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.