+        input_ids: torch.Tensor,
+        multimodal_embeddings: Optional[T] = None,
+        attn_metadata: Optional["AttentionMetadata"] = None,
+    ) -> torch.Tensor:
+        """
+        Returns the input embeddings merged from the text embeddings from
+        input_ids and the multimodal embeddings generated from multimodal
+        kwargs.
+        """
+        # 创建掩码，标记哪些 token_id 属于音频 Token
+        mask = input_ids < self.speech_token_size
+
+        # 获取 input_ids 的原始形状
+        input_shape = input_ids.shape
+        # 展平 input_ids 和掩码以便统一处理
+        flat_input_ids = input_ids.view(-1)
+        flat_mask = mask.view(-1)
+
+        inputs_embeds = self.inputs_embed_buffer[:flat_input_ids.shape[0]]
+        inputs_embeds.zero_()
+
+        # Process speech tokens
+        if flat_mask.any():
+            speech_token_ids = flat_input_ids[flat_mask]
+            inputs_embeds[flat_mask] = self.speech_embedding(speech_token_ids)
+
+        # 处理大于 delta 的 token_id
+        if (~flat_mask).any():
+            llm_token_ids = flat_input_ids[~flat_mask]
+            llm_embeds = torch.zeros_like(inputs_embeds[~flat_mask])
+
+            sos_eos_mask = llm_token_ids == self.sos_eos_token_id
+            task_mask = llm_token_ids == self.task_token_id
+            zero_mask = llm_token_ids == self.zero_token_id
+            normal_mask = ~(sos_eos_mask | task_mask | zero_mask)
+
+            # 分层处理逻辑
+            # 第一优先级：SOS/EOS标记
+            if sos_eos_mask.any():
+                llm_embeds[sos_eos_mask] = self.llm_embedding.weight[self.sos_eos].unsqueeze(0)
+
+            # 第二优先级：任务标记
+            if task_mask.any():
+                llm_embeds[task_mask] = self.llm_embedding.weight[self.task_id].unsqueeze(0)
+
+            # 第二优先级：空音频标记
+            if zero_mask.any():
+                llm_embeds[zero_mask] = self.zero_embed_buffer[:len(llm_embeds[zero_mask])]
+
+            # 常规LLM token
+            if normal_mask.any():
+                original_ids = llm_token_ids[normal_mask] - self.llm_token_id_delta
+                # print('original_ids: ',original_ids)
+                llm_embeds[normal_mask] = self.model.get_input_embeddings(original_ids)
+
+            inputs_embeds[~flat_mask] = llm_embeds
+
+        inputs_embeds = inputs_embeds.view(*input_shape, self.llm_input_size)
+
+        # 合并多模态嵌入（如果有）
+        if multimodal_embeddings is not None:
+            inputs_embeds = merge_multimodal_embeddings(
+                input_ids, inputs_embeds, multimodal_embeddings,
+                self.config.audio_token_index
+            )
+        return inputs_embeds
+
+    def forward(
+        self,
+        input_ids: torch.Tensor,
+        positions: torch.Tensor,
+        kv_caches: List[torch.Tensor],
+        attn_metadata: AttentionMetadata,
+        intermediate_tensors: Optional[IntermediateTensors] = None,
+        inputs_embeds: Optional[torch.Tensor] = None,
+    ) -> Union[torch.Tensor, IntermediateTensors]:
+        if inputs_embeds is None:
+            inputs_embeds = self.get_input_embeddings(
+                input_ids,
+                attn_metadata=attn_metadata,
+            )
+        return self.model(input_ids, positions, kv_caches,
+                        attn_metadata, intermediate_tensors,
+                        inputs_embeds)
+
+    def compute_logits(
+        self,
+        hidden_states: torch.Tensor,
+        sampling_metadata: SamplingMetadata,
+    ) -> Optional[torch.Tensor]:
+        logits = self.logits_processor(self.llm_decoder, hidden_states,
+                                       sampling_metadata)
+        return logits
+
+    def sample(
+        self,
+        logits: torch.Tensor,
+        sampling_metadata: SamplingMetadata,
+    ) -> Optional[SamplerOutput]:
+        next_tokens = self.sampler(logits, sampling_metadata)
+        return next_tokens
+
+    @staticmethod
+    def convert_weights(weights: Iterable[Tuple[str, torch.Tensor]]) -> Iterable[Tuple[str, torch.Tensor]]:
+        for name, param in weights:
+            # 处理Qwen2Model核心参数
+            if name.startswith("llm."):
+                if name.startswith("llm.model.model."):
+                    name = name.replace("llm.model.model.", "model.")
+                else:
+                    continue
+            # print('weights name: ', name)
+            yield name, param
+
+    def load_weights(self, weights: Iterable[Tuple[str, torch.Tensor]]):
+        weights = self.convert_weights(weights)
+        loader = AutoWeightsLoader(self)
+        loader.load_weights(weights)

cosyvoice/transformer/upsample_encoder.py

@@ -56,16 +56,11 @@ class Upsample1D(nn.Module):
         # In this mode, first repeat interpolate, than conv with stride=1
         self.conv = nn.Conv1d(self.channels, self.out_channels, stride * 2 + 1, stride=1, padding=0)
 
-    def forward(self, inputs: torch.Tensor, input_lengths: torch.Tensor, conv_cache: torch.Tensor = torch.zeros(0, 0, 0)) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
+    def forward(self, inputs: torch.Tensor, input_lengths: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor]:
         outputs = F.interpolate(inputs, scale_factor=float(self.stride), mode="nearest")
-        if conv_cache.size(2) == 0:
-            outputs = F.pad(outputs, (self.stride * 2, 0), value=0.0)
-        else:
-            assert conv_cache.size(2) == self.stride * 2
-            outputs = torch.concat([conv_cache, outputs], dim=2)
-        conv_cache_new = outputs[:, :, -self.stride * 2:]
+        outputs = F.pad(outputs, (self.stride * 2, 0), value=0.0)
         outputs = self.conv(outputs)
-        return outputs, input_lengths * self.stride, conv_cache_new
+        return outputs, input_lengths * self.stride
 
 
 class PreLookaheadLayer(nn.Module):
@@ -83,7 +78,7 @@ class PreLookaheadLayer(nn.Module):
             kernel_size=3, stride=1, padding=0,
         )
 
-    def forward(self, inputs: torch.Tensor, context: torch.Tensor = torch.zeros(0, 0, 0), conv2_cache: torch.Tensor = torch.zeros(0, 0, 0)) -> Tuple[torch.Tensor, torch.Tensor]:
+    def forward(self, inputs: torch.Tensor, context: torch.Tensor = torch.zeros(0, 0, 0)) -> torch.Tensor:
         """
         inputs: (batch_size, seq_len, channels)
         """
@@ -93,22 +88,18 @@ class PreLookaheadLayer(nn.Module):
         if context.size(2) == 0:
             outputs = F.pad(outputs, (0, self.pre_lookahead_len), mode='constant', value=0.0)
         else:
+            assert self.training is False, 'you have passed context, make sure that you are running inference mode'
             assert context.size(2) == self.pre_lookahead_len
             outputs = F.pad(torch.concat([outputs, context], dim=2), (0, self.pre_lookahead_len - context.size(2)), mode='constant', value=0.0)
         outputs = F.leaky_relu(self.conv1(outputs))
         # outputs
-        if conv2_cache.size(2) == 0:
-            outputs = F.pad(outputs, (self.conv2.kernel_size[0] - 1, 0), mode='constant', value=0.0)
-        else:
-            assert conv2_cache.size(2) == self.conv2.kernel_size[0] - 1
-            outputs = torch.concat([conv2_cache, outputs], dim=2)
-        conv2_cache_new = outputs[:, :, -(self.conv2.kernel_size[0] - 1):]
+        outputs = F.pad(outputs, (self.conv2.kernel_size[0] - 1, 0), mode='constant', value=0.0)
         outputs = self.conv2(outputs)
         outputs = outputs.transpose(1, 2).contiguous()
 
