Wolfram Computation Meets Knowledge

Wolfram Summer School


Aishwarya Praveen Das

Technology and Innovation

Class of 2017


I am a second-year student pursuing electronics and instrumentation at the Birla Institute of Technology & Science Pilani, Goa. My interests are in the fields of data science, genetic algorithms and artificial intelligence. I am currently working with my team to build a scalable Hyperloop Pod for the SpaceX Hyperloop Pod Competition II. I am greatly influenced by the accomplishments of Deepmind’s Alpha Go and would like to work on similar technologies in the future. In my free time, I write short stories and play football.

Computational Essay

Artificial Neural Networks »

Project: Music Genre Classifier

Goal of the project:

We aim to create a music genre classifier that allows the detection of the genre of audio/music files. The dataset used for training the model is the GTZAN dataset; it consists of 1,000 audio tracks, each 30 seconds long.

Summary of work:

We divided each song into two parts of 15 seconds each. This way we get more data and our dataset increases to 2,000 songs. We will be extracting the MFCC values of all the audio files by partitioning the song into 15 seconds each.

Results and future work:

We used the built-in classifier, which gives an accuracy of 39.5%, and we trained an RNN on the MFCC values with different architectures and got an accuracy of 75%. We also built a function that can take any song and classify it into a genre.