README.md

metadata

task_categories:
  - automatic-speech-recognition
configs:
  - config_name: clean
    data_files:
      - split: train.100
        path: clean/train.100/*.tar
      - split: train.360
        path: clean/train.360/*.tar
      - split: validation
        path: clean/validation/*.tar
      - split: test
        path: clean/test/*.tar
  - config_name: other
    data_files:
      - split: train.500
        path: other/train.500/*.tar
      - split: validation
        path: other/validation/*.tar
      - split: test
        path: other/test/*.tar
language:
  - en

LibriSpeech ASR

We performed some processing while sharding this dataset:

• Audios were pre-shuffled across the shards.