metadata

dataset_info:
  features:
    - name: audio
      dtype:
        audio:
          sampling_rate: 16000
    - name: text
      dtype: string
  splits:
    - name: train
      num_bytes: 11169954042.658998
      num_examples: 102263
  download_size: 10204609448
  dataset_size: 11169954042.658998
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: apache-2.0
task_categories:
  - automatic-speech-recognition
language:
  - ko
size_categories:
  - 100K<n<1M

Zeroth-STT-Korean Dataset

Description

This is a shuffled version of the Zeroth-STT-Ko dataset.

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