As an undergraduate I became interested in how timbre can be used to identify musical instruments. This was largely due to my first UROP (undergraduate research gig) with Keith Martin at the MIT Media Lab. Keith’s thesis was on identifying musical instruments from spectral features, and I worked a bit on this under Ryan Rifkin in a later UROP. I’ve been catching up on podcasts during my commute to campus this week, and a semi-recent Science Friday piece on the Steinway factory was on deck for this morning.
The piece talks about work in Agnieszka Roginska‘s lab at NYU, and in particular work from a paper from last year on measuring radiation patterns in piano soundboards. The radiation patterns are pretty but a bit hard to interpret, largely because I’m way out of the acoustical signal processing world. However, what’s interesting to me is that we’re still largely focused on overtones/cepstral coefficients. I wonder about how one might discover more interesting features to characterize this data. (I know someone will suggest deep learning but color me a little skeptical).
As a side note, one of the recent popular articles from JASA is on the acoustics of coffee roasting.