I am doing an undergraduate project where I want to first characterize an instrument based on its pitch and timbre and make an instrument recognition system afterward. Pitch characterization is pretty straightforward, I am planning to use the YIN algorithm to find the fundamental frequency of my samples. Now, timbre on the other hand is quite challenging. I read a lot of papers and they differ in their approaches; some use FFT and some use audio features. I guess I just need someone to steer me in the right direction on how to go about characterizing instrument timbre. Is it possible to extract various temporal and spectral features and characterize audio in terms of them? Do I need some statistical tools for this task? Is SOM or TSNE required? Thanks.
Asked
Active
Viewed 25 times
0
-
YIN is highly overrated. In fact, I don't know a single audio practitioner that uses YIN. (Admittedly besides myself, I only know two other audio DSP guys that have done pitch detectors.) But it essentially offers *nothing* other than that which AMDF or autocorrelation have. Maybe take a look at this, if you're interested. – robert bristow-johnson Dec 18 '23 at 22:20