         # residual connection
         outputs = outputs + inputs
-        return outputs, conv2_cache_new
+        return outputs
 
 
 class UpsampleConformerEncoder(torch.nn.Module):
@@ -253,6 +244,7 @@ class UpsampleConformerEncoder(torch.nn.Module):
         self,
         xs: torch.Tensor,
         xs_lens: torch.Tensor,
+        context: torch.Tensor = torch.zeros(0, 0, 0),
         decoding_chunk_size: int = 0,
         num_decoding_left_chunks: int = -1,
         streaming: bool = False,
@@ -280,20 +272,27 @@ class UpsampleConformerEncoder(torch.nn.Module):
             checkpointing API because `__call__` attaches all the hooks of the module.
             https://discuss.pytorch.org/t/any-different-between-model-input-and-model-forward-input/3690/2
         """
+        if hasattr(self, 'streaming'):
+            assert self.training is False, 'you have self.streaming attr, make sure that you are running inference mode'
+            streaming = self.streaming
         T = xs.size(1)
         masks = ~make_pad_mask(xs_lens, T).unsqueeze(1)  # (B, 1, T)
         if self.global_cmvn is not None:
             xs = self.global_cmvn(xs)
         xs, pos_emb, masks = self.embed(xs, masks)
+        if context.size(1) != 0:
+            assert self.training is False, 'you have passed context, make sure that you are running inference mode'
+            context_masks = torch.ones(1, 1, context.size(1)).to(masks)
+            context, _, _ = self.embed(context, context_masks, offset=xs.size(1))
         mask_pad = masks  # (B, 1, T/subsample_rate)
         chunk_masks = add_optional_chunk_mask(xs, masks, False, False, 0, self.static_chunk_size if streaming is True else 0, -1)
         # lookahead + conformer encoder
-        xs, _ = self.pre_lookahead_layer(xs)
+        xs = self.pre_lookahead_layer(xs, context=context)
         xs = self.forward_layers(xs, chunk_masks, pos_emb, mask_pad)
 
         # upsample + conformer encoder
         xs = xs.transpose(1, 2).contiguous()
-        xs, xs_lens, _ = self.up_layer(xs, xs_lens)
+        xs, xs_lens = self.up_layer(xs, xs_lens)
         xs = xs.transpose(1, 2).contiguous()
         T = xs.size(1)
         masks = ~make_pad_mask(xs_lens, T).unsqueeze(1)  # (B, 1, T)
@@ -322,100 +321,3 @@ class UpsampleConformerEncoder(torch.nn.Module):
         for layer in self.up_encoders:
             xs, chunk_masks, _, _ = layer(xs, chunk_masks, pos_emb, mask_pad)
         return xs
-
-    @torch.jit.export
-    def forward_chunk(
-        self,
-        xs: torch.Tensor,
-        xs_lens: torch.Tensor,
-        offset: int = 0,
-        context: torch.Tensor = torch.zeros(0, 0, 0),
-        pre_lookahead_layer_conv2_cache: torch.Tensor = torch.zeros(0, 0, 0),
-        encoders_kv_cache: torch.Tensor = torch.zeros(0, 0, 0, 0, 0),
-        upsample_offset: int = 0,
-        upsample_conv_cache: torch.Tensor = torch.zeros(0, 0, 0),
-        upsample_kv_cache: torch.Tensor = torch.zeros(0, 0, 0, 0, 0)
-    ) -> Tuple[torch.Tensor, torch.Tensor, Tuple[int, torch.Tensor, torch.Tensor, int, torch.Tensor, torch.Tensor]]:
-        """Embed positions in tensor.
-
-        Args:
-            xs: padded input tensor (B, T, D)
-            xs_lens: input length (B)
-            decoding_chunk_size: decoding chunk size for dynamic chunk
-                0: default for training, use random dynamic chunk.
-                <0: for decoding, use full chunk.
-                >0: for decoding, use fixed chunk size as set.
-            num_decoding_left_chunks: number of left chunks, this is for decoding,
-            the chunk size is decoding_chunk_size.
-                >=0: use num_decoding_left_chunks
-                <0: use all left chunks
-        Returns:
-            encoder output tensor xs, and subsampled masks
-            xs: padded output tensor (B, T' ~= T/subsample_rate, D)
-            masks: torch.Tensor batch padding mask after subsample
-                (B, 1, T' ~= T/subsample_rate)
-        NOTE(xcsong):
-            We pass the `__call__` method of the modules instead of `forward` to the
-            checkpointing API because `__call__` attaches all the hooks of the module.
-            https://discuss.pytorch.org/t/any-different-between-model-input-and-model-forward-input/3690/2
-        """
-        assert xs.size(0) == 1
-        # tmp_masks is just for interface compatibility
-        tmp_masks = torch.ones(1,
-                               xs.size(1),
-                               device=xs.device,
-                               dtype=torch.bool)
-        tmp_masks = tmp_masks.unsqueeze(1)
-        if self.global_cmvn is not None:
-            xs = self.global_cmvn(xs)
-        # NOTE(xcsong): Before embed, shape(xs) is (b=1, time, mel-dim)
-        xs, pos_emb, _ = self.embed(xs, tmp_masks, offset)
-        offset += xs.size(1)
-        tmp_masks = torch.ones(1,
-                               context.size(1),
-                               device=context.device,
-                               dtype=torch.bool)
-        tmp_masks = tmp_masks.unsqueeze(1)
-        if context.size(1) != 0:
-            context, _, _ = self.embed(context, tmp_masks, offset)
-
-        # lookahead + conformer encoder
-        xs, pre_lookahead_layer_conv2_cache = self.pre_lookahead_layer(xs, context, pre_lookahead_layer_conv2_cache)
-        # NOTE in cache mode we do not need to call add_optional_chunk_mask
-        chunk_masks = torch.ones((1, xs.size(1), offset), dtype=torch.bool, device=xs.device)
-        mask_pad = torch.ones((0, 0, 0), dtype=torch.bool, device=xs.device)
-        encoders_kv_cache_list = []
-        for index, layer in enumerate(self.encoders):
-            xs, chunk_masks, encoders_kv_cache_new, _ = layer(xs, chunk_masks, pos_emb, mask_pad, encoders_kv_cache[index])
-            encoders_kv_cache_list.append(encoders_kv_cache_new)
-        encoders_kv_cache = torch.stack(encoders_kv_cache_list, dim=0)
-
-        # upsample
-        xs = xs.transpose(1, 2).contiguous()
-        xs, xs_lens, upsample_conv_cache = self.up_layer(xs, xs_lens, upsample_conv_cache)
-        xs = xs.transpose(1, 2).contiguous()
-
-        # tmp_masks is just for interface compatibility
-        tmp_masks = torch.ones(1,
-                               xs.size(1),
-                               device=xs.device,
-                               dtype=torch.bool)
-        tmp_masks = tmp_masks.unsqueeze(1)
-        xs, pos_emb, masks = self.up_embed(xs, tmp_masks, upsample_offset)
-        upsample_offset += xs.size(1)
-
-        # conformer encoder
-        chunk_masks = torch.ones((1, xs.size(1), upsample_offset), dtype=torch.bool, device=xs.device)
-        mask_pad = torch.ones((0, 0, 0), dtype=torch.bool, device=xs.device)
-        upsample_kv_cache_list = []
-        for index, layer in enumerate(self.up_encoders):
-            xs, chunk_masks, upsample_kv_cache_new, _ = layer(xs, chunk_masks, pos_emb, mask_pad, upsample_kv_cache[index])
-            upsample_kv_cache_list.append(upsample_kv_cache_new)
-        upsample_kv_cache = torch.stack(upsample_kv_cache_list, dim=0)
-
-        if self.normalize_before:
-            xs = self.after_norm(xs)
-        # Here we assume the mask is not changed in encoder layers, so just
-        # return the masks before encoder layers, and the masks will be used
-        # for cross attention with decoder later
-        return xs, masks, (offset, pre_lookahead_layer_conv2_cache, encoders_kv_cache, upsample_offset, upsample_conv_cache, upsample_kv_cache)

cosyvoice/utils/common.py

@@ -1,5 +1,6 @@
 # Copyright (c) 2020 Mobvoi Inc (Binbin Zhang)
 #               2024 Alibaba Inc (authors: Xiang Lyu)
+#               2025 Alibaba Inc (authors: Xiang Lyu, Bofan Zhou)
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -15,6 +16,7 @@
 # Modified from ESPnet(https://github.com/espnet/espnet)
 """Unility functions for Transformer."""
 
