Phase (was: ToneShiftEQ v0.3.0 — First public release)

Hello Fons,

I (almost?) completely agree with what you say. Just some clarification.

On 2026-07-04 13:04, Fons Adriaensen wrote:
...
> In audio it doesn't matter at all except maybe in some very
> rare corner cases - those where very steep filtering is done
> in the region where human hearing transitions between T-domain
> and F-domain perception and things can become ambiguous.
...

Probably here for "audio" you mean "everything between the microphones 
output pins and the speaker input terminals". If this is the case, well, 
forget about the "almost?" above. It was already so even back in the 
good old analog days, with just few more corner cases.

Extending the range to the whole audio chain, regarding the mics I've 
seen some (competent) people affirming that they are minimum phase 
device, some other questioning it, using some measurements whose 
reliability I was unable to evaluate. In both cases I expect any phase 
alteration to be inaudible or close to inaudible, considering that they 
usually provide a smooth, almost flat magnitude response.

After the speaker terminals, things change. The audibility of a typical 
3 way 300Hz-3kHz LR24 crossover has been proved in some double blind 
listening tests, and went uncorfimed in some others, so it's probably 
close to the limits of audibility. Don't know of any reliable literature 
about the speakers themselves, didn't check recently, but I expect their 
phase response to be audible, even though way less important than other 
loudspeaker issues. For the room, the excess phase of a typical room has 
been proved audible, no surprise here.

In all cases, audibility strongly depends on the test signal used, 
because of temporal masking.

This is pretty much what I gathered about phase audibility in the last 
20 years or so. Hope it helps.

Bye,

-- 
Denis Sbragion