This is an important question with many possible answers.

A study of CNN for speech recognition (with implemenation) is given in

https://www.researchgate.net/publication/264387213_Convolutional_Neural_Networks_for_Speech_Recognition

and

https://www.researchgate.net/publication/307889463_Towards_End-to-End_Speech_Recognition_with_Deep_Convolutional_Neural_Networks

https://www.researchgate.net/publication/269295053_Data_Augmentation_for_Deep_Neural_Network_Acoustic_Modeling

https://www.researchgate.net/publication/264859599_Exploring_Convolutional_Neural_Network_Structures_and_Optimization_Techniques_for_Speech_Recognition

Article Convolutional Neural Networks for Speech Recognition

Conference Paper Towards End-to-End Speech Recognition with Deep Convolutiona...

Conference Paper Data Augmentation for Deep Neural Network Acoustic Modeling

Conference Paper Exploring Convolutional Neural Network Structures and Optimi...