blackteay/GPT-SoVITS
1
0
Fork 0
mirror of https://github.com/RVC-Boss/GPT-SoVITS.git synced 2025-04-06 03:57:44 +08:00
Code Issues Packages Projects Releases Wiki Activity

Fix some detail problems

Browse Source
This commit is contained in:
KamioRinn 2024-08-19 15:25:05 +08:00
parent 8b0fbe6d18
commit 49427c75bc
1 changed files with 5 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
api.py
View File

@ -325,14 +325,14 @@ def get_phones_and_bert(text,language,version,final=False):
if language == "zh": if language == "zh":
if re.search(r'[A-Za-z]', formattext): if re.search(r'[A-Za-z]', formattext):
formattext = re.sub(r'[a-z]', lambda x: x.group(0).upper(), formattext) formattext = re.sub(r'[a-z]', lambda x: x.group(0).upper(), formattext)
formattext = chinese.text_normalize(formattext) formattext = chinese.mix_text_normalize(formattext)
return get_phones_and_bert(formattext,"zh",version) return get_phones_and_bert(formattext,"zh",version)
else: else:
phones, word2ph, norm_text = clean_text_inf(formattext, language, version) phones, word2ph, norm_text = clean_text_inf(formattext, language, version)
bert = get_bert_feature(norm_text, word2ph).to(device) bert = get_bert_feature(norm_text, word2ph).to(device)
elif language == "yue" and re.search(r'[A-Za-z]', formattext): elif language == "yue" and re.search(r'[A-Za-z]', formattext):
formattext = re.sub(r'[a-z]', lambda x: x.group(0).upper(), formattext) formattext = re.sub(r'[a-z]', lambda x: x.group(0).upper(), formattext)
formattext = chinese.text_normalize(formattext) formattext = chinese.mix_text_normalize(formattext)
return get_phones_and_bert(formattext,"yue",version) return get_phones_and_bert(formattext,"yue",version)
else: else:
phones, word2ph, norm_text = clean_text_inf(formattext, language, version) phones, word2ph, norm_text = clean_text_inf(formattext, language, version)
@ -413,6 +413,9 @@ class DictToAttrRecursive(dict):
def get_spepc(hps, filename): def get_spepc(hps, filename):
audio,_ = librosa.load(filename, int(hps.data.sampling_rate)) audio,_ = librosa.load(filename, int(hps.data.sampling_rate))
audio = torch.FloatTensor(audio) audio = torch.FloatTensor(audio)
maxx=audio.abs().max()
if(maxx>1):
audio/=min(2,maxx)
audio_norm = audio audio_norm = audio
audio_norm = audio_norm.unsqueeze(0) audio_norm = audio_norm.unsqueeze(0)
spec = spectrogram_torch(audio_norm, hps.data.filter_length, hps.data.sampling_rate, hps.data.hop_length, spec = spectrogram_torch(audio_norm, hps.data.filter_length, hps.data.sampling_rate, hps.data.hop_length,