+import queue
 import random
 from typing import List
 
@@ -164,3 +166,20 @@ def mask_to_bias(mask: torch.Tensor, dtype: torch.dtype) -> torch.Tensor:
     #     chunk_masks = (1.0 - chunk_masks) * torch.finfo(dtype).min
     mask = (1.0 - mask) * -1.0e+10
     return mask
+
+
+class TrtContextWrapper:
+    def __init__(self, trt_engine, trt_concurrent=1):
+        self.trt_context_pool = queue.Queue()
+        self.trt_engine = trt_engine
+        for _ in range(trt_concurrent):
+            trt_context = trt_engine.create_execution_context()
+            assert trt_context is not None, 'failed to create trt context, maybe not enough CUDA memory, try reduce current trt concurrent {}'.format(trt_concurrent)
+            self.trt_context_pool.put(trt_context)
+        assert self.trt_context_pool.empty() is False, 'no avaialbe estimator context'
+
+    def acquire_estimator(self):
+        return self.trt_context_pool.get(), self.trt_engine
+
+    def release_estimator(self, context):
+        self.trt_context_pool.put(context)

cosyvoice/utils/executor_dpo.py

@@ -0,0 +1,184 @@
+# Copyright (c) 2020 Mobvoi Inc (Binbin Zhang)
+#               2024 Alibaba Inc (authors: Xiang Lyu)
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#   http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import logging
+from contextlib import nullcontext
+import os
+
+import torch
+import torch.distributed as dist
+
+from cosyvoice.utils.train_utils_dpo import update_parameter_and_lr, log_per_step, log_per_save, batch_forward, batch_backward, save_model, cosyvoice_join
+from cosyvoice.utils.losses_dpo import DPOLoss
+
+
+class Executor:
+
+    def __init__(self, gan: bool = False, dpo: bool = False, beta: float = 0.01, label_smoothing: float = 0.0, ipo: bool = False):
+        self.gan = gan
+        self.step = 0
+        self.epoch = 0
+        self.rank = int(os.environ.get('RANK', 0))
+        self.device = torch.device('cuda:{}'.format(self.rank))
+        self.dpo = dpo
+        if self.dpo:
+            self.dpo_loss = DPOLoss(beta, label_smoothing, ipo)
+        else:
+            self.dpo_loss = None
+
+    def train_one_epoc(self, model, optimizer, scheduler, train_data_loader, cv_data_loader, writer, info_dict, scaler, group_join, ref_model=None):
+        ''' Train one epoch
+        '''
+
+        lr = optimizer.param_groups[0]['lr']
+        logging.info('Epoch {} TRAIN info lr {} rank {}'.format(self.epoch, lr, self.rank))
+        logging.info('using accumulate grad, new batch size is {} times'
+                     ' larger than before'.format(info_dict['accum_grad']))
+        # A context manager to be used in conjunction with an instance of
+        # torch.nn.parallel.DistributedDataParallel to be able to train
+        # with uneven inputs across participating processes.
+        model.train()
+        if self.dpo:
+            assert ref_model is not None
+            ref_model.eval()
+        model_context = model.join if info_dict['train_engine'] == 'torch_ddp' else nullcontext
+        with model_context():
+            for batch_idx, batch_dict in enumerate(train_data_loader):
+                info_dict["tag"] = "TRAIN"
+                info_dict["step"] = self.step
+                info_dict["epoch"] = self.epoch
+                info_dict["batch_idx"] = batch_idx
+                if cosyvoice_join(group_join, info_dict):
+                    break
+
+                # Disable gradient synchronizations across DDP processes.
+                # Within this context, gradients will be accumulated on module
+                # variables, which will later be synchronized.
+                if info_dict['train_engine'] == 'torch_ddp' and (batch_idx + 1) % info_dict["accum_grad"] != 0:
+                    context = model.no_sync
+                # Used for single gpu training and DDP gradient synchronization
+                # processes.
+                else:
+                    context = nullcontext
+
+                with context():
+                    info_dict = batch_forward(model, batch_dict, scaler, info_dict, ref_model, self.dpo_loss)
+                    info_dict = batch_backward(model, scaler, info_dict)
+
+                info_dict = update_parameter_and_lr(model, optimizer, scheduler, scaler, info_dict)
+                log_per_step(writer, info_dict)
+                # NOTE specify save_per_step in cosyvoice.yaml if you want to enable step save
+                if info_dict['save_per_step'] > 0 and (self.step + 1) % info_dict['save_per_step'] == 0 and \
+                   (batch_idx + 1) % info_dict["accum_grad"] == 0:
+                    dist.barrier()
+                    self.cv(model, cv_data_loader, writer, info_dict, on_batch_end=False, ref_model=ref_model, dpo_loss=self.dpo_loss)
+                    model.train()
+                if (batch_idx + 1) % info_dict["accum_grad"] == 0:
+                    self.step += 1
+        dist.barrier()
+        self.cv(model, cv_data_loader, writer, info_dict, on_batch_end=True, ref_model=ref_model, dpo_loss=self.dpo_loss)
+
+    def train_one_epoc_gan(self, model, optimizer, scheduler, optimizer_d, scheduler_d, train_data_loader, cv_data_loader,
+                           writer, info_dict, scaler, group_join):
+        ''' Train one epoch
+        '''
+
+        lr = optimizer.param_groups[0]['lr']
+        logging.info('Epoch {} TRAIN info lr {} rank {}'.format(self.epoch, lr, self.rank))
+        logging.info('using accumulate grad, new batch size is {} times'
+                     ' larger than before'.format(info_dict['accum_grad']))
+        # A context manager to be used in conjunction with an instance of
+        # torch.nn.parallel.DistributedDataParallel to be able to train
+        # with uneven inputs across participating processes.
+        model.train()
+        model_context = model.join if info_dict['train_engine'] == 'torch_ddp' else nullcontext
+        with model_context():
+            for batch_idx, batch_dict in enumerate(train_data_loader):
+                info_dict["tag"] = "TRAIN"
+                info_dict["step"] = self.step
+                info_dict["epoch"] = self.epoch
+                info_dict["batch_idx"] = batch_idx
+                if cosyvoice_join(group_join, info_dict):
+                    break
+
+                # Disable gradient synchronizations across DDP processes.
+                # Within this context, gradients will be accumulated on module
+                # variables, which will later be synchronized.
+                if info_dict['train_engine'] == 'torch_ddp' and (batch_idx + 1) % info_dict["accum_grad"] != 0:
+                    context = model.no_sync
+                # Used for single gpu training and DDP gradient synchronization
+                # processes.
+                else:
+                    context = nullcontext
+
+                with context():
+                    batch_dict['turn'] = 'discriminator'
+                    info_dict = batch_forward(model, batch_dict, scaler, info_dict)
+                    info_dict = batch_backward(model, scaler, info_dict)
+                info_dict = update_parameter_and_lr(model, optimizer_d, scheduler_d, scaler, info_dict)
+                optimizer.zero_grad()
+                log_per_step(writer, info_dict)
+                with context():
+                    batch_dict['turn'] = 'generator'
+                    info_dict = batch_forward(model, batch_dict, scaler, info_dict)
+                    info_dict = batch_backward(model, scaler, info_dict)
+                info_dict = update_parameter_and_lr(model, optimizer, scheduler, scaler, info_dict)
+                optimizer_d.zero_grad()
+                log_per_step(writer, info_dict)
+                # NOTE specify save_per_step in cosyvoice.yaml if you want to enable step save
+                if info_dict['save_per_step'] > 0 and (self.step + 1) % info_dict['save_per_step'] == 0 and \
+                   (batch_idx + 1) % info_dict["accum_grad"] == 0:
+                    dist.barrier()
+                    self.cv(model, cv_data_loader, writer, info_dict, on_batch_end=False)
+                    model.train()
+                if (batch_idx + 1) % info_dict["accum_grad"] == 0:
+                    self.step += 1
+        dist.barrier()
+        self.cv(model, cv_data_loader, writer, info_dict, on_batch_end=True)
+
+    @torch.inference_mode()
+    def cv(self, model, cv_data_loader, writer, info_dict, on_batch_end=True, ref_model=None, dpo_loss=None):
+        ''' Cross validation on
+        '''
+        logging.info('Epoch {} Step {} on_batch_end {} CV rank {}'.format(self.epoch, self.step + 1, on_batch_end, self.rank))
+        model.eval()
+        if self.dpo:
+            assert ref_model is not None
+            ref_model.eval()
+        total_num_utts, total_loss_dict = 0, {}  # avoid division by 0
+        for batch_idx, batch_dict in enumerate(cv_data_loader):
+            info_dict["tag"] = "CV"
+            info_dict["step"] = self.step
+            info_dict["epoch"] = self.epoch
+            info_dict["batch_idx"] = batch_idx
+
+            num_utts = len(batch_dict["utts"])
+            total_num_utts += num_utts
+
+            if self.gan is True:
+                batch_dict['turn'] = 'generator'
+            info_dict = batch_forward(model, batch_dict, None, info_dict, ref_model, dpo_loss)
+
+            for k, v in info_dict['loss_dict'].items():
+                if k not in total_loss_dict:
+                    total_loss_dict[k] = []
+                total_loss_dict[k].append(v.item() * num_utts)
+            log_per_step(None, info_dict)
+        for k, v in total_loss_dict.items():
+            total_loss_dict[k] = sum(v) / total_num_utts
+        info_dict['loss_dict'] = total_loss_dict
+        log_per_save(writer, info_dict)
+        model_name = 'epoch_{}_whole'.format(self.epoch) if on_batch_end else 'epoch_{}_step_{}'.format(self.epoch, self.step + 1)
+        save_model(model, model_name, info_dict)

