My new survey is out. It reviews automatic techniques used for automatic speech recognition tasks from Large Corpora to Limited Data. It includes an opening to few-shot techniques (to use few data to train the model) and first promising results over speech/audio signals. Check it out.

Cheers, Vincent