killkan / README.md
ctaguchi's picture
Update README.md
b925c33 verified
|
raw
history blame
907 Bytes
metadata
dataset_info:
  features:
    - name: audio
      dtype: audio
    - name: sentence
      dtype: string
  splits:
    - name: train
      num_bytes: 184050207.062
      num_examples: 3002
    - name: valid
      num_bytes: 22788399
      num_examples: 375
    - name: test
      num_bytes: 22656433
      num_examples: 376
  download_size: 225437827
  dataset_size: 229495039.062
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: valid
        path: data/valid-*
      - split: test
        path: data/test-*
license: cc-by-4.0
task_categories:
  - automatic-speech-recognition
language:
  - qu
  - qvi
  - qug
  - qxl
  - qvj
  - qxr
  - qud
size_categories:
  - 1K<n<10K

Killkan: Speech Recognition dataset for Kichwa

Killkan (Kichwa uyachkata payllatak killkak anta) is the first automatic speech recognition (ASR) dataset for the Kichwa language. See also our paper (https://arxiv.org/abs/2404.15501).