| 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. | |