[sdiy] FPGA Synth Music

[Scott Nordlund]

>>In fact what stops you from gong all-out and making, say, 32-operator schemes? 
> 
> I don't know this from experience, but I understand from reading that more and more
> operators tend to produce more and more noise or noise-like sounds and that the most
> useful chains are 2, 3 and perhaps 4 operator chains. Not saying that more aren't useful.
> I do have plans for a synth that has the ability to chain up to 8 operators, or make
> smaller chains out of 8 operators (per voice). So something like 5-op and 3-op mixed
> would be possible.
 
I'd expect you should be able to get something interesting out of a very long FM chain
if you have subtlety in mind; very low modulation indices or lots of fixed/low frequency
"waveshaper" operators.  Wouldn't be much fun to program, though.
 
Anyway I guess you probably know that the DX7, etc. used a single phase adder/sine ROM/
accumulator thing with a 96 stage shift register to share it between all the operators.
It's not hard to get different algorithms this way; the data is routed differently 
depending which operator is, uh, operating.
 
You're working with vastly more horsepower than that, but I think a completely
parallelized architecture will make some things a lot more cumbersome.  Especially
if you tried to implement some sort of multitimbral, dynamic operator assignment 
thing (like the Synergy or K150?), you'd quickly run into trouble, even though it's
tempting territory.
 
The more I think about it, trying to develop something to the level of a commercially
available synth is really, really difficult... 		 	   		  
_________________________________________________________________
Hotmail: Trusted email with Microsoft’s powerful SPAM protection.
http://clk.atdmt.com/GBL/go/201469226/direct/01/