I'm a phd student in music theory, interested in computer modelling of intonation practices. In terms of the course I'd think it might be interesting to on something that would result in an improvisation performance enviroment - though I'm don't really have anything specifics in mind. My training is predominantly in music but I do have some programing skills (mainly C++) and am currently trying to get up to speed with matlab.
02:05:06 Mashup homework
I constrained the songs I used as source material in an attempt to achieve a certain amount of cohesion. I chose two mid 60's pop songs with a merseybeat sound Beatles "Eleanor Rigby" BeeGees "New York Mining Disaster, 1941" and replayed a late 70's post punk song Joy Division "She's Lost Control". I recorded each playthrough direct to a file and chose the file that I thought worked the best. Then I cheated(?) a little, I spliced up a single file into segments and removed the segments that I wasn't overly fond of. So the order of the resulant track, Eleanor lost control, 1941, is the same but parts are missing (hence it only being 1:08 sec long) - as you may notice, I became quite fond of the pizzicato effect that emerged on occasion from the Eleanor Rigby strings.
03:07:05 - Max externals (aka the Matlab-Max tête-à-tête ) Various searches on max externals that read matlab code yielded nothing more than an unanswered query on the max/msp forum. I did find the following tutorial, and would be interested in working on an external to make this 'conversation' possible.
02:20:05 - Video tracking stuff I've been focusing on the cv.jit extensions to Max/MSP/Jitter, mainly trying to get my head around the processes and the possibilities.
Some ideas:
- Style Remixer: this idea builds on Mike's idea of a dance party in a box and could make use of Ron's percussive
- extraction work. Essentially the inputted track is analyzed and then resynthesized in a variety of different styles - the choice of style being determined by the occurence of certain pre-determined gestures. These gestures would have to be distinctive enough that they can be recognized but also simple enough that the user could catch on. The style characteristics applied to the resynthesis would affect both the groove and/or swing of the rhythmic pattern and the effects applied to the drum sounds. Assuming that the original drum sounds could be almost completely filtered out and that the modifications to the rhythm for each style are 'tight' enough that they lock to the global metrical structure of the original; the resynthesized part could be played with the original (filtered) recording while (ideally) being modified more or less on the fly.
- events from an inputted audio file. But instead of modifying the rhythmic nature of the track the idea is to give the user control of which drum sounds are reintroduced. Effectively the space around the user would have to be mapped to correspond to the palette of drum sounds, when the user's moves their hand in that space the corresponding drum sound is triggered at the next (subdivision of the) beat - though I'm sure there would be some latency issues with this.
--> along the same lines the various regions around the user could be mapped to controllable musical
- parameters
Essentially I'm intersted in developing a system where the user's physical actions effect on the sound being produced, which is intutive enough that it can be easily learned.
jcd2102 columbia edu