100k_a / README.md
sin2piusc's picture
Update README.md
b6662bf verified
|
Raw History Blame Contribute Delete
1.15 kB
metadata
dataset_info:
  features:
    - name: audio
      dtype:
        audio:
          sampling_rate: 16000
    - name: sentence
      dtype: string
  splits:
    - name: train
      num_bytes: 14041063989
      num_examples: 100000
  download_size: 12813562837
  dataset_size: 14041063989
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: artistic-2.0
language:
  - ja
task_categories:
  - automatic-speech-recognition
  - translation
  - text-generation
  - feature-extraction
size_categories:
  - 10K<n<100K
tags:
  - whisper
  - audio
  - text
pretty_name: >-
  common_voice_17_0, google_fleurs, genshin, jsut, jsss, css10,
  japanese_anime_speech_v2 and v1

datasets : common_voice_17_0, google_fleurs, genshin, jsut, jsss, css10, japanese_anime_speech_v2 and v1.

Samples with any of these characters: [?\-\一\.\「\」\・-;:\“%\‘\”\ .~\♪\ー\…\~\!\ー?!!{}\?\。#$%&*\a-z\A-Z\0-9] have been removed from this dataset.

100k samples - Japanese (1of4) - No English - No punctuation - Sampling rate = 16,000 - No noise - sfw/nsfw mix - Reviewed by a human :) Still working on the onomatopoeias..