AI audio classification

This page does not yet provide a complete sound-classification training and deployment guide. Two advanced routes are currently available:

• Use a classification model with audio input and follow Port an Unsupported Model to add preprocessing and inference code.
• Convert audio into spectrogram images, then train an image classifier. A spectrogram shows how the strength of each frequency changes over time.

Both routes require your own training and data-processing code. If you only need to recognize a fixed command, start with Keyword Recognition.