cosyvoice/utils/file_utils.py

@@ -56,7 +56,7 @@ def convert_onnx_to_trt(trt_model, trt_kwargs, onnx_model, fp16):
     network = builder.create_network(network_flags)
     parser = trt.OnnxParser(network, logger)
     config = builder.create_builder_config()
-    config.set_memory_pool_limit(trt.MemoryPoolType.WORKSPACE, 1 << 33)  # 8GB
+    config.set_memory_pool_limit(trt.MemoryPoolType.WORKSPACE, 1 << 31)  # 1GB
     if fp16:
         config.set_flag(trt.BuilderFlag.FP16)
     profile = builder.create_optimization_profile()

cosyvoice/utils/losses_dpo.py

@@ -0,0 +1,57 @@
+import torch
+import torch.nn.functional as F
+from typing import Tuple
+
+
+def tpr_loss(disc_real_outputs, disc_generated_outputs, tau):
+    loss = 0
+    for dr, dg in zip(disc_real_outputs, disc_generated_outputs):
+        m_DG = torch.median((dr - dg))
+        L_rel = torch.mean((((dr - dg) - m_DG) ** 2)[dr < dg + m_DG])
+        loss += tau - F.relu(tau - L_rel)
+    return loss
+
+
+def mel_loss(real_speech, generated_speech, mel_transforms):
+    loss = 0
+    for transform in mel_transforms:
+        mel_r = transform(real_speech)
+        mel_g = transform(generated_speech)
+        loss += F.l1_loss(mel_g, mel_r)
+    return loss
+
+
+class DPOLoss(torch.nn.Module):
+    """
+    DPO Loss
+    """
+
+    def __init__(self, beta: float, label_smoothing: float = 0.0, ipo: bool = False) -> None:
+        super().__init__()
+        self.beta = beta
+        self.label_smoothing = label_smoothing
+        self.ipo = ipo
+
+    def forward(
+        self,
+        policy_chosen_logps: torch.Tensor,
+        policy_rejected_logps: torch.Tensor,
+        reference_chosen_logps: torch.Tensor,
+        reference_rejected_logps: torch.Tensor,
+    ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
+        pi_logratios = policy_chosen_logps - policy_rejected_logps
+        ref_logratios = reference_chosen_logps - reference_rejected_logps
+        logits = pi_logratios - ref_logratios
+        if self.ipo:
+            losses = (logits - 1 / (2 * self.beta)) ** 2  # Eq. 17 of https://arxiv.org/pdf/2310.12036v2.pdf
+        else:
+            # Eq. 3 https://ericmitchell.ai/cdpo.pdf; label_smoothing=0 gives original DPO (Eq. 7 of https://arxiv.org/pdf/2305.18290.pdf)
+            losses = (
+                -F.logsigmoid(self.beta * logits) * (1 - self.label_smoothing)
+                - F.logsigmoid(-self.beta * logits) * self.label_smoothing
+            )
+        loss = losses.mean()
+        chosen_rewards = self.beta * (policy_chosen_logps - reference_chosen_logps).detach()
+        rejected_rewards = self.beta * (policy_rejected_logps - reference_rejected_logps).detach()
+
+        return loss, chosen_rewards, rejected_rewards

cosyvoice/utils/mask.py

@@ -86,7 +86,7 @@ def subsequent_mask(
     return mask
 
 
-def subsequent_chunk_mask(
+def subsequent_chunk_mask_deprecated(
         size: int,
         chunk_size: int,
         num_left_chunks: int = -1,
@@ -124,6 +124,40 @@ def subsequent_chunk_mask(
     return ret
 
 
+def subsequent_chunk_mask(
+        size: int,
+        chunk_size: int,
+        num_left_chunks: int = -1,
+        device: torch.device = torch.device("cpu"),
+) -> torch.Tensor:
+    """Create mask for subsequent steps (size, size) with chunk size,
+       this is for streaming encoder
+
+    Args:
+        size (int): size of mask
+        chunk_size (int): size of chunk
+        num_left_chunks (int): number of left chunks
+            <0: use full chunk
+            >=0: use num_left_chunks
+        device (torch.device): "cpu" or "cuda" or torch.Tensor.device
+
+    Returns:
+        torch.Tensor: mask
+
+    Examples:
+        >>> subsequent_chunk_mask(4, 2)
+        [[1, 1, 0, 0],
+         [1, 1, 0, 0],
+         [1, 1, 1, 1],
+         [1, 1, 1, 1]]
+    """
+    # NOTE this modified implementation meets onnx export requirements, but it doesn't support num_left_chunks
+    pos_idx = torch.arange(size, device=device)
+    block_value = (torch.div(pos_idx, chunk_size, rounding_mode='trunc') + 1) * chunk_size
+    ret = pos_idx.unsqueeze(0) < block_value.unsqueeze(1)
+    return ret
+
+
 def add_optional_chunk_mask(xs: torch.Tensor,
                             masks: torch.Tensor,
                             use_dynamic_chunk: bool,
@@ -196,9 +230,6 @@ def add_optional_chunk_mask(xs: torch.Tensor,
     else:
         chunk_masks = masks
     assert chunk_masks.dtype == torch.bool
-    if (chunk_masks.sum(dim=-1) == 0).sum().item() != 0:
-        print('get chunk_masks all false at some timestep, force set to true, make sure they are masked in futuer computation!')
-        chunk_masks[chunk_masks.sum(dim=-1) == 0] = True
     return chunk_masks
 
