[music-dsp] Semi OT: How Much Math in Practical DSP?

[Veli-Pekka Tatila]

Hi,
First a little intro as I've mostly been lurking here apart from a couple 
newb posts. I'm a 20-year-old, visually-impaired Finnish computer enthusiast 
interested in practical synthesis. I have a number of synths, like 
programming them a lot, and as I do know C, CPP and Java somewhat, I think 
it would be nice to create some softsynths of my own at some point. So I'm 
approaching this more from the musician|programmer point of view rather than 
engineer|mathematician.

I'm primarily studying information processing science (which is 
unfortunately a less "scientific" form of computer science) and am pondering 
whether I should do math as a minor subject. My primary motivation is that 
it would appear any DSP work seems highly mathematical (more so than 
programming or practical math in general) and I have a vague feeling that it 
would be really useful to know some higher math in other contexts, too. I 
only did the short courses in highschool and they went well. However, I'm 
not usually into totally abstract math that doesn't have any practical value 
in real life or on the computer and being visually impaired means I gotta do 
much more work than the sighted in general. Geometry and anything purely 
graphical (e.g. even UML) is particularly difficult, of course.

I'm not quite sure how to put it but how much math do you need in order to 
do any practical DSP work related to synths? It would be nice to understand 
DSP as thoroughly as possible in order to really be able to design something 
new and know what one is really doing. However, I've heard that truely 
understanding say FFT would require years of study and I'm
probably not into math that much.

Can you really design something new and use say FFT as a programmer without 
really understanding what's going on behind the scenes? I suppose this comes 
down to a more philosophical and abstract question.

Apart from obvious computer aspects like coding and usability, how's 
creative DSP work like? Is doing a good sounding synth more an achiement in 
practical higher math or a feet of programming and intuition. I reckon it 
depends on the person and project, too.

On a side note, here's some info on projects I'd like to do some day:

-a NES synthesizer. there are a few emulating the SID chip but not much 
related to other computers or consoles. I actually have a ready-made NES 
sound lib from one guy and need only learn the basics of VSt as well as 
handling MIDI events, driving the synth, coding the GUI and including new 
virtual modulators like non-native extra LFOs and envelopes.

-an open-source modular synthesizer, the Linux equivalent of NI Reactor. 
However, it should be more about subtractive synths and as friendly as the 
Nord Modular, at least. AN alternative text-mode UI would be nice, 
specifying patches in  a special XML dialect. Coding new modules would be 
possible in some interpreted programming language.

-A relatively simple MIDI drum-machine plug-in. However, it would not do 
quantizing of realtime input and would also be able to analyze the rhythm 
from an audio source in real time. Also no arbitrary pattern length 
limitations.

-a MIDI to SAPI wrapper that would let you compose vocal parts for any 
MIcrosoft SAPI 4 or 5 compliant synth. MOstly useless but would be highly 
cool whatsoever.

-An MSAA wrapper for VST synths that would pass widget descriptions and 
names from totally custom synth GUis to screen reader programs through SAPI.

-A retro synth environment. It would let you run any older non-VST or DX 
audio app as a plug-in. The app would see the VST MIDI and audio ins and 
outs as virtual MIDI and DirectSound devices.

Loads more that I don't care to list right now.

Thanks for any help in advance.
If you think this is off-topic, feel free to post replies to me directly.

-- 
With kind regards Veli-Pekka Tätilä (vtatila at mail.student.oulu.fi)
Accessibility, game music, synthesizers and more:
http://www.student.oulu.fi/~vtatila