README.md

metadata

dataset_info:
  features:
    - name: audio
      dtype:
        audio:
          sampling_rate: 16000
    - name: sentence
      dtype: string
  splits:
    - name: train
      num_bytes: 5640823349.174486
      num_examples: 50004
  download_size: 12080511714
  dataset_size: 5640823349.174486
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: afl-3.0
task_categories:
  - translation
  - text-generation
  - automatic-speech-recognition
language:
  - ja
pretty_name: >-
  common voice, google fleurs, JSUTv1.1, JAS_v2
  (joujiboi/japanese-anime-speech-v2) processed for whisper.
size_categories:
  - 10K<n<100K

Dataset has -not- been striped of special characters or normalized.

Data set has been suffled and flattened. For faster version check out jgca_v2_50k_2.