deepgeek/CosyVoice
2
0
Fork 0
mirror of https://github.com/FunAudioLLM/CosyVoice.git synced 2026-02-04 17:39:25 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix lint

Browse Source
This commit is contained in:
lyuxiang.lx
2024-09-05 16:15:34 +08:00
parent eeebc45313
commit 90433f5373
35 changed files with 189 additions and 122 deletions
Download Patch File Download Diff File Expand all files Collapse all files

63
webui.py
View File

@@ -13,9 +13,6 @@
# limitations under the License.
import os
import sys
ROOT_DIR = os.path.dirname(os.path.abspath(__file__))
sys.path.append('{}/third_party/Matcha-TTS'.format(ROOT_DIR))
import argparse
import gradio as gr
import numpy as np
@@ -23,9 +20,19 @@ import torch
import torchaudio
import random
import librosa
ROOT_DIR = os.path.dirname(os.path.abspath(__file__))
sys.path.append('{}/third_party/Matcha-TTS'.format(ROOT_DIR))
from cosyvoice.cli.cosyvoice import CosyVoice
from cosyvoice.utils.file_utils import load_wav, speed_change, logging
from cosyvoice.utils.file_utils import load_wav, logging
inference_mode_list = ['预训练音色', '3s极速复刻', '跨语种复刻', '自然语言控制']
instruct_dict = {'预训练音色': '1. 选择预训练音色\n2. 点击生成音频按钮',
'3s极速复刻': '1. 选择prompt音频文件，或录入prompt音频，注意不超过30s，若同时提供，优先选择prompt音频文件\n2. 输入prompt文本\n3. 点击生成音频按钮',
'跨语种复刻': '1. 选择prompt音频文件，或录入prompt音频，注意不超过30s，若同时提供，优先选择prompt音频文件\n2. 点击生成音频按钮',
'自然语言控制': '1. 选择预训练音色\n2. 输入instruct文本\n3. 点击生成音频按钮'}
stream_mode_list = [('否', False), ('是', True)]
max_val = 0.8
def generate_seed():
seed = random.randint(1, 100000000)
@@ -34,13 +41,14 @@ def generate_seed():
"value": seed
}
def set_all_random_seed(seed):
random.seed(seed)
np.random.seed(seed)
torch.manual_seed(seed)
torch.cuda.manual_seed_all(seed)
max_val = 0.8
def postprocess(speech, top_db=60, hop_length=220, win_length=440):
speech, _ = librosa.effects.trim(
speech, top_db=top_db,
@@ -52,16 +60,13 @@ def postprocess(speech, top_db=60, hop_length=220, win_length=440):
speech = torch.concat([speech, torch.zeros(1, int(target_sr * 0.2))], dim=1)
return speech
inference_mode_list = ['预训练音色', '3s极速复刻', '跨语种复刻', '自然语言控制']
instruct_dict = {'预训练音色': '1. 选择预训练音色\n2. 点击生成音频按钮',
'3s极速复刻': '1. 选择prompt音频文件，或录入prompt音频，注意不超过30s，若同时提供，优先选择prompt音频文件\n2. 输入prompt文本\n3. 点击生成音频按钮',
'跨语种复刻': '1. 选择prompt音频文件，或录入prompt音频，注意不超过30s，若同时提供，优先选择prompt音频文件\n2. 点击生成音频按钮',
'自然语言控制': '1. 选择预训练音色\n2. 输入instruct文本\n3. 点击生成音频按钮'}
stream_mode_list = [('否', False), ('是', True)]
def change_instruction(mode_checkbox_group):
return instruct_dict[mode_checkbox_group]
def generate_audio(tts_text, mode_checkbox_group, sft_dropdown, prompt_text, prompt_wav_upload, prompt_wav_record, instruct_text, seed, stream, speed_factor):
def generate_audio(tts_text, mode_checkbox_group, sft_dropdown, prompt_text, prompt_wav_upload, prompt_wav_record, instruct_text,
seed, stream, speed_factor):
if prompt_wav_upload is not None:
prompt_wav = prompt_wav_upload
elif prompt_wav_record is not None:
@@ -72,31 +77,31 @@ def generate_audio(tts_text, mode_checkbox_group, sft_dropdown, prompt_text, pro
if mode_checkbox_group in ['自然语言控制']:
if cosyvoice.frontend.instruct is False:
gr.Warning('您正在使用自然语言控制模式, {}模型不支持此模式, 请使用iic/CosyVoice-300M-Instruct模型'.format(args.model_dir))
return (target_sr, default_data)
yield (target_sr, default_data)
if instruct_text == '':
gr.Warning('您正在使用自然语言控制模式, 请输入instruct文本')
return (target_sr, default_data)
yield (target_sr, default_data)
if prompt_wav is not None or prompt_text != '':
gr.Info('您正在使用自然语言控制模式, prompt音频/prompt文本会被忽略')
# if cross_lingual mode, please make sure that model is iic/CosyVoice-300M and tts_text prompt_text are different language
if mode_checkbox_group in ['跨语种复刻']:
if cosyvoice.frontend.instruct is True:
gr.Warning('您正在使用跨语种复刻模式, {}模型不支持此模式, 请使用iic/CosyVoice-300M模型'.format(args.model_dir))
return (target_sr, default_data)
yield (target_sr, default_data)
if instruct_text != '':
gr.Info('您正在使用跨语种复刻模式, instruct文本会被忽略')
if prompt_wav is None:
gr.Warning('您正在使用跨语种复刻模式, 请提供prompt音频')
return (target_sr, default_data)
