|
|
| """Transcribe a local audio file with NepTrans v4 and faster-whisper."""
|
| from __future__ import annotations
|
|
|
| import argparse
|
| import json
|
| from pathlib import Path
|
|
|
| from faster_whisper import WhisperModel
|
|
|
|
|
| def main():
|
| parser = argparse.ArgumentParser(description=__doc__)
|
| parser.add_argument("audio", type=Path)
|
| parser.add_argument("--model", default="DontHash/neptrans-v4-ct2")
|
| parser.add_argument("--revision", default="v4.0.0")
|
| parser.add_argument("--device", choices=["cpu", "cuda"], default="cpu")
|
| parser.add_argument("--compute-type", default=None)
|
| parser.add_argument("--language", choices=["ne", "en"], default=None)
|
| parser.add_argument("--beam-size", type=int, default=1)
|
| parser.add_argument("--local-files-only", action="store_true")
|
| parser.add_argument("--output", type=Path)
|
| parser.add_argument("--json-output", type=Path)
|
| args = parser.parse_args()
|
| if not args.audio.is_file():
|
| parser.error(f"Audio file not found: {args.audio}")
|
| if args.beam_size < 1:
|
| parser.error("--beam-size must be positive")
|
| if args.output and args.output.resolve() == args.audio.resolve():
|
| parser.error("Output must not overwrite the input audio")
|
| if args.json_output and args.json_output.resolve() == args.audio.resolve():
|
| parser.error("JSON output must not overwrite the input audio")
|
| if args.output and args.json_output and args.output.resolve() == args.json_output.resolve():
|
| parser.error("Text and JSON outputs must be different files")
|
| compute_type = args.compute_type or ("int8_float16" if args.device == "cuda" else "int8")
|
| model = WhisperModel(
|
| args.model, revision=args.revision, device=args.device,
|
| compute_type=compute_type, local_files_only=args.local_files_only,
|
| )
|
| segments, info = model.transcribe(
|
| str(args.audio), language=args.language, task="transcribe",
|
| beam_size=args.beam_size, condition_on_previous_text=False,
|
| )
|
| rows = [{"start": s.start, "end": s.end, "text": s.text.strip()} for s in segments]
|
| text = "\n".join(row["text"] for row in rows) + "\n"
|
| if args.output:
|
| args.output.write_text(text, encoding="utf-8")
|
| else:
|
| print(text, end="")
|
| if args.json_output:
|
| args.json_output.write_text(json.dumps({
|
| "language": info.language,
|
| "language_probability": info.language_probability,
|
| "duration": info.duration,
|
| "segments": rows,
|
| }, ensure_ascii=False, indent=2) + "\n", encoding="utf-8")
|
|
|
|
|
| if __name__ == "__main__":
|
| main()
|
|
|
