CLI Reference

The package installs one command:

audio-prep

It has two independent subcommands:

• convert: convert source audio files to WAV or FLAC, validate them, and optionally write a manifest.
• chunk: split source audio files into speech-only chunks and optionally write a chunk manifest.

audio-prep convert

audio-prep convert \
    --input-dir data/raw_mp3 \
    --output-dir data/wav16k \
    --format wav \
    --sample-rate 16000 \
    --channels 1 \
    --workers 4 \
    --min-duration-sec 0.5 \
    --manifest data/manifest.jsonl

Flag	Default	Description
--input-dir	required	Directory scanned recursively for source audio files.
--output-dir	required	Directory where converted output is written.
--extensions	common FFmpeg audio/video extensions	Comma-separated source extensions, or all to pass every regular file to FFmpeg.
--format	wav	Output format, wav or flac.
--sample-rate	16000	Target sample rate in Hz.
--channels	1	Target channel count.
--workers	4	Number of parallel conversion workers.
--min-duration-sec	0.5	Minimum accepted output duration.
--overwrite	off	Rebuild even when an existing output is valid.
--normalize-loudness	off	Apply single-pass EBU R128 loudness normalization.
--manifest	none	Path to write conversion manifest JSONL.

Exit codes:

Code	Meaning
0	Every discovered file converted and validated.
1	At least one file failed conversion or validation.
2	Argument parsing failed.

audio-prep chunk

audio-prep chunk \
    --input-dir data/raw_mp3 \
    --output-dir data/chunks \
    --format wav \
    --sample-rate 16000 \
    --min-duration-sec 5 \
    --max-duration-sec 20 \
    --workers 4 \
    --manifest data/chunk_manifest.jsonl

Flag	Default	Description
--input-dir	required	Directory scanned recursively for source audio files.
--output-dir	<input-dir>/chunks	Directory where chunks are written.
--extensions	common FFmpeg audio/video extensions	Comma-separated source extensions, or all to pass every regular file to FFmpeg.
--format	wav	Chunk format, wav or flac.
--sample-rate	16000	Resample before VAD and chunk writing.
--min-duration-sec	5.0	Drop chunks shorter than this duration.
--max-duration-sec	20.0	Split longer speech into windows no longer than this.
--workers	4	Number of parallel chunking workers.
--overwrite	off	Rebuild even when an existing chunk is valid.
--allow-energy-fallback	off	Use a lower-quality energy detector if Silero cannot load.
--manifest	none	Path to write chunk manifest JSONL.

Exit codes:

Code	Meaning
0	Every discovered file produced one or more chunks.
1	At least one file failed or produced no chunks.
2	Argument parsing failed.