[sdiy] Fixed filter bank questions

Ian Fritz ijfritz at comcast.net
Thu May 9 05:49:23 CEST 2019

2.1^.2 was first used by Ralph Burhans, IIRC, for his clavier resonator. He found this to give the fewest overlaps of harmonics of different notes.

A few months ago I ported my old 36-stage analog bank into Reaktor and added a variable spread parameter to experiment with. It can have a uniform spread, or random numbers can be folded in for a less regular one.  String resonators to formant filters and beyond. The things we can do!


> On May 8, 2019, at 6:55 PM, Dave Leith <dave.leith at gmail.com> wrote:
> Check out the frequencies of the Serge Resonant Filter. Intentionally does not use doubling of frequencies.
>> On Wed, May 8, 2019 at 5:24 PM Ben Bradley <ben.pi.bradley at gmail.com> wrote:
>> "Frequencies every half-octave around 1000 Hz (LP at 88, 125, 177,
>> 250, 354, 500, 707, 1000, 1414, 2000, 2828, 4000, 5657, HP at 8000
>> Hz)"
>> I spend a few minutes thinking about this, and what you'll have
>> musically is when a note "hits a frequency" of a filter, not only will
>> a fundamental be emphasized, but also its even and multiple of 3
>> harmonics as well. Other notes would have one or a very few harmonics
>> emphasized. I'm not sure how this would sound, but it did seem
>> somewhat lopsided.
>> I had read the previous discussions on filter banks and comb
>> filtering* and discovered at least one tidbit I didn't notice or had
>> forgotten. Several references use the formula 5th root of 2.1 (or in
>> this case you might use square root of 2.1, or to place them closer,
>> square root of 1.9) for the frequency ratios, so octave and other
>> harmonics of one filter frequency don't necessarily fall into other
>> filters.
>> For a one-off design, for the experimentation and development that you
>> appear to be doing, I'd want trimpots for frequency and/or Q, to be
>> able to more easily tweak things and hear how the sound changes. This
>> could help to determine if 1 percent capacitors is really worth it, or
>> even audible. I'd want to hear the difference, but I'd offhand guess
>> that 5 percent is good enough.
>> * Perhaps not a "good" name for this, since the comb filtering I'd
>> seen refers to dips in response caused by adding a phase-shifted or
>> time-delayed signal to the original, but it's also been used for fixed
>> filter banks.
>> _______________________________________________
>> Synth-diy mailing list
>> Synth-diy at synth-diy.org
>> http://synth-diy.org/mailman/listinfo/synth-diy
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://synth-diy.org/pipermail/synth-diy/attachments/20190508/a0c08cb5/attachment.htm>

More information about the Synth-diy mailing list