[sdiy] Waveform mixing - normalization? - Also multimodefilter level normalization

[Mattias Rickardsson]

Yes, filtering will ruin any serious plans of super-duper perception
application perfection. :-) In a way, we are already used to the fact
that filters do these things in synths, and we have learned to
compensate for it. It's a matter of learning to play an instrument.

When switching between waveforms, though, it's of course important to
have their levels designed roughly equal to avoid abrupt level
changes. I usually do this by ear, at mid frequencies, without
filtering. I.e., I would not normalize the fundamentals, and I would
not take filtering effects into account. If I'd design a synth with
simultaneous waveforms from the same oscillator, I'd probably want it
to mix these in some kind of percentage fashion, summing up to 100%
which would correspond to the level of any one waveform on its own.
I'd also be careful to keep the waveforms' fundamentals fairly
in-phase when mixing them, otherwise some unwanted cancelling effects
would occur causing lower levels and nasal spectra.

Btw on a simple analog monosynth with multiple mechanical switches as
waveform toggle selectors, I figure they can easily be connected so
that having two waveforms selected would mix them at 50/50% of their
respective levels. On a more modern machine it could be a bit
trickier, or easier, according to taste. :-)

On a similar brown note, I have actually had a tough time deciding
whether to take filtering effects into account in another
level-related dilemma: when designing a multimode filter. The output
from the different modes can be designed to have equal level in the
passband, but this is sometimes not perfect. On a strongly filtered
sound - say, a typical filtered synth-bass sound with a closing
lowpass filter - the sound is perceived louder if a mode with a less
steep lowpass slope is selected. The difference can be greater than
you first might expect, since such a small part of the sound is the
unfiltered passband. Here it might be advisable to consider raising
the levels a little bit of the filter mode with the steeper slope.
It's not unproblematic though, since a less radically filtered sound
would feel strange if it altered level when jumping between filter
modes that don't do much filtering to the sound.

/mr


On 11 May 2016 at 10:08, Oakley Sound <oakleylist at btinternet.com> wrote:
>> I mean consistent perceived volume from a waveform mixing section.
>
> Whilst this may sound good in theory I think the problem will come when you
> filter the mixed result - which we will almost certainly be doing. The
> relative harmonic structure will change and you'd be left out of kilter
> again.
>
> Personally, I'd stick with equal amplitude. Which I think if you're just
> after switched waveform selection can be done with a R-R ladder.
>
> Tony
>
> www.oakleysound.com
>
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy