dearwolf/LocalAI
1
0
Fork 0
mirror of https://github.com/mudler/LocalAI.git synced 2025-05-20 10:35:01 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 21

fix(vall-e-x): Fix voice cloning (#1696)

Browse source
This commit is contained in:
Ettore Di Giacinto 2024-02-11 11:20:00 +01:00 • committed by GitHub
parent 58cdf97361
commit fd68bf7084
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 7 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

4
backend/python/vall-e-x/ttsvalle.py
View file

@ -55,6 +55,7 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
print("Preparing models, please wait", file=sys.stderr)
# download and load all models
preload_models()
self.clonedVoice = False
# Assume directory from request.ModelFile.
# Only if request.LoraAdapter it's not an absolute path
if request.AudioPath and request.ModelFile != "" and not os.path.isabs(request.AudioPath):
@ -65,6 +66,7 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
if request.AudioPath != "":
print("Generating model", file=sys.stderr)
make_prompt(name=model_name, audio_prompt_path=request.AudioPath)
self.clonedVoice = True
### Use given transcript
##make_prompt(name=model_name, audio_prompt_path="paimon_prompt.wav",
## transcript="Just, what was that? Paimon thought we were gonna get eaten.")
@ -91,6 +93,8 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
try:
audio_array = None
if model != "":
if self.clonedVoice:
model = os.path.basename(request.model)
audio_array = generate_audio(request.text, prompt=model)
else:
audio_array = generate_audio(request.text)