deepgeek/CosyVoice
2
0
Fork 0
mirror of https://github.com/FunAudioLLM/CosyVoice.git synced 2026-02-05 18:09:24 +08:00
Code Issues Packages Projects Releases Wiki Activity

add WeTextProcessing

Browse Source
This commit is contained in:
lyuxiang.lx
2024-07-09 23:37:54 +08:00
parent 5e97398d38
commit 7981796523
4 changed files with 12 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
README.md
View File

@@ -22,6 +22,8 @@ git submodule update --init --recursive
``` sh ``` sh
conda create -n cosyvoice python=3.8 conda create -n cosyvoice python=3.8
conda activate cosyvoice conda activate cosyvoice
# pynini is required by WeTextProcessing, use conda to install it as it can be executed on all platform.
conda install -y -c conda-forge pynini==2.1.5
pip install -r requirements.txt -i https://mirrors.aliyun.com/pypi/simple/ --trusted-host=mirrors.aliyun.com pip install -r requirements.txt -i https://mirrors.aliyun.com/pypi/simple/ --trusted-host=mirrors.aliyun.com
# If you encounter sox compatibility issues # If you encounter sox compatibility issues
@@ -55,9 +57,9 @@ git clone https://www.modelscope.cn/iic/CosyVoice-300M-Instruct.git pretrained_m
git clone https://www.modelscope.cn/iic/CosyVoice-ttsfrd.git pretrained_models/CosyVoice-ttsfrd git clone https://www.modelscope.cn/iic/CosyVoice-ttsfrd.git pretrained_models/CosyVoice-ttsfrd
``` ```
Optionaly, you can unzip `ttsfrd` resouce and install `ttsfrd` package. Optionaly, you can unzip `ttsfrd` resouce and install `ttsfrd` package for better text normalization performance.
Notice that this step is not necessary. If you do not install `ttsfrd` package, you need to normalize input text manually. Notice that this step is not necessary. If you do not install `ttsfrd` package, we will use WeTextProcessing by default.
``` sh ``` sh
cd pretrained_models/CosyVoice-ttsfrd/ cd pretrained_models/CosyVoice-ttsfrd/

14
cosyvoice/cli/frontend.py
View File

@@ -21,13 +21,13 @@ import torchaudio.compliance.kaldi as kaldi
import torchaudio import torchaudio
import os import os
import inflect import inflect
from tn.chinese.normalizer import Normalizer as ZhNormalizer
from tn.english.normalizer import Normalizer as EnNormalizer
try: try:
import ttsfrd import ttsfrd
use_ttsfrd = True use_ttsfrd = True
except: except ImportError:
print("failed to import ttsfrd, please normalize input text manually") print("failed to import ttsfrd, use WeTextProcessing instead")
from tn.chinese.normalizer import Normalizer as ZhNormalizer
from tn.english.normalizer import Normalizer as EnNormalizer
use_ttsfrd = False use_ttsfrd = False
from cosyvoice.utils.frontend_utils import contains_chinese, replace_blank, replace_corner_mark, remove_bracket, spell_out_number, split_paragraph from cosyvoice.utils.frontend_utils import contains_chinese, replace_blank, replace_corner_mark, remove_bracket, spell_out_number, split_paragraph
@@ -64,8 +64,8 @@ class CosyVoiceFrontEnd:
self.frd.enable_pinyin_mix(True) self.frd.enable_pinyin_mix(True)
self.frd.set_breakmodel_index(1) self.frd.set_breakmodel_index(1)
else: else:
self.zh_tn_model = ZhNormalizer(remove_erhua=False,full_to_half=False) self.zh_tn_model = ZhNormalizer(remove_erhua=False, full_to_half=False)
self.en_tn_model = EnNormalizer() self.en_tn_model = EnNormalizer()
def _extract_text_token(self, text): def _extract_text_token(self, text):
text_token = self.tokenizer.encode(text, allowed_special=self.allowed_special) text_token = self.tokenizer.encode(text, allowed_special=self.allowed_special)
@@ -103,7 +103,7 @@ class CosyVoiceFrontEnd:
if self.use_ttsfrd: if self.use_ttsfrd:
text = self.frd.get_frd_extra_info(text, 'input') text = self.frd.get_frd_extra_info(text, 'input')
else: else:
text = self.zh_tn_model.normalize(text) text = self.zh_tn_model.normalize(text)
text = text.replace("\n", "") text = text.replace("\n", "")
text = replace_blank(text) text = replace_blank(text)
text = replace_corner_mark(text) text = replace_corner_mark(text)

1
cosyvoice/dataset/processor.py
View File

@@ -22,7 +22,6 @@ from torch.nn.utils.rnn import pad_sequence
import torch.nn.functional as F import torch.nn.functional as F
torchaudio.set_audio_backend('soundfile') torchaudio.set_audio_backend('soundfile')
torchaudio.utils.sox_utils.set_buffer_size(16500)
AUDIO_FORMAT_SETS = set(['flac', 'mp3', 'm4a', 'ogg', 'opus', 'wav', 'wma']) AUDIO_FORMAT_SETS = set(['flac', 'mp3', 'm4a', 'ogg', 'opus', 'wav', 'wma'])

2
requirements.txt
View File

@@ -26,4 +26,4 @@ tensorboard==2.14.0
torch==2.0.1 torch==2.0.1
torchaudio==2.0.2 torchaudio==2.0.2
wget==3.2 wget==3.2
WeTextProcessing WeTextProcessing==1.0.3