The new AudioIdentify function is using a trained neural net to identify about 600 sound categories from an audio signal. Audio identification can be used to create audio captioning, monitor surveillance data and more.
Perform identification on a drum recording.
Restrict the identification to a specific category in the ontology.
You can also get multiple identifications, along with their associated probabilities.
Create a signal containing different sounds.
Make identifications over 1-second intervals using AudioBlockMap.
Plot the results.