[sdiy] Patchable polyphonic synth with FM or AM transmission idea

mskala at ansuz.sooke.bc.ca mskala at ansuz.sooke.bc.ca
Sat Apr 27 17:21:00 CEST 2019

On Fri, 26 Apr 2019, Brian Willoughby wrote:
> A quick search of the internets gave me the impression that group delay
> can be 1 ms for even simple filters at certain frequencies. I’m curious
> to know more about audio filters in particular, such as the worst case
> group delay “latency” for any sort of practical filter that would be
> used in a synthesizer. Is 1 ms common? Rare? Is it small compared to
> what’s possible?

You can make the group delay arbitrarily large at just one frequency for a
one-pole all-pass filter with a high Q centred on that frequency; but it
will only be true over a narrow frequency band.  So then the question
becomes all about what you consider a practical or realistic filter.  I
don't have a solid answer to that but in my Web log entry I was measuring
1/10 of a cycle per pole for the two filters I tested:  about 100us for a
one-pole low-pass and 500us for a five-pole low-pass, both tuned to
oscillate at 1kHz and fed a signal at 1kHz.  I think that's a reasonable
estimate for typical group delay of synth filters not designed
specifically for high or low group delay.  Since it goes by number of
cycles rather than by an absolute amount of time, it will be longer at low

It would be reasonable to expect the delay to be a bit longer for the
Dome filters used in an analog frequency shifter but I haven't done the
math on it.

Matthew Skala
mskala at ansuz.sooke.bc.ca                 People before tribes.

More information about the Synth-diy mailing list