mrq/ai-voice-cloning
31
74
Fork 43
Code Issues 223 Pull Requests 5 Packages Projects Releases Wiki Activity

I don't know how I did not commit the 'sample from the voices to construct the input prompt for vall-e' change but this helps

Browse Source
This commit is contained in:
mrq 2023-08-25 04:26:48 +00:00
parent a657623cbc
commit b4dc103931
1 changed files with 11 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
src/utils.py
View File

@ -679,14 +679,23 @@ def generate_valle(**kwargs):
def fetch_voice( voice ): def fetch_voice( voice ):
if voice in voice_cache: if voice in voice_cache:
return voice_cache[voice] return voice_cache[voice]
"""
voice_dir = f'./training/{voice}/audio/' voice_dir = f'./training/{voice}/audio/'
if not os.path.isdir(voice_dir) or len(os.listdir(voice_dir)) == 0: if not os.path.isdir(voice_dir) or len(os.listdir(voice_dir)) == 0:
voice_dir = f'./voices/{voice}/' voice_dir = f'./voices/{voice}/'
files = [ f'{voice_dir}/{d}' for d in os.listdir(voice_dir) if d[-4:] == ".wav" ] files = [ f'{voice_dir}/{d}' for d in os.listdir(voice_dir) if d[-4:] == ".wav" ]
"""
if os.path.isdir(f'./training/{voice}/audio/'):
files = get_voice(name="audio", dir=f"./training/{voice}/", load_latents=False)
else:
files = get_voice(name=voice, load_latents=False)
# return files # return files
voice_cache[voice] = random.choice(files) voice_cache[voice] = random.sample(files, k=min(3, len(files)))
return voice_cache[voice] return voice_cache[voice]
def get_settings( override=None ): def get_settings( override=None ):
@ -707,7 +716,7 @@ def generate_valle(**kwargs):
continue continue
settings[k] = override[k] settings[k] = override[k]
settings['references'] = [ fetch_voice(voice=selected_voice) for _ in range(3) ] settings['references'] = fetch_voice(voice=selected_voice) # [ fetch_voice(voice=selected_voice) for _ in range(3) ]
return settings return settings
if not parameters['delimiter']: if not parameters['delimiter']: