Developed a deep learning stack for genre classification using MFCC features of the audio. Designed an architecture with inception, convolution, bidirectional GRU layers and obtained 15% improvement in accuracy when compared to feed forward networks.