blackteay/GPT-SoVITS
1
0
Fork 0
mirror of https://github.com/RVC-Boss/GPT-SoVITS.git synced 2025-04-05 04:22:46 +08:00
Code Issues Packages Projects Releases Wiki Activity

Optimize short sentence (#1430)

Browse Source
This commit is contained in:
KamioRinn 2024-08-10 12:28:53 +08:00 committed by GitHub
parent 893b45246b
commit 2310bcde53
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
3 changed files with 10 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
GPT_SoVITS/inference_webui.py
View File

@ -299,7 +299,7 @@ def get_first(text):
return text return text
from text import chinese from text import chinese
def get_phones_and_bert(text,language,version): def get_phones_and_bert(text,language,version,final=False):
if language in {"en", "all_zh", "all_ja", "all_ko", "all_yue"}: if language in {"en", "all_zh", "all_ja", "all_ko", "all_yue"}:
language = language.replace("all_","") language = language.replace("all_","")
if language == "en": if language == "en":
@ -366,6 +366,9 @@ def get_phones_and_bert(text,language,version):
phones = sum(phones_list, []) phones = sum(phones_list, [])
norm_text = ''.join(norm_text_list) norm_text = ''.join(norm_text_list)
if not final and len(phones) < 6:
return get_phones_and_bert("." + text,language,version,final=True)
return phones,bert.to(dtype),norm_text return phones,bert.to(dtype),norm_text
@ -408,7 +411,7 @@ def get_tts_wav(ref_wav_path, prompt_text, prompt_language, text, text_language,
if (prompt_text[-1] not in splits): prompt_text += "。" if prompt_language != "en" else "." if (prompt_text[-1] not in splits): prompt_text += "。" if prompt_language != "en" else "."
print(i18n("实际输入的参考文本:"), prompt_text) print(i18n("实际输入的参考文本:"), prompt_text)
text = text.strip("\n") text = text.strip("\n")
if (text[0] not in splits and len(get_first(text)) < 4): text = "。" + text if text_language != "en" else "." + text # if (text[0] not in splits and len(get_first(text)) < 4): text = "。" + text if text_language != "en" else "." + text
print(i18n("实际输入的目标文本:"), text) print(i18n("实际输入的目标文本:"), text)
zero_wav = np.zeros( zero_wav = np.zeros(

2
GPT_SoVITS/text/cleaner.py
View File

@ -45,7 +45,7 @@ def clean_text(text, language, version=None):
elif language == "en": elif language == "en":
phones = language_module.g2p(norm_text) phones = language_module.g2p(norm_text)
if len(phones) < 4: if len(phones) < 4:
phones = [','] * (4 - len(phones)) + phones phones = [','] + phones
word2ph = None word2ph = None
else: else:
phones = language_module.g2p(norm_text) phones = language_module.g2p(norm_text)

5
api.py
View File

@ -275,7 +275,7 @@ def get_bert_inf(phones, word2ph, norm_text, language):
return bert return bert
from text import chinese from text import chinese
def get_phones_and_bert(text,language,version): def get_phones_and_bert(text,language,version,final=False):
if language in {"en", "all_zh", "all_ja", "all_ko", "all_yue"}: if language in {"en", "all_zh", "all_ja", "all_ko", "all_yue"}:
language = language.replace("all_","") language = language.replace("all_","")
if language == "en": if language == "en":
@ -340,6 +340,9 @@ def get_phones_and_bert(text,language,version):
phones = sum(phones_list, []) phones = sum(phones_list, [])
norm_text = ''.join(norm_text_list) norm_text = ''.join(norm_text_list)
if not final and len(phones) < 6:
return get_phones_and_bert("." + text,language,version,final=True)
return phones,bert.to(torch.float16 if is_half == True else torch.float32),norm_text return phones,bert.to(torch.float16 if is_half == True else torch.float32),norm_text