[sdiy] Chorus question

[rburnett at richieburnett.co.uk]

As well as Tom Wiltshire's analysis, I seem to remember Antti 
Huovilainen presented a paper at DAFx'05 called something like "Digital 
Models of Analog Modulation Effects".  I can't find it online as a PDF 
anywhere though.

The link for DAFx'05 seems to be broken!

-Richie,

On 2014-04-23 19:17, Scott Nordlund wrote:
> I was actually just thinking about this the other day before getting
> confused and giving up. If you're doing a DSP delay with a fixed
> sample rate and interpolated delay read, it's pretty straightforward.
> The pitch change is proportional to the derivative of the modulating
> signal, so a triangle wave modulator results in square wave pitch
> change. There's some additional complexity here if you want to ensure
> that the detuning is symmetric, i.e. the amount of detuning in cents
> is the same for the rising and falling portions of the modulator. The
> solution is to vary the shape of the modulator depending on modulation
> depth.
> 
> But anyway, you'd think that since a triangle wave is frequently used
> as a modulation signal in BBD delays that it would be the same. It's
> similar but not identical, and I think this is part of why BBDs are
> sometimes preferred for chorus effects.
> 
> So here's the problem: a modulated DSP delay is like a tape delay with
> a movable playback head. A modulated BBD delay is like a tape delay
> with variable tape speed. I was trying to formulate some way to make
> these equivalent without using an interpolated delay write (which
> would be the more or less straightforward way to do it). It's easy to
> demonstrate the difference if you have, for example, a VCO-clocked
> digital delay. You can turn the delay time knob and notice that the
> pitch change persists after you stop turning for the length of the
> delay. For a DSP delay, the pitch change would stop instantaneously.
> 
> I tried to sort of restate the problem, but didn't work through it all
> the way. Basically, you can think of the variable rate case as
> traversing a fixed distance d at a variable speed v(t). Distance
> traveled D(t) is the integral of v(t) with respect to time. I think
> the goal is to solve for the time interval T(t) that it takes to
> traverse
>  the distance (i.e. D(t) = d), and use this to make the two cases 
> equivalent.
> 
> I think you could emulate the BBD behavior in DSP by putting the
> modulator through some sort of self-modulated delay. But I'm hesitant
> to really give this the attention it warrants, as it's distracting me
> from other stuff...
> 
> 
>> From: tom at electricdruid.net
>> Date: Tue, 22 Apr 2014 17:17:58 +0100
>> To: tom at electricdruid.net
>> CC: Synth-diy at dropmix.xs4all.nl
>> Subject: Re: [sdiy] Chorus question
>> 
>> Ok, here's a link to what I've got so far:
>> 
>> http://www.electricdruid.net/ChorusStudy.html
>> 
>> You can see the way a simple sine modulation gets bend out of shape by 
>> the changing delay.
>> 
>> T.
>> 
>  		 	   		
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy