[sdiy] Resonator type filters

[Tom Wiltshire]

Is there some reason why string filters use a large number of fixed bands, rather than fewer variable ones?

Mostly, we're trying to simulate certain resonances that occur in the bodies instruments (brass/woodwind/strings/etc) and there aren't 30+ of those.

It seems to me that (like the PolyMoog resonator from earlier) three, four, or five variable filters (SVF is the obvious choice) would be able to produce all the complex responses you might need, but with a saving in circuitry.

What are the differences between the two approaches? Why might you choose one over the other?

Thanks,
Tom

==================
       Electric Druid
Synth & Stompbox DIY
==================


On 30 Jun 2017, at 03:31, Ian Fritz <ijfritz at comcast.net> wrote:

> Just to be sure we are all together on this, a string filter bank needs high enough Q's to give something around 10-12dB peak-to-valley ratios. And a common adjacent channel frequency ratio may be taken as the fifth root of 2.1, following the original Matthews/Kohut/Burhans research. Note that octave repeating frequencies would be a disaster! If anyone knows of a graphic EQ that can do this, I would love to hear about it. My personal take on a string filter is described in Electronotes EN#107 (1979). It was built a couple of years earlier. It has 36 resonanators in banks of six, with adjustable amplitude for each frequency and adjustable Q for each bank.
> 
> Ian
> 
> 
> On 6/29/2017 2:44 PM, Richie Burnett wrote:
>> The Q factor of each filter is obviously higher for something like a 30 band graphic EQ where the bands are 1/3rd octave spaced, than for something like a 10 band EQ with 1 oct wide bands.  As you've said the Q factors aren't particularly high though in a typical graphic EQ.  In fact I don't think they are even fixed in low-cost GEQs, because i've seen much more expensive GEQs that are specifically marketed as "constant Q" and are supposedly superior.
>> If you want to simulate body resonances of musical instruments the graphic EQ probably isn't the most precise tool to use, and may even fall well short in terms of the required Q factor.  I've never tried, but it's quite a coarse tool.  I have tried using graphic EQs to correct for nasty room modes over the years, and it doesn't work very well.  In both cases I suspect that a decent multi-band parametric EQ would be much more precise and better suited to the job.  Room modes at bass frequencies tend to have quite high Q-factors and are very narrowband, so they demand very narrow notches from an equaliser, if room equalisation is your thing.  A graphic EQ is nowhere near precise enough in my experience, at least for small home studios.
> ijfritz.byethost4.com
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy