I have made audio processing projects in the past but it has been a while since I practiced my librosa. I basically tried to make spectrogram generator for some audio recordings. I used a mel spectrogram for my project. I did not find there to be a lot of math used in my project and I found it to be more “theory” heavy with having to learn about FFT and how a mel spectrogram uses logarithm scale to plot the audio. I could instead link you the source I used to learn about librosa. It is pretty theory heavy with some “useless” stuff (even though all of it is pretty interesting). https://www.youtube.com/watch?v=iCwMQJnKk2c&list=PL-wATfeyAMNqIee7cH3q1bh4QJFAaeNv0. This course is pretty good tbh but also remember to skip some parts or speed up the video a bit cause (atleast for me), it gets a bit boring if you are grinding through the videos. It covers all the formulas and the theory used to explain audio processing. I wish you good luck with your project!

@R10 I have recently just started looking into audio signal processing for machine learning which this project is based on and i’ve realised the thing i want to make will need both maths and theory specially in music information retrieval . so yea first i’ll complete ASP and then move on to the MIR . And Thanks a lot , the videos are really interesting and helpful.