I was thinking about the problem of source separation and was wondering if audio fingerprinting could solve it.

What it does

It takes in a mashup song and attempts to identify each song being used

How I built it

Lots and lots of Python audio input/output.

Challenges I ran into

I was originally going to use HPE on demand, but their Speech to Text kept giving me bad music. So I had to figure an alternative method.

Accomplishments that I'm proud of

Learning how to do FIR/IIR and effects in SciPy, also somehow doing the separation

What I learned


What's next for Mashup Smasher

Probably going to try and attach more ML techniques to it.

Built With

Share this project: