"The secret of big speakers"

[Cosmik]

... now, if we just had that IR database...

It's been done already for users of the Devialet SAM system:
https://www.devialet.com/en-gb/expert-pro-sam-ready-speakers/

 

[Wombat]

On LF phase audibility, I am still waiting for irrefutable evidence vs conjecture and opinion.

The longer the explanations, the less credible are the posits.

 

[Juhazi]

Wombat, you just have to trust your ears And if you don't hear the difference - be happy! I have read about experience with Devialet SAM, and everyone doesn't hear the difference.

Actually it is really funny, how we pay attention to so different things and flaws in music reproduction, even those who call themselves hifi enthusiasts. I am happy to fail recognizing different cables, cd players etc.

I guess that I should test phase correction, but I harly ever use laptop as source. I use Foobar and Spotify Premium, is it possible to add a phase convolver plugin? My main speakers utilize minidsp 4x10HD which can't handle FIR.

 

[andreasmaaan]

On LF phase audibility, I am still waiting for irrefutable evidence vs conjecture and opinion.

The longer the explanations, the less credible are the posits.

I agree. We began with a plausible statement from @KSTR about his own research and various published studies (non-peer reviewed, which is fine IMO) that he said existed and were available.

But those studies haven't materialised, and now we're back in the domain of subjective generalisations about "how it sounds".

Honestly, I almost want to believe that the established research is wrong on this. But still we have absolutely no evidence that it is wrong, and plenty of evidence that it's right (for example, good agreement between the results of a number of different studies, with consistently similar thresholds being found using numerous different test setups and methods).

 

[andreasmaaan]

Wombat, you just have to trust your ears And if you don't hear the difference - be happy! I have read about experience with Devialet SAM, and everyone doesn't hear the difference.

You mean everyone doesn't think/say they hear a difference. Or have audible differences been demonstrated under controlled conditions?

I guess that I should test phase correction, but I harly ever use laptop as source. I use Foobar and Spotify Premium, is it possible to add a phase convolver plugin? My main speakers utilize minidsp 4x10HD which can't handle FIR.

You should be able to do it with rePhase. Link on use with miniDSP here.

 

[andreasmaaan]

If we can make a linear phase speaker , what’s the advantage in not doing so?

We already split hairs with our DAC reviews/requirements , wanting best theoretical performance so why should this be different in this instance ...

Make it the best possible and worry about if you can hear it ,,, later or ,,never.

I totally agree with you, but as a speaker designer you're trying to make the best possible design trade-offs within given limitations, and in particular a specific budget.

Phase linearisation is not "free" in the sense that it introduces pre-ringing (although of course this is very unlikely to be audible in the ranges we're talking about here) and latency.

More importantly though, it's not free in terms of financial cost, since it requires DSP capabilities that are more expensive than alternatives which are not phase-linearisation capable. So the question whether this extra money should be spent (and charged to customers) is an important one from this perspective, at least.

Or to put it another way, if transparency in this domain is achievable for X cost and phase linearity adds additional costs without increasing transparency, it's a poor investment.

 

[Juhazi]

I am very happy with LR2 and the step response it has, but I just noticed that I have a 2x4HD box lying around - I can use it to do FIR for my main speakers AINOgradients! Putting a 2x4HD between my pre and 4x10HD adds also a new ADDA process, but who cares - at least for testing...

 

[andreasmaaan]

I am very happy with LR2 and the step response it has, but I just noticed that I have a 2x4HD box lying around - I can use it to do FIR for my main speakers AINOgradients! At least for testing...View attachment 15694

Are you sure that's the step response of a speaker with LR2 crossovers there? It looks to me like there's 360° phase rotation at each crossover point.

 

[Juhazi]

Oops! Only lowest xo is LR2, others LR4. That outdoor measurement is quite old, looks like I haven't copied newer LR2 version measurements on this computer. Let's wait if I can run some tests tomorrow!

 

[hvbias]

While not on the thread topic The discussion on phase is interesting, there are more comments in this thread: https://audiosciencereview.com/forum/index.php?threads/interesting-article-by-mitchco.661/

(Just pointing it out for additional discussion. I myself like the idea of linear phase throughout and will be aiming for it)

 

[Cosmik]

On LF phase audibility, I am still waiting for irrefutable evidence vs conjecture and opinion.

The longer the explanations, the less credible are the posits.

I never cease to be surprised at the idea that the phase-correct proponents have to defend or argue their position, while the phase-mangling brigade swan about as if they own the place.

 

[svart-hvitt]

I never cease to be surprised at the idea that the phase-correct proponents have to defend or argue their position, while the phase-mangling brigade swan about as if they own the place.

We also have a discussion on high resolution material, where some think one should produce and consume at max 16/44 because the majority shouldn’t have any use of higher resolution, now and in eternity.

So it seems like some people put practical use over theoretical correctness.

Not surprising, is it?

 

[andreasmaaan]

I never cease to be surprised at the idea that the phase-correct proponents have to defend or argue their position, while the phase-mangling brigade swan about as if they own the place.

I think you're looking at it the wrong way if you think there are two opposing positions here.

