auracaster/piper
3
0
Fork 0
mirror of https://github.com/pstrueb/piper.git synced 2026-04-17 22:05:30 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add extra options to infer script

Browse Source
This commit is contained in:
Michael Hansen
2023-10-30 11:34:13 -05:00
parent c4b7fd3b5a
commit 1d3e8b2901
1 changed files with 6 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
src/python/piper_train/infer.py
View File

@@ -24,6 +24,11 @@ def main():
) )
parser.add_argument("--output-dir", required=True, help="Path to write WAV files") parser.add_argument("--output-dir", required=True, help="Path to write WAV files")
parser.add_argument("--sample-rate", type=int, default=22050) parser.add_argument("--sample-rate", type=int, default=22050)
#
parser.add_argument("--noise-scale", type=float, default=0.667)
parser.add_argument("--length-scale", type=float, default=1.0)
parser.add_argument("--noise-w", type=float, default=0.8)
#
args = parser.parse_args() args = parser.parse_args()
args.output_dir = Path(args.output_dir) args.output_dir = Path(args.output_dir)
@@ -49,7 +54,7 @@ def main():
text = torch.LongTensor(phoneme_ids).unsqueeze(0) text = torch.LongTensor(phoneme_ids).unsqueeze(0)
text_lengths = torch.LongTensor([len(phoneme_ids)]) text_lengths = torch.LongTensor([len(phoneme_ids)])
scales = [0.667, 1.0, 0.8] scales = [args.noise_scale, args.length_scale, args.noise_w]
sid = torch.LongTensor([speaker_id]) if speaker_id is not None else None sid = torch.LongTensor([speaker_id]) if speaker_id is not None else None
start_time = time.perf_counter() start_time = time.perf_counter()