prepare_reference.py

import argparse
from pathlib import Path

import librosa
import soundfile as sf


def main():
    parser = argparse.ArgumentParser(
        description="Normalize/trim a reference file for better XTTS cloning quality."
    )
    parser.add_argument("--input", required=True, help="Input audio path.")
    parser.add_argument(
        "--output",
        default="data/reference_clean.wav",
        help="Output cleaned WAV path.",
    )
    parser.add_argument(
        "--target_sr",
        type=int,
        default=24000,
        help="Output sample rate.",
    )
    parser.add_argument(
        "--top_db",
        type=int,
        default=30,
        help="Silence trimming aggressiveness.",
    )
    args = parser.parse_args()

    y, sr = librosa.load(args.input, sr=args.target_sr, mono=True)
    y_trimmed, _ = librosa.effects.trim(y, top_db=args.top_db)

    output_path = Path(args.output)
    output_path.parent.mkdir(parents=True, exist_ok=True)
    sf.write(str(output_path), y_trimmed, args.target_sr)

    print(f"Saved cleaned reference: {output_path}")


if __name__ == "__main__":
    main()