HarkonCollider/tortoise-tts
1
0
Fork 0
forked from mrq/tortoise-tts
Code Pull Requests Activity

arg to skip voice latents for grabbing voice lists (for preparing datasets)

Browse Source
This commit is contained in:
mrq 2023-02-17 04:50:02 +00:00
parent b255a77a05
commit 00cb19b6cf
1 changed files with 5 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
tortoise/utils/audio.py
View File

@ -77,7 +77,7 @@ def dynamic_range_decompression(x, C=1):
return torch.exp(x) / C return torch.exp(x) / C
def get_voices(extra_voice_dirs=[]): def get_voices(extra_voice_dirs=[], load_latents=True):
dirs = [get_voice_dir()] + extra_voice_dirs dirs = [get_voice_dir()] + extra_voice_dirs
voices = {} voices = {}
for d in dirs: for d in dirs:
@ -85,7 +85,9 @@ def get_voices(extra_voice_dirs=[]):
for sub in subs: for sub in subs:
subj = os.path.join(d, sub) subj = os.path.join(d, sub)
if os.path.isdir(subj): if os.path.isdir(subj):
voices[sub] = list(glob(f'{subj}/*.wav')) + list(glob(f'{subj}/*.mp3')) + list(glob(f'{subj}/*.pth')) voices[sub] = list(glob(f'{subj}/*.wav')) + list(glob(f'{subj}/*.mp3'))
if load_latents:
voices[sub] = voices[sub] + list(glob(f'{subj}/*.pth'))
return voices return voices
@ -93,7 +95,7 @@ def load_voice(voice, extra_voice_dirs=[], load_latents=True, sample_rate=22050,
if voice == 'random': if voice == 'random':
return None, None return None, None
voices = get_voices(extra_voice_dirs) voices = get_voices(extra_voice_dirs=extra_voice_dirs, load_latents=load_latents)
paths = voices[voice] paths = voices[voice]
mtime = 0 mtime = 0