On Sun, 12 Sep 2010, Pierre Massat wrote:
Now please, could someone tell me how this works? Like this tan distortion, this table that switches from 0 to 1 with a variably soft transition, what is this supposed to model, and what does it do to the raw waveform?
The use of the table is just a shortcut so that tanh doesn't have to be used live, in which case it may take a good chunk of CPU (especially if you want to use many of them at once).
As I was trying to allude to in this thread, tanh doesn't directly model much, but is a good approximation to several patterns appearing when you are using analogue electronics close to its max capacity. The graph of tanh vs several forms of exp shows this. (see one of the mails).
I'd like to try the sigmoid function but i don't really understand the way the whole thing works. Also, i'm assuming the size of the table matters. Miller's Chebyshev table in the help patch is 129 points wide, i guess the sound would improve if i made it larger, wouldn't it?
The table is optional. If you use a table, more points will be more precise, but also, if you use [tabread4~], you will need many less points than [tabread~] to get a comparable level of precision, and with [tabread4c~] (external) you can get a bit better than that (though for the case of tanh the difference doesn't show as much as for steep or spiky functions)
This is all very exciting for me right now! I always thought distortion could not be properly modelised,
Just because the formulas don't normally show up in a book about the basics of dsp, doesn't mean it can't be done. You see, when mapping from the original theory of a [lop~] or [hip~], to analog electronics, the supposition was that 1-exp(-k*x) can be rounded to k*x. As long as you assume it is true, you can't see where the distorsion can be coming from.
It's like what I learnt about lenses and curved mirrors, in grade 11 and again in grade 13... it's all based on the supposition that tan(x) can be rounded to x, but then it only works well for viewing angles that are quite close to the centre...
| Mathieu Bouchard ------------------------------ Villeray, Montréal, QC