What CNN algorithms can I use for Denoise Speech?

Hello everyone,

I applied the example in the link for denoise speech. But can I set up the CNN structure with other methods? Like AlexNet for example.

Answer 1

[Walter Roberson]

Data is data. You can reshape it or chop it up however you want. Deep Learning algorithms do not care whether the data was an image, or was a video of a rock opera, or was tracking where your cat went while you were not home.

You should be less immediately concerned with what could be done, and more concerned with whether it makes sense to use a particular CNN.

[Walter Roberson]

https://en.m.wikipedia.org/wiki/AlexNet

Alexnet expects rgb images — 227 x 227 x 3 arrays of data, and targets 1000 classes.

If you were to create spectrographs over short sounds you could possibly classify down to the phoneme.