metadata

license: apache-2.0
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
dataset_info:
  features:
    - name: audio
      dtype:
        audio:
          sampling_rate: 16000
    - name: text
      dtype: string
  splits:
    - name: train
      num_bytes: 11169873357.152
      num_examples: 102263
  download_size: 10186919269
  dataset_size: 11169873357.152
task_categories:
  - automatic-speech-recognition
language:
  - ko
size_categories:
  - 100K<n<1M
pretty_name: zeroth-STT-Ko

Zeroth-STT-Ko Dataset

Description

This dataset combines the following publicly available Korean language datasets:

Junhoee/STT_Korean_Dataset_80000 and Zeroth-Korean Dataset (from Project: Zeroth, by GoodAtlas and Gridspace)

This provides over 102K rows of data (sentences) in total.

Citation

Zeroth-Korean Dataset, created by [Lucas Jo(@Atlas Guide Inc.) and Wonkyum Lee(@Gridspace Inc.)], 2023. Available at https://github.com/goodatlas/zeroth under CC-BY-4.0 license.

Junhoee/STT_Korean_Dataset_80000 Dataset, created by [Junhoee], 2024. Available at https://huggingface.co/datasets/Junhoee/STT_Korean_Dataset_80000