metadata

license: cc-by-nc-4.0
dataset_info:
  features:
    - name: audio_filename
      dtype:
        audio:
          sampling_rate: 16000
    - name: labels
      sequence: string
    - name: scores
      sequence: float64
  splits:
    - name: chunk_0.5s
      num_bytes: 2335914119.828
      num_examples: 408828
    - name: 500k_part1_chunk_0.5s
      num_bytes: 10729355414.456
      num_examples: 3103548
    - name: chunk_0.25s
      num_bytes: 2570845459.725
      num_examples: 692209
    - name: 500k_part2_chunk_0.5s
      num_bytes: 10547345488.462
      num_examples: 3073862
  download_size: 30054318161
  dataset_size: 26183460482.470993
configs:
  - config_name: default
    data_files:
      - split: chunk_0.5s
        path: data/chunk_0.5s-*
      - split: chunk_0.25s
        path: data/chunk_0.25s-*
      - split: 500k_part1_chunk_0.5s
        path: data/500k_part1_chunk_0.5s-*
      - split: 500k_part2_chunk_0.5s
        path: data/500k_part2_chunk_0.5s-*
task_categories:
  - audio-classification
pretty_name: AudioSet-Chunk

AudioSet Chunk

Chunk AudioSet every 0.25 seconds and 0.5 seconds and predict using MIT/ast-finetuned-audioset-10-10-0.4593, after that only filter and labels predicted overlapped with the gold labels.