[music-dsp] Is beating the same thing as flanging?

[Peter Schoffhauzer]

Richard Dobson wrote:
> OK. let me ask what may be the same question in a different way. It 
> seems to be agreed that a flanger is characterised by lots of 
> peaks/notches distributed over the range. That is of course a 
> consequence of long delay times (reinforced by feedback etc etc etc). So 
> what do we call it if the delay is short (say, under a millisecond?). It 
> fails that definition of a flanger, in having but a small number of 
> notches, and if we are not to call it a type of phaser (despite some IMO 
> obvious similarities), what do we call it? A semi-flanger?
> 
> In short  - are we required to characterise these things by what they 
> sound like, in general terms, or by implementation method? And what then 
> do we call those few (so-called) flangers that were implemented using (a 
> lot of) allpass filters?

Good questions. It seems that there are several gray areas here.

First, if a flanger has only a small number of notches, they're still 
placed equdistant in the spectrum. So as long as there are at least 2 
notches in a flanger, that definition still applies. But you're right 
that there's a degenerate case when the delay becomes 2 samples or less 
- then there's only one notch.

So there's a gray area between a delay and a filter, in fact all delays 
can be considered filters. But it seems that there's a semantic 
distinction at some point where people start calling it a 'delay' 
instead of a 'filter'. When an end user of the effect - a musician, 
composer, producer or studio technician - says "delay" to another end 
user, in at least 99.99% of the time he or she will mean z^(-n) and not 
z^(-1), n being a larger number. This is why I prefer not to consider 
effects with only sample delays in it "delay-based".

In fact in digital audio, nearly all effects have delays in them. I 
can't really think of anything else than multiplication (amplitude 
modulation and panning/stereo effects) an distortion (clipping / 
waveshaping / bitrcushing) that does *not* have delays in them. Pretty 
much every other audio effect has delays in them, just a few examples:

- compressors/dynamics: delays in the detector, either IIR or FIR, 
feedback compressors also have a delay before the feedback

- FFT: in a realtime application, incoming samples are first stored in a 
buffer, so the whole thing is happening in a delay buffer

- resampling: the interpolator remembers previous samples, at least the 
last one in the zero-order case

- frequency shifter: there are delays in the Hilbert-transformer

... and so on. So I guess I can safely say that basically pretty much 
the whole digital audio is "delay-based" in that sense that all effects 
contain delay(s), except amplitude modulation, panning and distortion.

For this reason, I like to make a semantic distinction somewhere, and 
for me that's z^(-n), n being higher than 1. It would be easier if we 
simply called everything a 'filter', but probably the reason for the 
distinction is that higher order delays are semantically different for 
the human perception. I think there's no distinct borderline between 
'filter' and 'delay'.

So, in short, I can't answer those questions. I just put my 2 cents here.

Peter