Open JamesRick opened 4 years ago
We should investigate using an AutoEncoder trained on the data to draw affinity functions from. We could encode different portions of the spectrogram to produce multiple prototypes from one sample.
We should investigate using an AutoEncoder trained on the data to draw affinity functions from. We could encode different portions of the spectrogram to produce multiple prototypes from one sample.