 

cosyvoice/utils/train_utils_dpo.py

@@ -0,0 +1,364 @@
+# Copyright (c) 2021 Mobvoi Inc. (authors: Binbin Zhang)
+#               2023 Horizon Inc. (authors: Xingchen Song)
+#               2024 Alibaba Inc (authors: Xiang Lyu)
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import logging
+import os
+import torch
+import json
+import re
+import datetime
+import yaml
+
+import deepspeed
+import torch.optim as optim
+import torch.distributed as dist
+
+from torch.utils.tensorboard import SummaryWriter
+from torch.utils.data import DataLoader
+from torch.nn.utils import clip_grad_norm_
+
+from deepspeed.runtime.zero.stage_1_and_2 import estimate_zero2_model_states_mem_needs_all_live
+
+from cosyvoice.dataset.dataset import Dataset
+from cosyvoice.utils.scheduler import WarmupLR, NoamHoldAnnealing, ConstantLR
+
+
+def init_distributed(args):
+    world_size = int(os.environ.get('WORLD_SIZE', 1))
+    local_rank = int(os.environ.get('LOCAL_RANK', 0))
+    rank = int(os.environ.get('RANK', 0))
+    logging.info('training on multiple gpus, this gpu {}'.format(local_rank) +
+                 ', rank {}, world_size {}'.format(rank, world_size))
+    if args.train_engine == 'torch_ddp':
+        torch.cuda.set_device(local_rank)
+        dist.init_process_group(args.dist_backend)
+    else:
+        deepspeed.init_distributed(dist_backend=args.dist_backend)
+    return world_size, local_rank, rank
+
+
+def init_dataset_and_dataloader(args, configs, gan):
+    data_pipeline = configs['data_pipeline_gan'] if gan is True else configs['data_pipeline']
+    train_dataset = Dataset(args.train_data, data_pipeline=data_pipeline, mode='train', gan=gan, shuffle=True, partition=True)
+    cv_dataset = Dataset(args.cv_data, data_pipeline=data_pipeline, mode='train', gan=gan, shuffle=False, partition=False)
+
+    # do not use persistent_workers=True, as whisper tokenizer opens tiktoken file each time when the for loop starts
+    train_data_loader = DataLoader(train_dataset,
+                                   batch_size=None,
+                                   pin_memory=args.pin_memory,
+                                   num_workers=args.num_workers,
+                                   prefetch_factor=args.prefetch)
+    cv_data_loader = DataLoader(cv_dataset,
+                                batch_size=None,
+                                pin_memory=args.pin_memory,
+                                num_workers=args.num_workers,
+                                prefetch_factor=args.prefetch)
+    return train_dataset, cv_dataset, train_data_loader, cv_data_loader
+
+
+def check_modify_and_save_config(args, configs):
+    if args.train_engine == "torch_ddp":
+        configs['train_conf']["dtype"] = 'fp32'
+    else:
+        with open(args.deepspeed_config, 'r') as fin:
+            ds_configs = json.load(fin)
+        if "fp16" in ds_configs and ds_configs["fp16"]["enabled"]:
+            configs['train_conf']["dtype"] = "fp16"
+        elif "bf16" in ds_configs and ds_configs["bf16"]["enabled"]:
+            configs['train_conf']["dtype"] = "bf16"
+        else:
+            configs['train_conf']["dtype"] = "fp32"
+        assert ds_configs["train_micro_batch_size_per_gpu"] == 1
+        # if use deepspeed, override ddp config
+        configs['train_conf']['save_per_step'] = int(configs['train_conf']['save_per_step'] *
+                                                     configs['train_conf']['accum_grad'] / ds_configs["gradient_accumulation_steps"])
+        configs['train_conf']['accum_grad'] = ds_configs["gradient_accumulation_steps"]
+        configs['train_conf']['grad_clip'] = ds_configs["gradient_clipping"]
+        configs['train_conf']['log_interval'] = ds_configs["steps_per_print"]
+    return configs
+
+
+def wrap_cuda_model(args, model):
+    local_world_size = int(os.environ.get('LOCAL_WORLD_SIZE', 1))
+    world_size = int(os.environ.get('WORLD_SIZE', 1))
+    if args.train_engine == "torch_ddp":  # native pytorch ddp
+        assert (torch.cuda.is_available())
+        model.cuda()
+        model = torch.nn.parallel.DistributedDataParallel(model, find_unused_parameters=True)
+    else:
+        if int(os.environ.get('RANK', 0)) == 0:
+            logging.info("Estimating model states memory needs (zero2)...")
+            estimate_zero2_model_states_mem_needs_all_live(
+                model,
+                num_gpus_per_node=local_world_size,
+                num_nodes=world_size // local_world_size)
+    return model
+
+
+def init_optimizer_and_scheduler(args, configs, model, gan):
+    if gan is False:
+        if configs['train_conf']['optim'] == 'adam':
+            optimizer = optim.Adam(model.parameters(), **configs['train_conf']['optim_conf'])
+        elif configs['train_conf']['optim'] == 'adamw':
+            optimizer = optim.AdamW(model.parameters(), **configs['train_conf']['optim_conf'])
+        else:
+            raise ValueError("unknown optimizer: " + configs['train_conf'])
+
+        if configs['train_conf']['scheduler'] == 'warmuplr':
+            scheduler_type = WarmupLR
+            scheduler = WarmupLR(optimizer, **configs['train_conf']['scheduler_conf'])
+        elif configs['train_conf']['scheduler'] == 'NoamHoldAnnealing':
+            scheduler_type = NoamHoldAnnealing
+            scheduler = NoamHoldAnnealing(optimizer, **configs['train_conf']['scheduler_conf'])
+        elif configs['train_conf']['scheduler'] == 'constantlr':
+            scheduler_type = ConstantLR
+            scheduler = ConstantLR(optimizer)
+        else:
+            raise ValueError("unknown scheduler: " + configs['train_conf'])
+
+        # use deepspeed optimizer for speedup
+        if args.train_engine == "deepspeed":
+            def scheduler(opt):
+                return scheduler_type(opt, **configs['train_conf']['scheduler_conf'])
+            model, optimizer, _, scheduler = deepspeed.initialize(
+                args=args,
+                model=model,
+                optimizer=None,
+                lr_scheduler=scheduler,
+                model_parameters=model.parameters())
+
+        optimizer_d, scheduler_d = None, None
+
+    else:
+        # currently we wrap generator and discriminator in one model, so we cannot use deepspeed
+        if configs['train_conf']['optim'] == 'adam':
+            optimizer = optim.Adam(model.module.generator.parameters(), **configs['train_conf']['optim_conf'])
+        elif configs['train_conf']['optim'] == 'adamw':
+            optimizer = optim.AdamW(model.module.generator.parameters(), **configs['train_conf']['optim_conf'])
+        else:
+            raise ValueError("unknown optimizer: " + configs['train_conf'])
+
+        if configs['train_conf']['scheduler'] == 'warmuplr':
+            scheduler_type = WarmupLR
+            scheduler = WarmupLR(optimizer, **configs['train_conf']['scheduler_conf'])
+        elif configs['train_conf']['scheduler'] == 'NoamHoldAnnealing':
+            scheduler_type = NoamHoldAnnealing
+            scheduler = NoamHoldAnnealing(optimizer, **configs['train_conf']['scheduler_conf'])
+        elif configs['train_conf']['scheduler'] == 'constantlr':
+            scheduler_type = ConstantLR
+            scheduler = ConstantLR(optimizer)
+        else:
+            raise ValueError("unknown scheduler: " + configs['train_conf'])
+
+        if configs['train_conf']['optim_d'] == 'adam':
+            optimizer_d = optim.Adam(model.module.discriminator.parameters(), **configs['train_conf']['optim_conf'])
+        elif configs['train_conf']['optim_d'] == 'adamw':
+            optimizer_d = optim.AdamW(model.module.discriminator.parameters(), **configs['train_conf']['optim_conf'])
+        else:
+            raise ValueError("unknown optimizer: " + configs['train_conf'])
+
+        if configs['train_conf']['scheduler_d'] == 'warmuplr':
+            scheduler_type = WarmupLR
+            scheduler_d = WarmupLR(optimizer_d, **configs['train_conf']['scheduler_conf'])
+        elif configs['train_conf']['scheduler_d'] == 'NoamHoldAnnealing':
+            scheduler_type = NoamHoldAnnealing
