kbd_speech / README.md
anzorq's picture
Update README.md
3791001
metadata
language:
  - kbd
task_categories:
  - automatic-speech-recognition
  - text-to-speech
dataset_info:
  features:
    - name: audio
      dtype: audio
    - name: transcription
      dtype: string
    - name: gender
      dtype: string
    - name: country
      dtype: string
    - name: speaker_id
      dtype: int64
  splits:
    - name: train
      num_bytes: 193658385.11
      num_examples: 20555
  download_size: 518811329
  dataset_size: 193658385.11
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

Dataset Card for "kbd_speech"

More Information needed