diff --git a/.local/bin/video2srt b/.local/bin/video2srt index f5a1e84..6db5c1f 100755 --- a/.local/bin/video2srt +++ b/.local/bin/video2srt @@ -19,9 +19,6 @@ DEFAULT_MODEL = Path( DEFAULT_BIN = Path( os.environ.get("WHISPER_BIN", WHISPER_DIR / "build/bin/whisper-cli") ) -DEFAULT_VAD = Path( - os.environ.get("VAD_MODEL", WHISPER_DIR / "models/ggml-silero-v6.2.0.bin") -) MIN_WORDS_TO_WRAP = 2 SRT_HEADER_LINES = 2 @@ -125,20 +122,6 @@ def process(video: Path, args: argparse.Namespace) -> None: cmd.append("--translate") if args.max_len > 0: cmd += ["--max-len", str(args.max_len), "--split-on-word"] - if args.vad: - if args.vad_model.exists(): - cmd += ["--vad", "--vad-model", str(args.vad_model)] - if args.vad_max_speech > 0: - cmd += ["-vmsd", str(args.vad_max_speech)] - else: - print( - f"warn: VAD model not found at {args.vad_model}, running without VAD", - file=sys.stderr, - ) - print( - f" download with: sh {WHISPER_DIR}/models/download-vad-model.sh silero-v6.2.0", - file=sys.stderr, - ) subprocess.run(cmd, check=True) @@ -182,18 +165,6 @@ def main() -> int: p.add_argument( "-f", "--force", action="store_true", help="Overwrite existing .srt" ) - p.add_argument( - "--no-vad", - dest="vad", - action="store_false", - help="Disable Silero VAD pre-filtering (VAD reduces hallucination loops)", - ) - p.add_argument( - "--vad-max-speech", - type=float, - default=15.0, - help="Max seconds of speech per VAD chunk; shorter values give tighter timestamps (default: 15)", - ) p.add_argument( "--max-len", type=int, @@ -209,9 +180,6 @@ def main() -> int: p.add_argument( "--bin", type=Path, default=DEFAULT_BIN, help=argparse.SUPPRESS ) - p.add_argument( - "--vad-model", type=Path, default=DEFAULT_VAD, help=argparse.SUPPRESS - ) args = p.parse_args()