dataset_info:
  features:
  - name: id
    dtype: string
  - name: audio
    dtype: 'struct<bytes: binary, path: string>'
  - name: text
    dtype: string
  splits:
  - name: dev
    num_bytes: 889581958
  - name: test
    num_bytes: 945497798
  - name: train
    num_bytes: 4819120779
  download_size: 6654200535
  dataset_size: 6654200535