[sdiy] Digital noise generation

[Magnus Danielson]

From: Scott Gravenhorst <music.maker at gte.net>
Subject: Re: [sdiy] Digital noise generation
Date: Tue, 08 Mar 2005 08:53:17 -0700
Message-ID: <200503081653.j28GrGc11418 at linux6.lan>

> >What polynomials did you use?
> 
> This may be very important.  I've listened to 24 and 32 bit _maximum_length
> LFSR noise generators, personally, I found even the 24 bit to be as smooth as
> silk.  Shhhhhhhhhhhhhhhhhh...

Yeah, that should be enought for most people. The longer sequence, the less
impressive ka-chunk you get and less often. The sequence length grows much
faster (exponential) than the ka-chunk amplitude.

> >If you dig in the Synth-DIY archive I published a list up to length of 64.
> 
> And I have a list on my site up to 168 stages.  (c:
> 
> http://home1.gte.net/res0658s/electronics/LFSRtaps.html
> 
> - though it appears to be down right now...  grrrr...

Well, well... ;O)

> >
> >Did you use a parallelized version?
> >I can help you with that if you need.
> >
> >Did you amplitude weigthed the outputs of the parallelized version? 
> >Again, I can help you, but it is really a quick-and-dirty approximation of
> >filtering with a one-pole lowpass filter. Exponential drop-off in amplitude
> >should work the "older" the bit is supposed to be.
> 
> Heh, I did it the cheesey way, the output is one bit which is then fed to a
> low pass filter.  It works for me...

That is what I would do in a hardware implementation yes, but when hacking
software it is a bit different optimizations you need to do.

Cheers,
Magnus