dataset_info:
  features:
  - name: id
    dtype: string
  - name: audio
    dtype: 'struct<bytes: binary, path: string>'
  - name: dialect
    dtype: string
  splits:
  - name: dev
    num_bytes: 2188937354
  - name: test
    num_bytes: 3030622009
  - name: train
    num_bytes: 258047188724
  download_size: 263266748087
  dataset_size: 263266748087