Add a device argument to the tts pytorch script.

This commit is contained in:
laurent 2025-07-07 08:31:03 +02:00
parent 91a4d120cb
commit 8927c0d9eb

View File

@ -44,12 +44,18 @@ def main():
help="The voice to use, relative to the voice repo root. "
f"See {DEFAULT_DSM_TTS_VOICE_REPO}",
)
parser.add_argument(
"--device",
type=str,
default="cuda",
help="Device on which to run, defaults to 'cuda'.",
)
args = parser.parse_args()
print("Loading model...")
checkpoint_info = CheckpointInfo.from_hf_repo(args.hf_repo)
tts_model = TTSModel.from_checkpoint_info(
checkpoint_info, n_q=32, temp=0.6, device=torch.device("cuda")
checkpoint_info, n_q=32, temp=0.6, device=args.device
)
if args.inp == "-":