README.md

metadata

dataset_info:
  - config_name: dev_clean
    features:
      - name: transcript
        dtype: string
      - name: phonemes
        list: int32
      - name: durations
        list: float32
      - name: feats
        dtype:
          array2_d:
            shape:
              - null
              - 39
            dtype: float32
    splits:
      - name: dev_clean
        num_bytes: 314037457
        num_examples: 2703
    download_size: 304617180
    dataset_size: 314037457
  - config_name: dev_other
    features:
      - name: transcript
        dtype: string
      - name: phonemes
        list: int32
      - name: durations
        list: float32
      - name: feats
        dtype:
          array2_d:
            shape:
              - null
              - 39
            dtype: float32
    splits:
      - name: dev_other
        num_bytes: 298491930
        num_examples: 2864
    download_size: 289508236
    dataset_size: 298491930
  - config_name: test_clean
    features:
      - name: transcript
        dtype: string
      - name: phonemes
        list: int32
      - name: durations
        list: float32
      - name: feats
        dtype:
          array2_d:
            shape:
              - null
              - 39
            dtype: float32
    splits:
      - name: test_clean
        num_bytes: 314958920
        num_examples: 2620
    download_size: 305315915
    dataset_size: 314958920
  - config_name: test_other
    features:
      - name: transcript
        dtype: string
      - name: phonemes
        list: int32
      - name: durations
        list: float32
      - name: feats
        dtype:
          array2_d:
            shape:
              - null
              - 39
            dtype: float32
    splits:
      - name: test_other
        num_bytes: 311132600
        num_examples: 2938
    download_size: 301644425
    dataset_size: 311132600
  - config_name: train_clean_100
    features:
      - name: transcript
        dtype: string
      - name: phonemes
        list: int32
      - name: durations
        list: float32
      - name: feats
        dtype:
          array2_d:
            shape:
              - null
              - 39
            dtype: float32
    splits:
      - name: train_clean_100
        num_bytes: 5859947190
        num_examples: 28538
    download_size: 5683391884
    dataset_size: 5859947190
  - config_name: train_clean_360
    features:
      - name: transcript
        dtype: string
      - name: phonemes
        list: int32
      - name: durations
        list: float32
      - name: feats
        dtype:
          array2_d:
            shape:
              - null
              - 39
            dtype: float32
    splits:
      - name: train_clean_360
        num_bytes: 21178355013
        num_examples: 104008
    download_size: 20536800326
    dataset_size: 21178355013
  - config_name: train_other_500
    features:
      - name: transcript
        dtype: string
      - name: phonemes
        list: int32
      - name: durations
        list: float32
      - name: feats
        dtype:
          array2_d:
            shape:
              - null
              - 39
            dtype: float32
    splits:
      - name: train_other_500
        num_bytes: 28926977167
        num_examples: 148645
    download_size: 28044280349
    dataset_size: 28926977167
configs:
  - config_name: dev_clean
    data_files:
      - split: dev_clean
        path: dev_clean/dev_clean-*
  - config_name: dev_other
    data_files:
      - split: dev_other
        path: dev_other/dev_other-*
  - config_name: test_clean
    data_files:
      - split: test_clean
        path: test_clean/test_clean-*
  - config_name: test_other
    data_files:
      - split: test_other
        path: test_other/test_other-*
  - config_name: train_clean_100
    data_files:
      - split: train_clean_100
        path: train_clean_100/train_clean_100-*
  - config_name: train_clean_360
    data_files:
      - split: train_clean_360
        path: train_clean_360/train_clean_360-*
  - config_name: train_other_500
    data_files:
      - split: train_other_500
        path: train_other_500/train_other_500-*
task_categories:
  - automatic-speech-recognition
language:
  - en
pretty_name: Preprocessed MFCCs of English Librispeech with force-aligned phonemes