This notebook was published in a Kaggle competition as an introduction to MFCC and how to implement it.
You can check the competition and get the dataset in the link: https://www.kaggle.com/c/freesound-audio-tagging
Also, you can check another notebook that I wrote in the same competition. It is a farly simple code for pre processing and removing uninteresting parts of the sample to reduce leasrning time: https://www.kaggle.com/ilyamich/remove-uninformative-parts-from-the-audio-files