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

Update frontend.py

Browse Source 
Fix the bug in handling anomalies for synthetic text ending with Chinese and English commas.
This commit is contained in:
zhuyunfeng
2024-07-14 22:33:56 +08:00
committed by GitHub
parent d1df60590a
commit 6ae6ba3f77
1 changed files with 2 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
cosyvoice/cli/frontend.py
View File

@@ -20,6 +20,7 @@ from typing import Callable
import torchaudio.compliance.kaldi as kaldi import torchaudio.compliance.kaldi as kaldi
import torchaudio import torchaudio
import os import os
import re
import inflect import inflect
try: try:
import ttsfrd import ttsfrd
@@ -110,6 +111,7 @@ class CosyVoiceFrontEnd:
text = text.replace(".", "、") text = text.replace(".", "、")
text = text.replace(" - ", "，") text = text.replace(" - ", "，")
text = remove_bracket(text) text = remove_bracket(text)
text = re.sub(r'[，,]+$', '。', text)
texts = [i for i in split_paragraph(text, partial(self.tokenizer.encode, allowed_special=self.allowed_special), "zh", token_max_n=80, texts = [i for i in split_paragraph(text, partial(self.tokenizer.encode, allowed_special=self.allowed_special), "zh", token_max_n=80,
token_min_n=60, merge_len=20, token_min_n=60, merge_len=20,
comma_split=False)] comma_split=False)]