Is That Right?

In the supervised case, a neural network can kind of figure out exactly what should it look at in the speech stream to identify the speech sound. But in the unsupervised case, how do you do this? Because you don't know. So you need to think about unsupervised models that can do something like clusstering. And i think yor am ou slipping through things very carefully. I think that's good. Ok, so let me try to to recap that to see what i was able to capture.