ben_mkiv/ai-voice-cloning
1
0
Fork 0
forked from mrq/ai-voice-cloning
Code Pull Requests Activity

indentation fix

Browse Source
This commit is contained in:
mrq 2023-04-28 15:56:57 +00:00
parent 99387920e1
commit eddb8aaa9a
1 changed files with 27 additions and 27 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

52
src/utils.py
View File

@ -106,16 +106,16 @@ if BARK_ENABLED:
self.output_sample_rate = args.output_sample_rate self.output_sample_rate = args.output_sample_rate
preload_models( preload_models(
text_use_gpu=True, text_use_gpu=True,
coarse_use_gpu=True, coarse_use_gpu=True,
fine_use_gpu=True, fine_use_gpu=True,
codec_use_gpu=True, codec_use_gpu=True,
text_use_small=small, text_use_small=small,
coarse_use_small=small, coarse_use_small=small,
fine_use_small=small, fine_use_small=small,
force_reload=False force_reload=False
) )
def create_voice( self, voice, device='cuda' ): def create_voice( self, voice, device='cuda' ):
@ -150,7 +150,7 @@ if BARK_ENABLED:
# Extract discrete codes from EnCodec # Extract discrete codes from EnCodec
with torch.no_grad(): with torch.no_grad():
encoded_frames = model.encode(wav) encoded_frames = model.encode(wav)
codes = torch.cat([encoded[0] for encoded in encoded_frames], dim=-1).squeeze().cpu().numpy() # [n_q, T] codes = torch.cat([encoded[0] for encoded in encoded_frames], dim=-1).squeeze().cpu().numpy() # [n_q, T]
# get seconds of audio # get seconds of audio
@ -2285,30 +2285,30 @@ def phonemizer( text, language="en-us" ):
from phonemizer.backend import BACKENDS from phonemizer.backend import BACKENDS
def _get_backend( language="en-us", backend="espeak" ): def _get_backend( language="en-us", backend="espeak" ):
key = f'{language}_{backend}' key = f'{language}_{backend}'
if key in cached_backends: if key in cached_backends:
return cached_backends[key] return cached_backends[key]
if backend == 'espeak': if backend == 'espeak':
phonemizer = BACKENDS[backend]( language, preserve_punctuation=True, with_stress=True) phonemizer = BACKENDS[backend]( language, preserve_punctuation=True, with_stress=True)
elif backend == 'espeak-mbrola': elif backend == 'espeak-mbrola':
phonemizer = BACKENDS[backend]( language ) phonemizer = BACKENDS[backend]( language )
else: else:
phonemizer = BACKENDS[backend]( language, preserve_punctuation=True ) phonemizer = BACKENDS[backend]( language, preserve_punctuation=True )
cached_backends[key] = phonemizer cached_backends[key] = phonemizer
return phonemizer return phonemizer
if language == "en": if language == "en":
language = "en-us" language = "en-us"
backend = _get_backend(language=language, backend=args.phonemizer_backend) backend = _get_backend(language=language, backend=args.phonemizer_backend)
if backend is not None: if backend is not None:
tokens = backend.phonemize( text, strip=True ) tokens = backend.phonemize( text, strip=True )
else: else:
tokens = phonemize( text, language=language, strip=True, preserve_punctuation=True, with_stress=True ) tokens = phonemize( text, language=language, strip=True, preserve_punctuation=True, with_stress=True )
return tokens[0] if len(tokens) == 0 else tokens return tokens[0] if len(tokens) == 0 else tokens
tokenized = " ".join( tokens ) tokenized = " ".join( tokens )
def should_phonemize(): def should_phonemize():
should = args.tokenizer_json is not None and args.tokenizer_json[-8:] == "ipa.json" should = args.tokenizer_json is not None and args.tokenizer_json[-8:] == "ipa.json"