Steven
Backer
sbacker @ ccrma
March
17, 2005
MUSIC
220B - Synthesis Techniques, Compositional Algorithms,
Psychoacoustics and Spatial Processing
CCRMA,
Stanford University
My starting point for the “final project” was the lisp code developed in Assignment #4 – mainly some code that swept frequency from one point to another. The idea for the project was to extend this code to include rhythmic sweeps as well, varying tempo smoothly in a similar manner to frequency. Using oscillators for synthesis is somewhat primitive, but seems to allow for an increased amount of freedom in the resultant sounds. After incorporating the tempo traversals into the code, I found that a variety of “interesting” sweep sounds could be realized by choosing different parameters such as:
initial and final tempo
tempo increment
initial and final frequencies
number of oscillators used in each 'sweep'
Although I obtained some mildly interesting sounds using a large number of oscillators moving to a very high tempo (e.g. 10000 oscillators going from 1bpm to 10000bpm), I still found most of the sounds I was producing rather dull and boring. In an attempt to liven up the oscillator sounds, I added some Moog filters here and there. One strange thing I noted was that the placement of the Moog filter in the code could greatly alter its performance. It seemed that if I put the filter directly in the with-sound call, I would only get that “classic sound” I wanted during the initial and final moments during which the sound was filtered. I'm not 100% sure, but I think this corresponds to the transient states of the filtering versus the steady state. In other words, the “cool sounds” seemed to be mostly when the filter was being initialized and emptied out – not in the steady state region. Moving the filter inside of another loop gave repeated transient response with the same number of oscillators. This may also have to do with the length of the filter. Probably delving deeper into the Moog code would've cleared up any confusion...
Still, the final product was lacking a bit, and so I decided to further experiment with rhythm and add some accompaniment using the fm-violin and fm-drum instruments.
All in all, I'm of the opinion that “THX sound synthesis” is more musical than the results obtained here, but if I had more time (of course) to work on this that could possibly change. For me, lisp programming came very slow, but was rewarding when it worked out as intended (or at least close to what was intended). Often I spent either too much or too little time with the code, as I attempted to complete assignments for many classes at once. Possibly the most important lesson learned from the lisp language comes from using algorithms to save time, and of course, balancing parentheses!