Hi,

here are some recent research works of colleagues of mine who used unsupervised deep learning for the automatic clustering of speech sounds:

Conference Paper Unsupervised Speech Unit Discovery Using K-means and Neural Networks

Conference Paper CNN-Based Phone Segmentation Experiments in a Less-Represent...

Hope this helps

Lionel

i would suggest the following :

http://deeplearning.stanford.edu/tutorial/

both for Supervised and Unsupervised

Read this book

http://www.deeplearningitalia.com/wp-content/uploads/2017/12/Dropbox_Chollet.pdf