Haible Juergen Juergen.Haible at
Mon Aug 14 16:03:55 CEST 2000

You don't need a delay, you need a chain of all pass filters:
Product of (s_i - s)/(s_i + s), i running from 1 to N, s being the complex
s_i being the (normally real) poles and zeroes of each stage.
What you're going to modulate over time are the s_i values. 
And add an overall feedback around the whole chain or around parts of the
feedback with factors between -1 and +1, and with a slight rolloff for
frequencies: s_fb/(s_fb + s).
All linear so far. Now comes the tricky part: Implement a soft clipping
to avoid nasty overload when the input signal hits a resonant peak. Make
that you don't get aliasing effects from the (nonlinear) soft clipping
and your sampling rate (;->).


	>last night I was working on my software synth again, and to thicken
up the
	>sound I added a chorus, which is just a few delay lines of which
the delay
	>times are slowly modulated. A phaser is almost the same, where a
very short
	>delay (below 10ms) is also modulated slowly, but the effect,
compared to the
	>chorus, is absolutely unsatisfactory. It just doesn't go 'deep'
	>What's wrong with my algorithm? My Boss SE-50 multieffector has
quite a set
	>of parameters. Speed and depth are easy. Depth controls the maximum
	>time. But what about manual and resonance? Probably some filter is
	>incorporated. Or some strong feedback path? Or are more delay lines

More information about the Synth-diy mailing list