turicreate.sound_classifier.get_deep_features(audio_data, verbose=True)

Calculates the deep features used by the Sound Classifier.

Internally the Sound Classifier calculates deep features for both model creation and predictions. If the same data will be used multiple times, calculating the deep features just once will result in a significant speed up.

audio_data : SArray

Audio data is represented as dicts with key ‘data’ and ‘sample_rate’, see turicreate.load_audio(…).


>>> my_audio_data['deep_features'] = get_deep_features(my_audio_data['audio'])
>>> train, test = my_audio_data.random_split(.8)
>>> model = tc.sound_classifier.create(train, 'label', 'deep_features')
>>> predictions = model.predict(test)