README.md

metadata

configs:
  - config_name: default
    data_files:
      - split: train
        path: train/*.parquet

AVSpeech Video + Audio

This dataset stores AVSpeech media as separate columns:

• video: video-only stream (no audio), stream-copied without re-encoding.
• audio: audio-only stream, stream-copied without re-encoding.