I’ve udpated my deeplearning repo with Tensorflow Speech Recognition challenge. This update includes notebooks that one can run step by step in order see how a audio challenge is solved using CNNs. Also, python scripts are provided under scripts folder for further experimentation through terminal. At the moment results are around ~0.90 on validation with simple CNN architecture but it will be updated by Kaggle winning solutions and also fastai library will be used to leverage more advanced features.
Here is the link which might be helpful to others, I hope https://github.com/KeremTurgutlu/deeplearning/tree/master/tensorflow_speech