My Exploration of Deep Learning and Audio Analysis

The capstone project is the final hurdle to jump to become a General Assembly DSI graduate. We had the freedom to decide on pretty much any topic that we want so it was a bit overwhelming given the breadth of areas to explore in Data Science. Coming from a physics background, I knew I wanted to incorporate some of my prior knowledge. I’m always fascinated by the physics of sound (I did my senior research project on echolocation and bats) and since I love music I easily made my decision to do a project regarding audio analysis.

Next was deciding the problem or question I wanted to answer. I often hear tracks that I like and want to find similar tracks but I don’t know what style or sub-genre is so I figured it would be fun to build a classifier and potentially an audio tagger.

Another exciting thing was I would be exploring Neural Networks. I hadn’t ever heard of them before this class and the more I found out the more fascinated I became. I’ll have to use a recurrent neural network and possibly convolution neural network and I’m really excited to dive into those.

So far the biggest roadblock is that I don’t actually know much about neural networks yet; we will learn them formally in our last week but that will leave me with a little less than two weeks to build a model and package it into a concise presentation. I’m also a little rusty on the technical side of waveform and audio analysis since it’s been a few years since I was in school. However, there are a wealth of resources online and I’m excited to make the most of it and learn all I can in the next couple weeks.

If you love music like me (especially more obscure types) then I hope this can be a simple tool to discover new artists and learn what types of music you really enjoy. Even if the outcome for this project itself isn’t a robust and effective model, I am looking forward to continue building upon it and see what improvements I can make beyond GA as my skills grow.

Related posts:

Mobile application is a software program that is developed by mobile developer or programmer for use on a particular platform or device like Android, iOS and Windows. Absolute mobile app means the…

Thank you for stopping by and taking your time to read this article that contains all the information you need to begin/continue submitting to Know Thyself, Heal Thyself (KTHT). I am updating the…

To get the training data for machine learning in agriculture first of all you need to decide what type of AI model you are working or which type of agriculture related datasets you need for your…