This Space showcases two audio classification models fine-tuned on the speech_commands dataset:

  • wav2vec2-conformer: 97.2% accuracy, added in transformers 4.20.0.
  • audio-spectrogram-transformer: 98.12% accuracy, added in transformers 4.25.1.

They can spot one of the following keywords: 'Yes', 'No', 'Up', 'Down', 'Left', 'Right', 'On', 'Off', 'Stop', 'Go', 'Zero', 'One', 'Two', 'Three', 'Four', 'Five', 'Six', 'Seven', 'Eight', 'Nine', 'Bed', 'Bird', 'Cat', 'Dog', 'Happy', 'House', 'Marvin', 'Sheila', 'Tree', 'Wow', 'Backward', 'Forward', 'Follow', 'Learn', 'Visual'.
Model selection
Examples
Speech input Model selection