+            scheduler_d = NoamHoldAnnealing(optimizer_d, **configs['train_conf']['scheduler_conf'])
+        elif configs['train_conf']['scheduler'] == 'constantlr':
+            scheduler_type = ConstantLR
+            scheduler_d = ConstantLR(optimizer_d)
+        else:
+            raise ValueError("unknown scheduler: " + configs['train_conf'])
+    return model, optimizer, scheduler, optimizer_d, scheduler_d
+
+
+def init_summarywriter(args):
+    writer = None
+    if int(os.environ.get('RANK', 0)) == 0:
+        os.makedirs(args.model_dir, exist_ok=True)
+        writer = SummaryWriter(args.tensorboard_dir)
+    return writer
+
+
+def save_model(model, model_name, info_dict):
+    rank = int(os.environ.get('RANK', 0))
+    model_dir = info_dict["model_dir"]
+    save_model_path = os.path.join(model_dir, '{}.pt'.format(model_name))
+
+    if info_dict["train_engine"] == "torch_ddp":
+        if rank == 0:
+            torch.save({**model.module.state_dict(), 'epoch': info_dict['epoch'], 'step': info_dict['step']}, save_model_path)
+    else:
+        with torch.no_grad():
+            model.save_checkpoint(save_dir=model_dir,
+                                  tag=model_name,
+                                  client_state=info_dict)
+    if rank == 0:
+        info_path = re.sub('.pt$', '.yaml', save_model_path)
+        info_dict['save_time'] = datetime.datetime.now().strftime('%d/%m/%Y %H:%M:%S')
+        with open(info_path, 'w') as fout:
+            data = yaml.dump(info_dict)
+            fout.write(data)
+        logging.info('[Rank {}] Checkpoint: save to checkpoint {}'.format(rank, save_model_path))
+
+
+def cosyvoice_join(group_join, info_dict):
+    world_size = int(os.environ.get('WORLD_SIZE', 1))
+    local_rank = int(os.environ.get('LOCAL_RANK', 0))
+    rank = int(os.environ.get('RANK', 0))
+
+    if info_dict["batch_idx"] != 0:
+        # we try to join all rank in both ddp and deepspeed mode, in case different rank has different lr
+        try:
+            dist.monitored_barrier(group=group_join,
+                                   timeout=group_join.options._timeout)
+            return False
+        except RuntimeError as e:
+            logging.info("Detected uneven workload distribution: {}\n".format(e) +
+                         "Break current worker to manually join all workers, " +
+                         "world_size {}, current rank {}, current local_rank {}\n".
+                         format(world_size, rank, local_rank))
+            return True
+    else:
+        return False
+
+
+def batch_forward(model, batch, scaler, info_dict, ref_model=None, dpo_loss=None):
+    device = int(os.environ.get('LOCAL_RANK', 0))
+
+    dtype = info_dict["dtype"]
+    if dtype == "fp16":
+        dtype = torch.float16
+    elif dtype == "bf16":
+        dtype = torch.bfloat16
+    else:  # fp32
+        dtype = torch.float32
+
+    if info_dict['train_engine'] == 'torch_ddp':
+        autocast = torch.cuda.amp.autocast(enabled=scaler is not None)
+    else:
+        autocast = torch.cuda.amp.autocast(enabled=True, dtype=dtype, cache_enabled=False)
+
+    with autocast:
+        info_dict['loss_dict'] = model(batch, device)
+        if ref_model and dpo_loss:
+            chosen_logps = info_dict['loss_dict']["chosen_logps"]
+            rejected_logps = info_dict['loss_dict']["rejected_logps"]
+            sft_loss = info_dict['loss_dict']['loss']
+            with torch.no_grad():
+                ref_model = ref_model.to(device)
+                ref_loss_dict = ref_model(batch, device)
+            reference_chosen_logps = ref_loss_dict["chosen_logps"]
+            reference_rejected_logps = ref_loss_dict["rejected_logps"]
+            preference_loss, chosen_reward, reject_reward = dpo_loss(
+                chosen_logps, rejected_logps, reference_chosen_logps, reference_rejected_logps
+            )
+            dpo_acc = (chosen_reward > reject_reward).float().mean()
+            info_dict['loss_dict']["loss"] = preference_loss + sft_loss
+            info_dict['loss_dict']["sft_loss"] = sft_loss
+            info_dict['loss_dict']["dpo_loss"] = preference_loss
+            info_dict['loss_dict']["dpo_acc"] = dpo_acc
+            info_dict['loss_dict']["chosen_reward"] = chosen_reward.mean()
+            info_dict['loss_dict']["reject_reward"] = reject_reward.mean()
+    return info_dict
+
+
+def batch_backward(model, scaler, info_dict):
+    if info_dict["train_engine"] == "deepspeed":
+        scaled_loss = model.backward(info_dict['loss_dict']['loss'])
+    else:
+        scaled_loss = info_dict['loss_dict']['loss'] / info_dict['accum_grad']
+        if scaler is not None:
+            scaler.scale(scaled_loss).backward()
+        else:
+            scaled_loss.backward()
+
+    info_dict['loss_dict']['loss'] = scaled_loss
+    return info_dict
+
+
+def update_parameter_and_lr(model, optimizer, scheduler, scaler, info_dict):
+    grad_norm = 0.0
+    if info_dict['train_engine'] == "deepspeed":
+        info_dict["is_gradient_accumulation_boundary"] = model.is_gradient_accumulation_boundary()
+        model.step()
+        grad_norm = model.get_global_grad_norm()
+    elif (info_dict['batch_idx'] + 1) % info_dict["accum_grad"] == 0:
+        # Use mixed precision training
+        if scaler is not None:
+            scaler.unscale_(optimizer)
+            grad_norm = clip_grad_norm_(model.parameters(), info_dict['grad_clip'])
+            # We don't check grad here since that if the gradient
+            # has inf/nan values, scaler.step will skip
+            # optimizer.step().
+            if torch.isfinite(grad_norm):
+                scaler.step(optimizer)
+            scaler.update()
+        else:
+            grad_norm = clip_grad_norm_(model.parameters(), info_dict['grad_clip'])
+            if torch.isfinite(grad_norm):
+                optimizer.step()
+        optimizer.zero_grad()
+        scheduler.step()
+    info_dict["lr"] = optimizer.param_groups[0]['lr']
+    info_dict["grad_norm"] = grad_norm
+    return info_dict
+
+
+def log_per_step(writer, info_dict):
+    tag = info_dict["tag"]
+    epoch = info_dict.get('epoch', 0)
+    step = info_dict["step"]
+    batch_idx = info_dict["batch_idx"]
+    loss_dict = info_dict['loss_dict']
+    rank = int(os.environ.get('RANK', 0))
+
+    # only rank 0 write to tensorboard to avoid multi-process write
+    if writer is not None:
+        if (info_dict['train_engine'] == 'deepspeed' and info_dict['is_gradient_accumulation_boundary'] is True) or \
+           (info_dict['train_engine'] == 'torch_ddp' and (info_dict['batch_idx'] + 1) % info_dict['accum_grad'] == 0):
+            for k in ['epoch', 'lr', 'grad_norm']:
+                writer.add_scalar('{}/{}'.format(tag, k), info_dict[k], step + 1)
+            for k, v in loss_dict.items():
+                writer.add_scalar('{}/{}'.format(tag, k), v, step + 1)
+
+    # TRAIN & CV, Shell log (stdout)
+    if (info_dict['batch_idx'] + 1) % info_dict['log_interval'] == 0:
+        log_str = '{} Batch {}/{} '.format(tag, epoch, batch_idx + 1)
+        for name, value in loss_dict.items():
+            log_str += '{} {:.6f} '.format(name, value)
+        if tag == "TRAIN":
+            log_str += 'lr {:.8f} grad_norm {:.6f}'.format(
+                info_dict["lr"], info_dict['grad_norm'])
+        log_str += ' rank {}'.format(rank)
+        logging.debug(log_str)
+
+
+def log_per_save(writer, info_dict):
+    tag = info_dict["tag"]
+    epoch = info_dict["epoch"]
+    step = info_dict["step"]
+    loss_dict = info_dict["loss_dict"]
+    lr = info_dict['lr']
+    rank = int(os.environ.get('RANK', 0))
+    logging.info(
+        'Epoch {} Step {} CV info lr {} {} rank {}'.format(
+            epoch, step + 1, lr, rank, ' '.join(['{}_{}'.format(k, v) for k, v in loss_dict.items()])))
+
+    if writer is not None:
+        for k in ['epoch', 'lr']:
+            writer.add_scalar('{}/{}'.format(tag, k), info_dict[k], step + 1)
+        for k, v in loss_dict.items():
+            writer.add_scalar('{}/{}'.format(tag, k), v, step + 1)

