|
|
|
@ -1269,11 +1269,11 @@ def phonemizer( text, language="eng" ):
|
|
|
|
|
return ["_" if p in ignored else p for p in phones]
|
|
|
|
|
"""
|
|
|
|
|
|
|
|
|
|
def prepare_dataset( voice, use_segments=False, text_length=0, audio_length=0, normalize=True, progress=gr.Progress() ):
|
|
|
|
|
def prepare_dataset( voice, use_segments=False, text_length=0, audio_length=0, progress=gr.Progress() ):
|
|
|
|
|
indir = f'./training/{voice}/'
|
|
|
|
|
infile = f'{indir}/whisper.json'
|
|
|
|
|
messages = []
|
|
|
|
|
|
|
|
|
|
normalize = True
|
|
|
|
|
phonemize = args.tokenizer_json is not None and args.tokenizer_json[-8:] == "ipa.json"
|
|
|
|
|
if args.tts_backend == "vall-e":
|
|
|
|
|
phonemize = True
|
|
|
|
@ -1301,7 +1301,7 @@ def prepare_dataset( voice, use_segments=False, text_length=0, audio_length=0, n
|
|
|
|
|
|
|
|
|
|
normalizer = None
|
|
|
|
|
if normalize:
|
|
|
|
|
normalizer = EnglishTextNormalizer() if language.lower()[:2] == "en" else BasicTextNormalizer()
|
|
|
|
|
normalizer = EnglishTextNormalizer() if language and language == "english" else BasicTextNormalizer()
|
|
|
|
|
|
|
|
|
|
# check if unsegmented text exceeds 200 characters
|
|
|
|
|
if not use_segment:
|
|
|
|
@ -2225,7 +2225,8 @@ def unload_tts():
|
|
|
|
|
do_gc()
|
|
|
|
|
|
|
|
|
|
def reload_tts():
|
|
|
|
|
load_tts( restart=True )
|
|
|
|
|
unload_tts()
|
|
|
|
|
load_tts()
|
|
|
|
|
|
|
|
|
|
def get_current_voice():
|
|
|
|
|
global current_voice
|
|
|
|
|