Feb 8
I got a version of the synthesis/Q component working in Csound. It's a pretty weird language, but that fact has made it fun to work with. In order to have the instrument run continuously, a single instance of the instrument plays for an hour. Basically, the instrument sustains a single note for the entire performance. The instrument can receive OSC messages to change its parameters, most importantly amplitude and fundamental frequency (pitch). Rapid shifts in amplitude could be a way to separate notes for the listener.
Some work to be done on Q:
- Add more formants for more unique sound. From what I understand, the first formant determines pitch and the second through fifth formants determine the sounds being created (such as different vowels). The sixth formant and up shape the voice's unique timbre. Ideally, I would like to make the synthesis sound like me, so I'm going to see if there techniques for estimating the formants in a particular voice. If I can't do that, I'll just experiment with formants to make a richer sound.
- Figure out how to best work with amplitude.
More broadly, I need to decide how I want to incorporate timbre. This will likely require changes across all three main files. So far, I haven't been using the MFCC data at all, and now is the time to figure out how best to use it to estimate formants.
I've been working on an outline for the paper. I'll post it here before our meeting.
![[paper clip]](/courses/source/wiki_images/paper_clip_tilt.png)
name last modified size