yield (target_sr, default_data)
gr.Info('您正在使用跨语种复刻模式, 请确保合成文本和prompt文本为不同语言')
# if in zero_shot cross_lingual, please make sure that prompt_text and prompt_wav meets requirements
if mode_checkbox_group in ['3s极速复刻', '跨语种复刻']:
if prompt_wav is None:
gr.Warning('prompt音频为空，您是否忘记输入prompt音频？')
return (target_sr, default_data)
yield (target_sr, default_data)
if torchaudio.info(prompt_wav).sample_rate < prompt_sr:
gr.Warning('prompt音频采样率{}低于{}'.format(torchaudio.info(prompt_wav).sample_rate, prompt_sr))
return (target_sr, default_data)
yield (target_sr, default_data)
# sft mode only use sft_dropdown
if mode_checkbox_group in ['预训练音色']:
if instruct_text != '' or prompt_wav is not None or prompt_text != '':
@@ -105,7 +110,7 @@ def generate_audio(tts_text, mode_checkbox_group, sft_dropdown, prompt_text, pro
if mode_checkbox_group in ['3s极速复刻']:
if prompt_text == '':
gr.Warning('prompt文本为空，您是否忘记输入prompt文本？')
return (target_sr, default_data)
yield (target_sr, default_data)
if instruct_text != '':
gr.Info('您正在使用3s极速复刻模式，预训练音色/instruct文本会被忽略！')
@@ -113,28 +118,32 @@ def generate_audio(tts_text, mode_checkbox_group, sft_dropdown, prompt_text, pro
logging.info('get sft inference request')
set_all_random_seed(seed)
for i in cosyvoice.inference_sft(tts_text, sft_dropdown, stream=stream):
yield (target_sr, i['tts_speech'].numpy().flatten())
yield (target_sr, i['tts_speech'].numpy().flatten())
elif mode_checkbox_group == '3s极速复刻':
logging.info('get zero_shot inference request')
prompt_speech_16k = postprocess(load_wav(prompt_wav, prompt_sr))
set_all_random_seed(seed)
for i in cosyvoice.inference_zero_shot(tts_text, prompt_text, prompt_speech_16k, stream=stream):
yield (target_sr, i['tts_speech'].numpy().flatten())
yield (target_sr, i['tts_speech'].numpy().flatten())
elif mode_checkbox_group == '跨语种复刻':
logging.info('get cross_lingual inference request')
prompt_speech_16k = postprocess(load_wav(prompt_wav, prompt_sr))
set_all_random_seed(seed)
for i in cosyvoice.inference_cross_lingual(tts_text, prompt_speech_16k, stream=stream):
yield (target_sr, i['tts_speech'].numpy().flatten())
yield (target_sr, i['tts_speech'].numpy().flatten())
else:
logging.info('get instruct inference request')
set_all_random_seed(seed)
for i in cosyvoice.inference_instruct(tts_text, sft_dropdown, instruct_text, stream=stream):
yield (target_sr, i['tts_speech'].numpy().flatten())
yield (target_sr, i['tts_speech'].numpy().flatten())
def main():
with gr.Blocks() as demo:
gr.Markdown("### 代码库 [CosyVoice](https://github.com/FunAudioLLM/CosyVoice) 预训练模型 [CosyVoice-300M](https://www.modelscope.cn/models/iic/CosyVoice-300M) [CosyVoice-300M-Instruct](https://www.modelscope.cn/models/iic/CosyVoice-300M-Instruct) [CosyVoice-300M-SFT](https://www.modelscope.cn/models/iic/CosyVoice-300M-SFT)")
gr.Markdown("### 代码库 [CosyVoice](https://github.com/FunAudioLLM/CosyVoice) \
预训练模型 [CosyVoice-300M](https://www.modelscope.cn/models/iic/CosyVoice-300M) \
[CosyVoice-300M-Instruct](https://www.modelscope.cn/models/iic/CosyVoice-300M-Instruct) \
[CosyVoice-300M-SFT](https://www.modelscope.cn/models/iic/CosyVoice-300M-SFT)")
gr.Markdown("#### 请输入需要合成的文本，选择推理模式，并按照提示步骤进行操作")
tts_text = gr.Textbox(label="输入合成文本", lines=1, value="我是通义实验室语音团队全新推出的生成式语音大模型，提供舒适自然的语音合成能力。")
@@ -160,12 +169,14 @@ def main():
seed_button.click(generate_seed, inputs=[], outputs=seed)
generate_button.click(generate_audio,
inputs=[tts_text, mode_checkbox_group, sft_dropdown, prompt_text, prompt_wav_upload, prompt_wav_record, instruct_text, seed, stream, speed_factor],
inputs=[tts_text, mode_checkbox_group, sft_dropdown, prompt_text, prompt_wav_upload, prompt_wav_record, instruct_text,
seed, stream, speed_factor],
outputs=[audio_output])
mode_checkbox_group.change(fn=change_instruction, inputs=[mode_checkbox_group], outputs=[instruction_text])
demo.queue(max_size=4, default_concurrency_limit=2)
demo.launch(server_name='0.0.0.0', server_port=args.port)
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--port',