blackteay/GPT-SoVITS
1
0
Fork 0
mirror of https://github.com/RVC-Boss/GPT-SoVITS.git synced 2025-06-04 07:09:17 +08:00
Code Issues Packages Projects Releases Wiki Activity

Fix bug in #1660 and #1667 (#1670)

Browse Source 
* fix #1660 #1667

* add japanese cmudict-katakana dict
This commit is contained in:
AkitoLiu 2024-10-02 10:47:07 -04:00 committed by GitHub
parent a95b2b85f7
commit eee607b71d
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 134228 additions and 25 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

134200
GPT_SoVITS/text/ja_userdic/userdict.csv
View File

File diff suppressed because it is too large Load Diff

53
GPT_SoVITS/text/japanese.py
View File

@ -1,29 +1,34 @@
# modified from https://github.com/CjangCjengh/vits/blob/main/text/japanese.py
import re
import pyopenjtalk
import os
import hashlib
current_file_path = os.path.dirname(__file__)
def get_hash(fp: str) -> str:
hash_md5 = hashlib.md5()
with open(fp, "rb") as f:
for chunk in iter(lambda: f.read(4096), b""):
hash_md5.update(chunk)
return hash_md5.hexdigest()
try:
import pyopenjtalk
current_file_path = os.path.dirname(__file__)
def get_hash(fp: str) -> str:
hash_md5 = hashlib.md5()
with open(fp, "rb") as f:
for chunk in iter(lambda: f.read(4096), b""):
hash_md5.update(chunk)
return hash_md5.hexdigest()
USERDIC_CSV_PATH = os.path.join(current_file_path, "ja_userdic", "userdict.csv")
USERDIC_BIN_PATH = os.path.join(current_file_path, "ja_userdic", "user.dict")
USERDIC_HASH_PATH = os.path.join(current_file_path, "ja_userdic", "userdict.md5")
# 如果没有用户词典，就生成一个；如果有，就检查md5，如果不一样，就重新生成
if os.path.exists(USERDIC_CSV_PATH):
if not os.path.exists(USERDIC_BIN_PATH) or get_hash(USERDIC_CSV_PATH) != open(USERDIC_HASH_PATH, "r",encoding='utf-8').read():
pyopenjtalk.mecab_dict_index(USERDIC_CSV_PATH, USERDIC_BIN_PATH)
with open(USERDIC_HASH_PATH, "w", encoding='utf-8') as f:
f.write(get_hash(USERDIC_CSV_PATH))
USERDIC_CSV_PATH = os.path.join(current_file_path, "ja_userdic", "userdict.csv")
USERDIC_BIN_PATH = os.path.join(current_file_path, "ja_userdic", "user.dict")
USERDIC_HASH_PATH = os.path.join(current_file_path, "ja_userdic", "userdict.md5")
# 如果没有用户词典，就生成一个；如果有，就检查md5，如果不一样，就重新生成
if os.path.exists(USERDIC_CSV_PATH):
if not os.path.exists(USERDIC_BIN_PATH) or get_hash(USERDIC_CSV_PATH) != open(USERDIC_HASH_PATH, "r",encoding='utf-8').read():
pyopenjtalk.mecab_dict_index(USERDIC_CSV_PATH, USERDIC_BIN_PATH)
with open(USERDIC_HASH_PATH, "w", encoding='utf-8') as f:
f.write(get_hash(USERDIC_CSV_PATH))
if os.path.exists(USERDIC_BIN_PATH):
pyopenjtalk.update_global_jtalk_with_user_dict(USERDIC_BIN_PATH)
if os.path.exists(USERDIC_BIN_PATH):
pyopenjtalk.update_global_jtalk_with_user_dict(USERDIC_BIN_PATH)
except Exception as e:
# print(e)
import pyopenjtalk
# failed to load user dictionary, ignore.
pass
from text.symbols import punctuation
@ -80,10 +85,6 @@ def post_replace_ph(ph):
if ph in rep_map.keys():
ph = rep_map[ph]
# if ph in symbols:
# return ph
# if ph not in symbols:
# ph = "UNK"
return ph
@ -103,6 +104,8 @@ def symbols_to_japanese(text):
def preprocess_jap(text, with_prosody=False):
"""Reference https://r9y9.github.io/ttslearn/latest/notebooks/ch10_Recipe-Tacotron.html"""
text = symbols_to_japanese(text)
# English words to lower case, should have no influence on japanese words.
text = text.lower()
sentences = re.split(_japanese_marks, text)
marks = re.findall(_japanese_marks, text)
text = []
@ -219,5 +222,5 @@ def g2p(norm_text, with_prosody=True):
if __name__ == "__main__":
phones = g2p("こんにちは, hello, AKITOです,よろしくお願いしますね！")
phones = g2p("Hello.こんにちは！今日もNiCe天気ですね！tokyotowerに行きましょう！")
print(phones)