examples/libritts/cosyvoice/conf/cosyvoice_dpo.yaml

@@ -0,0 +1,226 @@
+# set random seed, so that you may reproduce your result.
+__set_seed1: !apply:random.seed [1986]
+__set_seed2: !apply:numpy.random.seed [1986]
+__set_seed3: !apply:torch.manual_seed [1986]
+__set_seed4: !apply:torch.cuda.manual_seed_all [1986]
+
+# fixed params
+sample_rate: 24000   # 16000 for llm, 24000 for cfm
+llm_input_size: 896
+llm_output_size: 896
+spk_embed_dim: 192
+qwen_pretrain_path: 'CosyVoice2-0.5B/CosyVoice-BlankEN'
+
+# model params
+# for all class/function included in this repo, we use !<name> or !<new> for intialization, so that user may find all corresponding class/function according to one single yaml.
+# for system/third_party class/function, we do not require this.
+llm: !new:cosyvoice.llm.llm_dpo.Qwen2LM
+    llm_input_size: !ref <llm_input_size>
+    llm_output_size: !ref <llm_output_size>
+    speech_token_size: 6561
+    length_normalized_loss: True
+    lsm_weight: 0
+    dpo: True
+    llm: !new:cosyvoice.llm.llm.Qwen2Encoder
+        pretrain_path: !ref <qwen_pretrain_path>
+    sampling: !name:cosyvoice.utils.common.ras_sampling
+        top_p: 0.8
+        top_k: 25
+        win_size: 10
+        tau_r: 0.1
+flow: !new:cosyvoice.flow.flow.CausalMaskedDiffWithXvec
+    input_size: 512
+    output_size: 80
+    spk_embed_dim: !ref <spk_embed_dim>
+    output_type: 'mel'
+    vocab_size: 6561
+    input_frame_rate: 25
+    only_mask_loss: True
+    token_mel_ratio: 2
+    pre_lookahead_len: 3
+    encoder: !new:cosyvoice.transformer.upsample_encoder.UpsampleConformerEncoder
+        output_size: 512
+        attention_heads: 8
+        linear_units: 2048
+        num_blocks: 6
+        dropout_rate: 0.1
+        positional_dropout_rate: 0.1
+        attention_dropout_rate: 0.1
+        normalize_before: True
+        input_layer: 'linear'
+        pos_enc_layer_type: 'rel_pos_espnet'
+        selfattention_layer_type: 'rel_selfattn'
+        input_size: 512
+        use_cnn_module: False
+        macaron_style: False
+    decoder: !new:cosyvoice.flow.flow_matching.CausalConditionalCFM
+        in_channels: 240
+        n_spks: 1
+        spk_emb_dim: 80
+        cfm_params: !new:omegaconf.DictConfig
+            content:
+                sigma_min: 1e-06
+                solver: 'euler'
+                t_scheduler: 'cosine'
+                training_cfg_rate: 0.2
+                inference_cfg_rate: 0.7
+                reg_loss_type: 'l1'
+        estimator: !new:cosyvoice.flow.decoder.ConditionalDecoder
+            in_channels: 320
+            out_channels: 80
+            causal: True
+            channels: [256]
+            dropout: 0.0
+            attention_head_dim: 64
+            n_blocks: 4
+            num_mid_blocks: 12
+            num_heads: 8
+            act_fn: 'gelu'
+
+hift: !new:cosyvoice.hifigan.generator.HiFTGenerator
+    in_channels: 80
+    base_channels: 512
+    nb_harmonics: 8
+    sampling_rate: !ref <sample_rate>
+    nsf_alpha: 0.1
+    nsf_sigma: 0.003
+    nsf_voiced_threshold: 10
+    upsample_rates: [8, 5, 3]
+    upsample_kernel_sizes: [16, 11, 7]
+    istft_params:
+        n_fft: 16
+        hop_len: 4
+    resblock_kernel_sizes: [3, 7, 11]
+    resblock_dilation_sizes: [[1, 3, 5], [1, 3, 5], [1, 3, 5]]
+    source_resblock_kernel_sizes: [7, 7, 11]
+    source_resblock_dilation_sizes: [[1, 3, 5], [1, 3, 5], [1, 3, 5]]
+    lrelu_slope: 0.1
+    audio_limit: 0.99
+    f0_predictor: !new:cosyvoice.hifigan.f0_predictor.ConvRNNF0Predictor
+        num_class: 1
+        in_channels: 80
+        cond_channels: 512
+
+# gan related module
+mel_spec_transform1: !name:matcha.utils.audio.mel_spectrogram
+    n_fft: 1024
+    num_mels: 80
+    sampling_rate: !ref <sample_rate>
+    hop_size: 256
+    win_size: 1024
+    fmin: 0
+    fmax: null
+    center: False
+hifigan: !new:cosyvoice.hifigan.hifigan.HiFiGan
+    generator: !ref <hift>
+    discriminator: !new:cosyvoice.hifigan.discriminator.MultipleDiscriminator
+        mpd: !new:matcha.hifigan.models.MultiPeriodDiscriminator
+        mrd: !new:cosyvoice.hifigan.discriminator.MultiResolutionDiscriminator
+    mel_spec_transform: [
+        !ref <mel_spec_transform1>
+    ]
+
+# processor functions
+parquet_opener: !name:cosyvoice.dataset.processor.parquet_opener
+get_tokenizer: !name:whisper.tokenizer.get_tokenizer # change to !name:cosyvoice.tokenizer.tokenizer.get_tokenizer if you want to train with CosyVoice-300M-25Hz recipe
+    multilingual: True
+    num_languages: 100
+    language: 'en'
+    task: 'transcribe'
+allowed_special: 'all'
+tokenize: !name:cosyvoice.dataset.processor.tokenize
+    get_tokenizer: !ref <get_tokenizer>
+    allowed_special: !ref <allowed_special>
+filter: !name:cosyvoice.dataset.processor.filter
+    max_length: 40960
+    min_length: 0
+    token_max_length: 200
+    token_min_length: 1
+resample: !name:cosyvoice.dataset.processor.resample
+    resample_rate: !ref <sample_rate>
+truncate: !name:cosyvoice.dataset.processor.truncate
+    truncate_length: 24576 # must be a multiplier of hop_size
+feat_extractor: !name:matcha.utils.audio.mel_spectrogram
+    n_fft: 1024
+    num_mels: 80
+    sampling_rate: !ref <sample_rate>
+    hop_size: 256
+    win_size: 1024
+    fmin: 0
+    fmax: 8000
+    center: False
+compute_fbank: !name:cosyvoice.dataset.processor.compute_fbank
+    feat_extractor: !ref <feat_extractor>
+compute_f0: !name:cosyvoice.dataset.processor.compute_f0
+    sample_rate: !ref <sample_rate>
+    hop_size: 256
+parse_embedding: !name:cosyvoice.dataset.processor.parse_embedding
+    normalize: True
+shuffle: !name:cosyvoice.dataset.processor.shuffle
+    shuffle_size: 1000
+sort: !name:cosyvoice.dataset.processor.sort
+    sort_size: 500  # sort_size should be less than shuffle_size
+batch: !name:cosyvoice.dataset.processor.batch
+    batch_type: 'dynamic'
+    max_frames_in_batch: 2000 # change to 1400 in gan train on v100 16g
+padding: !name:cosyvoice.dataset.processor.padding
+    use_spk_embedding: True # change to True during sft
+    dpo: True
+
+# dataset processor pipeline
+data_pipeline: [
+    !ref <parquet_opener>,
+    !ref <tokenize>,
+    !ref <filter>,
+    !ref <resample>,
+    !ref <compute_fbank>,
+    !ref <parse_embedding>,
+    !ref <shuffle>,
+    !ref <sort>,
+    !ref <batch>,
+    !ref <padding>,
+]
+data_pipeline_gan: [
+    !ref <parquet_opener>,
+    !ref <tokenize>,
+    !ref <filter>,
+    !ref <resample>,
+    !ref <truncate>,
+    !ref <compute_fbank>,
+    !ref <compute_f0>,
+    !ref <parse_embedding>,
+    !ref <shuffle>,
+    !ref <sort>,
+    !ref <batch>,
+    !ref <padding>,
+]
+
+# llm flow train conf
+train_conf:
+    optim: adam
+    optim_conf:
+        lr: 0.00001 # change to 1e-5 during sft
+    scheduler: warmuplr # change to constantlr during sft
+    scheduler_conf:
+        warmup_steps: 25000
+    max_epoch: 200
+    grad_clip: 5
+    accum_grad: 2
+    log_interval: 100
+    save_per_step: -1
+
+# gan train conf
+train_conf_gan:
+    optim: adam
+    optim_conf:
+        lr: 0.0002 # use small lr for gan training
+    scheduler: constantlr
+    optim_d: adam
+    optim_conf_d:
+        lr: 0.0002 # use small lr for gan training
+    scheduler_d: constantlr
+    max_epoch: 200
+    grad_clip: 5
+    accum_grad: 1 # in gan training, accum_grad must be 1
+    log_interval: 100
+    save_per_step: -1

