audio-katana

A tool to slice your audio files into chunks using the Voice Activity Detection technique

Install the project

git clone https://github.com/egorsmkv/audio-katana
cd audio-katana

pip install librosa torch torchaudio tqdm

Basic command:

mkdir demo-chunks

python audio_katana.py --source ./demo --format wav --destination ./demo-chunks

Create own folder for chunks:

python audio_katana.py --random_subfolder --source ./demo --format wav --destination ./demo-chunks

Find audio files recursively:

python audio_katana.py --recursive --random_subfolder --source ./demo --format wav --destination ./demo-chunks

Chunk audios and skip chunks out of limits: --min_chunk_len, --max_chunk_len:

python audio_katana.py --random_subfolder --source ./demo --format wav --destination ./demo-chunks --min_chunk_len 1.5 --max_chunk_len 5.5

Chunk audios and skip chunks out of WADA SNR values: --min_wada_snr, --max_wada_snr:

python audio_katana.py --random_subfolder --source ./demo --format wav --destination ./demo-chunks --min_wada_snr 1.5 --max_wada_snr 5.5