gaudreau@juggler.East.Sun.COM (Joe Gaudreau (Spaced for Rent)) (03/02/91)
About a month ago someone posted an article chock full of math and sin's and golly, all sorts of stuff. Fool that I was, I didn't save it. The article converned itself with how a sound is put together and how to mathematically take it apart... Could someone PLEASE mail this article to me!!!!???? I'd really^a-lot appreciate it! Thanx. --- On another note, does anyone have any idea how to: (a) decompose a "sound" into the various parameters that a Dx7 uses? Or pretty close. It should be possible to get an okay approximation with a lot of processing. (b) come up with the "sound" or a Dx7 patch, based purely on the patch info... I think there is a commercial product for this even... Joe -=-
dbell@cup.portal.com (David J Bell) (03/04/91)
>About a month ago someone posted an article chock full of math and sin's >and golly, all sorts of stuff. Fool that I was, I didn't save it. >The article converned itself with how a sound is put together and how >to mathematically take it apart... >Could someone PLEASE mail this article to me!!!!???? I have no copy of the article, though I seem to remember one... >On another note, does anyone have any idea how to: >(a) decompose a "sound" into the various parameters that a Dx7 uses? Or > pretty close. It should be possible to get an okay approximation with > a lot of processing. This is a REALLY tall order. The decomposttion of a subjectively- interpreted sound into parameters that could reproduce the effect is an extremely complex task. What it really requires is a lot of experience building your own patches and modifying others', to see the effect of various "tweaks". There are some basic approaches: First, you can probably analyze the amplitude envelope fairly easily, even without instrumentation; from this you can create the ADSR parameters and get pretty close. Second, the overall timbre may be recognized as some common patch algorithm, with some hint as to the relative frequency and amplitude of the carrier and principal modulator. This is where the experience starts to really come into play. A third level of analysis may expose the time-varying frequency or timbre changes, allowing the programmer to tweak the ADSR envelope of the modulators, or to add in some LFO. From this point on, it's a matter of experiment, adjust, and repeat ad infinitum... >(b) come up with the "sound" or a Dx7 patch, based purely on the patch > info... I think there is a commercial product for this even... Sure! This one is easy: A Yamaha DX-7!!!! Seriously, you certainly can simulate the tone generators of a DX, even in software. I don't know of any commercial product that does this, but I'd be surprised if one didn't exist somewhere... Over all, I'd suggest you look up some of the original work done at Stanford University and San Jose State. The initial work was by John M. Chowning at Stanford (early article published ca. 1973), and more expansion of the technique by Alan Strange at San Jose. This was really the origin of the Yamaha-style FM synthesis. It does get heavily mathematical, but there are plenty of practical examples in the literature. Dave dbell@cup.portal.com