README.md

metadata

dataset_info:
  - config_name: test_a
    features:
      - name: audio
        dtype:
          audio:
            sampling_rate: 22050
            decode: false
      - name: librivox_source
        list:
          - name: item
            struct:
              - name: reader_id
                dtype: int32
              - name: section_id
                dtype: int32
              - name: book_id
                dtype: int32
              - name: offset_seconds
                dtype: float32
              - name: speed_factor
                dtype: float32
      - name: audioset_source_tar_url
        dtype: string
      - name: audioset_source_filename
        dtype: string
configs:
  - config_name: test_a
    data_files:
      - split: test_a
        path: data/test_a/train-*