requirements_vllm.txt

@@ -0,0 +1,40 @@
+vllm==0.7.3
+pydantic==2.10.6
+torch==2.5.1
+torchaudio==2.5.1
+
+conformer==0.3.2
+
+diffusers==0.32.2
+gdown==5.1.0
+grpcio==1.57.0
+grpcio-tools==1.57.0
+hydra-core==1.3.2
+HyperPyYAML==1.2.2
+inflect==7.3.1
+librosa==0.10.2
+
+lightning==2.5.0.post0
+matplotlib==3.7.5
+modelscope==1.15.0
+
+networkx==3.4.2
+omegaconf==2.3.0
+onnx==1.17.0
+
+onnxruntime-gpu==1.19.0; sys_platform == 'linux'
+
+#openai-whisper==20231117
+openai-whisper==20240930
+protobuf==4.25
+pyworld==0.3.4
+rich==13.7.1
+soundfile==0.12.1
+tensorboard==2.14.0
+wget==3.2
+WeTextProcessing==1.0.3
+
+# trt use
+tensorrt-cu12==10.0.1
+tensorrt-cu12-bindings==10.0.1
+tensorrt-cu12-libs==10.0.1

tools/make_parquet_list_dpo.py

@@ -0,0 +1,125 @@
+#!/usr/bin/env python3
+# Copyright (c) 2024 Alibaba Inc (authors: Xiang Lyu)
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#   http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+import argparse
+import logging
+import os
+import json
+from tqdm import tqdm
+import pandas as pd
+import multiprocessing
+import time
+import torch
+
+
+def job(utt_list, parquet_file, utt2parquet_file, spk2parquet_file):
+    start_time = time.time()
+    data_list = []
+    for utt in tqdm(utt_list):
+        data = open(utt2wav[utt], 'rb').read()
+        data_list.append(data)
+    wav_list = [utt2wav[utt] for utt in utt_list]
+    text_list = [utt2text[utt] for utt in utt_list]
+    spk_list = [utt2spk[utt] for utt in utt_list]
+    uttembedding_list = [utt2embedding[utt] for utt in utt_list]
+    spkembedding_list = [spk2embedding[utt2spk[utt]] for utt in utt_list]
+    speech_token_list = [utt2speech_token[utt] for utt in utt_list]
+    if utt2reject_speech_token:
+        reject_speech_token_list = [utt2reject_speech_token[utt] for utt in utt_list]
+
+    # 保存到parquet,utt2parquet_file,spk2parquet_file
+    df = pd.DataFrame()
+    df['utt'] = utt_list
+    df['wav'] = wav_list
+    df['audio_data'] = data_list
+    df['text'] = text_list
+    df['spk'] = spk_list
+    df['utt_embedding'] = uttembedding_list
+    df['spk_embedding'] = spkembedding_list
+    df['speech_token'] = speech_token_list
+    if utt2reject_speech_token:
+        df['reject_speech_token'] = reject_speech_token_list
+    df.to_parquet(parquet_file)
+    with open(utt2parquet_file, 'w') as f:
+        json.dump({k: parquet_file for k in utt_list}, f, ensure_ascii=False, indent=2)
+    with open(spk2parquet_file, 'w') as f:
+        json.dump({k: parquet_file for k in list(set(spk_list))}, f, ensure_ascii=False, indent=2)
+    logging.info('spend time {}'.format(time.time() - start_time))
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument('--num_utts_per_parquet',
+                        type=int,
+                        default=1000,
+                        help='num utts per parquet')
+    parser.add_argument('--num_processes',
+                        type=int,
+                        default=1,
+                        help='num processes for make parquets')
+    parser.add_argument('--src_dir',
+                        type=str)
+    parser.add_argument('--des_dir',
+                        type=str)
+    parser.add_argument('--dpo',
+                        action='store_true',
+                        default=False,
+                        help='Use Direct Preference Optimization')
+    args = parser.parse_args()
+
+    utt2wav, utt2text, utt2spk = {}, {}, {}
+    with open('{}/wav.scp'.format(args.src_dir)) as f:
+        for l in f:
+            l = l.replace('\n', '').split()
+            utt2wav[l[0]] = l[1]
+    with open('{}/text'.format(args.src_dir)) as f:
+        for l in f:
+            l = l.replace('\n', '').split()
+            utt2text[l[0]] = ' '.join(l[1:])
+    with open('{}/utt2spk'.format(args.src_dir)) as f:
+        for l in f:
+            l = l.replace('\n', '').split()
+            utt2spk[l[0]] = l[1]
+    utt2embedding = torch.load('{}/utt2embedding.pt'.format(args.src_dir))
+    spk2embedding = torch.load('{}/spk2embedding.pt'.format(args.src_dir))
+    utt2speech_token = torch.load('{}/utt2speech_token.pt'.format(args.src_dir))
+    if args.dpo:
+        utt2reject_speech_token = torch.load('{}/utt2reject_speech_token.pt'.format(args.src_dir))
+    else:
+        utt2reject_speech_token = None
+    utts = list(utt2wav.keys())
+
+    # Using process pool to speedup
+    pool = multiprocessing.Pool(processes=args.num_processes)
+    parquet_list, utt2parquet_list, spk2parquet_list = [], [], []
+    for i, j in enumerate(range(0, len(utts), args.num_utts_per_parquet)):
+        parquet_file = os.path.join(args.des_dir, 'parquet_{:09d}.tar'.format(i))
+        utt2parquet_file = os.path.join(args.des_dir, 'utt2parquet_{:09d}.json'.format(i))
+        spk2parquet_file = os.path.join(args.des_dir, 'spk2parquet_{:09d}.json'.format(i))
+        parquet_list.append(parquet_file)
+        utt2parquet_list.append(utt2parquet_file)
+        spk2parquet_list.append(spk2parquet_file)
+        pool.apply_async(job, (utts[j: j + args.num_utts_per_parquet], parquet_file, utt2parquet_file, spk2parquet_file))
+    pool.close()
+    pool.join()
+
+    with open('{}/data.list'.format(args.des_dir), 'w', encoding='utf8') as f1, \
+            open('{}/utt2data.list'.format(args.des_dir), 'w', encoding='utf8') as f2, \
+            open('{}/spk2data.list'.format(args.des_dir), 'w', encoding='utf8') as f3:
+        for name in parquet_list:
+            f1.write(name + '\n')
+        for name in utt2parquet_list:
+            f2.write(name + '\n')
+        for name in spk2parquet_list:
+            f3.write(name + '\n')