In-Depth Audio Analysis with Librosa (?)

Hello! My name is Max, I’m currently working on my senior thesis, and would love any help I can get (even if it’s just pointing me in a good direction).

I want to create an interactive experience where people speak into a microphone or play an instrument, and a visual reacts to the audio signal. This is pretty typical TouchDesigner stuff from what I know, but I want to do a more detailed analysis. For example, I want to explore mapping the spectral centroid or timbre to specific RGB values (spectral centroid is red, timbre is green, etc.), so that each sound truly has a more unique visual than just mapping spectral centroid to the entire color value, for instance.

The issue I’m running into is that I don’t know how to measure those metrics within TouchDesigner, and I have no Python knowledge. I found this GitHub project that measures all of these metrics (https://github.com/devbret/detailed-audio-analysis), but I don’t know what do with them. There was also a post on here about Librosa from 7 years ago, but without any Python knowledge, it is hard to understand.

I was wondering if anyone would like to help me or if they have any good resources to help me build this out. I’m open to video calling, emailing, or even just responding in this forum. Thanks!

If you have no Python knowledge, designing a realtime system with librosa reading&writing from/to CHOPs in TD is perhaps a bit much as a first project.

Maybe this project is useful to you:

Also if you have a mac I found this VST3 plugin (which you can load in TD) of the same creator which sends its analytics data like spectral centroid and others out via OSC: