[sdiy] Help with math, please

[Matthew Smith]

Quoth Tom Wiltshire at 10/08/11 07:23...
...
> For your envelope generator, you can use your linear value as an index
 > into "exponential curve" lookup table and get the required value out
 > quickly and easily. If necessary, interpolation can be used between
 > table entries to prevent stepping and provide a piecewise-linear
 > approximation to the required curve. As a side-benefit, this
 > method works for arbitrary curves, so you can easily alter your
 > alter your envelope generator to tweak the response,
 > or provide responses which aren't available on standard envelopes.

I was about to say the second part of this until I re-read it properly 
and saw that Tom already had ;-)

 From discussions earlier this year regarding my digital ADSR design, it 
became apparent that attack/decay/etcetera parts could be of any shape
required (including ones that would be hard to calculate on-the-fly with 
limited resources) using lookup tables (LUTs) - and that a shape can be 
used in both directions, simply by reading the table in reverse 
(decrementing the counter.)

Really no different from the way I am implementing my wavetable DCO 
design: a series of samples (whether waves or curves) of equal length 
are held in memory, one after the other. Selecting the appropriate 
wave/curve is just a matter of adding an index value to the address 
counter stepping through the memory. [My current DCO design uses the 
otherwise unused upper three most significant bits of an EPROM address. 
These are switched to select one of eight different waveforms.]

Cheers

M

-- 
Matthew Smith

Business: http://www.smiffytech.com
Blog:     http://www.smiffysplace.com
Linkedin: http://www.linkedin.com/in/smiffy
Flickr:   http://www.flickr.com/photos/msmiffy
Twitter:  http://twitter.com/smiffy