This dataset contains packed codec-token .npz files generated by
audioagi tokens pack-training-dataset. It does not contain raw waveform audio.
token_manifest.jsonl
pack_manifest.jsonl
packs/*.npz
summary.json
run_record.json
Codec: encodec_24khz
Storage format: npz-packed-rvq:frames
Records: 116500
Packs: 15
Frames: 76902664
Source token manifest:… See the full description on the dataset page:
https://huggingface.co/datasets/SolidSnake123/audio-agi-libritts-train-clean-360-encodec-tokens-packed-v0.
