README.md

metadata

dataset_info:
  features:
    - name: audio
      dtype: audio
    - name: speaker
      dtype: string
    - name: transcription
      dtype: string
    - name: translation
      dtype: string
    - name: Class
      dtype: string
    - name: Class_index
      dtype: float64
  splits:
    - name: train
      num_bytes: 515094059
      num_examples: 750
    - name: test
      num_bytes: 171698025
      num_examples: 250
  download_size: 590736573
  dataset_size: 686792084
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: test
        path: data/test-*

concatentation of "thomasavare/waste-classification-audio-deepl" and "thomasavare/waste-classification-audio-deepl2", so its easier to use.

there are 3 duplicates, idk why but it's too long to remake the datasets (maybe later) and not necessary.