We all must agree on the following points:

• non-constant group delay is a theoretically well-understood and measurable result of conventional crossovers, and of minimum phase changes in the frequency domain, including those resulting from placing drivers in sealed or ported boxes
• as humans, there are certain thresholds below which non-constant group delay becomes inaudible

I and @Wombat (if I understand him correctly) are saying that a transparent device is one that does not exceed these audibility thresholds.

Are you agreeing with this, but saying that the experimentally obtained thresholds are wrong?

Or are you saying that transparency is irrelevant? That even if two speakers sound exactly the same to all listeners, one can still be better because it exhibits measured superiority below thresholds of audibility?

 

[andreasmaaan]

And further @Cosmik, if the latter is your position, are you saying that a speaker with constant group delay but X degree of pre-ringing is theoretically superior to a speaker with zero pre-ringing but X degree of non-constant group delay?

If so (and assuming that both speakers sound exactly the same to all listeners), on what basis do you say this?

 

[DonH56]

Things are getting a bit fuzzy... I think you mean "phase changes" instead of "group delay" but again this is not my day job.

Group delay (in units of time, e.g. seconds) is the negative of the derivative of phase with frequency. That is, it is (-) the change in phase divided by the change in frequency. If phase changes linearly over frequency, a straight line, then the group delay is a constant (the negative of the slope of that line). Constant group delay is desirable in systems where things like pulse integrity is important (much of my work life). If you put a pulse, or step, into a system with constant group delay, then all frequency components of the signal are delayed equally, and you get the same pulse or step out, just delayed in time.

Minimum phase systems have a specific definition that is a bit hard to follow if you aren't up on your mathematics or systems classes. See e.g. https://en.wikipedia.org/wiki/Minimum_phase . A minimum-phase system also has minimum group delay but that does not mean it is linear phase thus does not in general mean the group delay is constant (constant meaning it's just a number, like 1, or 2, or 3.14159265...)

There are arguments for minimum phase systems (including crossovers) but I personally would focus on linear phase as that maintains the best time-domain response. But, how much that matters in the real world of audio I could not say. Various research, at least in passing, goes both ways. Paraphrasing, Toole and many others say phase is not really important, and with real source material such as music and movies we aren't bothered by phase changes since our ears and brain process mainly frequency (magnitude) information. Others say we are very sensitive to phase as evidenced by the ability to locate sources in space based upon differences in phase (not sure how they distinguish from pure time delays and I've always thought that argument a little lacking) and point to their ability to detect "smeared" transient responses when the phase is not linear (the latter at least matches what the system actually does to the time domain response).

So my little pea-brain suggests that minimum phase, while may be desirable (and may be required to implement certain filter structures), does not really matter much. The magnitude of group delay is not an issue since we can compensate for it using DSP or analog all-pass filters. A system that is not linear phase, that is does not have constant group delay, will not have as good time-domain response as a linear-phase system but there is debate about whether we can hear the difference.

Hopefully a real expert can comment and clear things up.

 

[andreasmaaan]

I agree with your definitions @DonH56. It's normal (albeit not technically accurate perhaps) IME to take the point at which the impulse begins to excite the speaker as the zero point for the purposes of describing group delay. This is what I meant by "zero group delay". You're right though, "constant group delay" is a clearer and more accurate term (have edited previous posts accordingly).

Which particular aspect of the discussion do you believe needs clearing up?

 

[DonH56]

Zero group delay, while technically a constant, is not in general the same as constant group delay. It's also not realizable in practice but I do understand using a relative reference point and going from there, no problem with that. I had trouble relating your post to what I understand from my background but I think our fundamental definitions and understanding simply differ. I also think what I know is not really relevant to where this discussion has gone and frankly, except for maybe trying to help possibly clarify some definitions from an anal design engineer's standpoint, I do not feel my participation would help anyone.

 

[andreasmaaan]

Zero group delay, while technically a constant, is not in general the same as constant group delay. It's also not realizable in practice but I do understand using a relative reference point and going from their, no problem with that. I had trouble relating your post to what I understand from my background but I think our fundamental definitions and understanding simply differ. I also think what I know is not really relevant to where this discussion has gone and frankly, except for maybe trying to help possibly clarify some definitions from an anal design engineer's standpoint, I do not feel my participation would help anyone.

It was a fair point, and it would have been better of me to be more careful with my language in my previous posts, which I've now edited.

I think (hope) that my posts are now clear, and that we're on the same page.

The reason I didn't choose the term "phase change" is that research into audibility in this area tends to express results in terms of group delay; in any case as you point out the two are different ways of representing the same information.

 

[Wombat]

I never cease to be surprised at the idea that the phase-correct proponents have to defend or argue their position, while the phase-mangling brigade swan about as if they own the place.

I am happy either way re Lf phase audibility but without credible confirmation the arguments, including "phase-mangling", are moot.

 

[hvbias]

Linkwitz has what I have found an exceptional paper on linear phase- http://www.linkwitzlab.com/Attributes_Of_Linear_Phase_Loudspeakers.pdf

A combination of subjective observation plus published papers. I am still digesting it now, 30 pages with many more once all the additional references are read!