captioned-audio-1k / README.md
cappuch's picture
Update README.md
9ebee0d verified
metadata
task_categories:
  - audio-classification
  - text-to-audio
language:
  - en
size_categories:
  - 1K<n<10K
configs:
  - config_name: default
    data_files:
      - split: train
        path: dataset.csv

This is a dataset intended for audio captioning/generation. It is formatted using spectograms. It is based off the Clotho dataset, with many other sizes coming soon. We used this dataset to train the embedding model, along with some other songs.