AudioTokenBench / README.md
martinma's picture
Update README.
f2f8364
metadata
dataset_info:
  features:
    - name: audio
      dtype: audio
    - name: label
      dtype:
        class_label:
          names:
            '0': audiophile
            '1': music
            '2': sound_event
            '3': speech
  splits:
    - name: test
      num_bytes: 1965782403.25
      num_examples: 3150
  download_size: 1609823419
  dataset_size: 1965782403.25
configs:
  - config_name: default
    data_files:
      - split: test
        path: data/test-*

AudioTokenBench

This is the evaluation dataset for HiggsTokenizer. It contains 3150 24khz audio samples across 4 subsets:

  • Speech: 1,000 clips of 10 seconds audio, randomly sampled from DAPS.
  • Music: 1,000 clips of 10 seconds audio, randomly sampled from MUSDB.
  • Sound Event: 1,000 clips of 10 seconds audio, randomly sampled from AudioSet.
  • Audiophile: Contains 150 clips of 30 seconds audio, curated from eleven high-fidelity test discs. The clips feature both music and sound events, selected for high-quality audio evaluation.

For detailed evaluation metrics, please refer to our blog and github.