Science thread: audibility of phase distortion in loudspeakers

amirm

Banned
Apr 2, 2010
15,813
37
0
Seattle, WA
There was a good discussion started in another thread, http://www.whatsbestforum.com/showt...nglish-version&p=338075&viewfull=1#post338075, on audibility and requirement of proper phase response in loudspeakers. Good suggestion was made by Michael to create a separate thread on that and I thought this would be a good place for it.

So let's have the necessary theory and listening tests results posted here and have a good discussion on this usually murky topic. There are definitely two schools of thought here as evidenced by loudspeakers with stepped drivers and those without it for example.
 

Groucho

New Member
Aug 18, 2012
680
3
0
UK
There was a good discussion started in another thread, http://www.whatsbestforum.com/showt...nglish-version&p=338075&viewfull=1#post338075, on audibility and requirement of proper phase response in loudspeakers. Good suggestion was made by Michael to create a separate thread on that and I thought this would be a good place for it.

So let's have the necessary theory and listening tests results posted here and have a good discussion on this usually murky topic. There are definitely two schools of thought here as evidenced by loudspeakers with stepped drivers and those without it for example.

Yes, I think it is an extremely murky topic. There have possibly been listening tests in the past, but comparing what with what I am not sure.

Here's my take on some of the murkiness:

One of the main areas of confusion is that of phase versus pure delay. These are not the same thing. The stepped drivers you mention are giving time alignment between drivers, compensating for actual physical delays. but this does not mean that the speaker has an overall neutral phase response. The individual drivers have their own characteristic phase response, and in passive speakers are being driven via crossover filters which are also phase-shifting networks (not delay networks). The idea is to get the drivers to cross over with the same phase shift or, I think (having never tried to do this myself), cross over such that the resulting acoustic sum has the correct amplitude even if the individual drivers are not behaving ideally. The overall phase shift that occurs is usually not considered an issue. The resulting system is extremely complex, and what comes out in the time domain looks nothing like what went in. The 'get out of jail free card' is that someone at some time has decreed that humans are not able to hear phase shifts.

There is also a common argument that "The room completely swamps any phase shifts in the speaker", but this is based on a flawed assumption: that delays are the same as phase shifts. In the room, the ear is hearing the direct sound, plus delayed (not phase-shifted) sound. Transients are not affected by delays in the same way as they are by phase shifts, and the ear seems to be able to hear past the delays to the direct sound pretty effortlessly.

None of this can be proved with mathematics (which often impress and of course always 'prove' something, but not necessarily the right thing).

All I know is that with the advent of DSP it is possible to take a punt, bypass the entire question of how audible this stuff is in theory, and actually try it. If we are half-expecting reviews of DSP active systems (e.g. Kii Three) to conclude that these systems are clinical, harsh, flat, grey, unmusical etc. then it is a pleasant surprise to find that they don't. In the words of Michael Fremer:

...what I heard from this system absolutely astonished me. There was nothing 'digital' about the presentation. Nothing. The top end was about as perfectly rendered as I've heard from all of these tracks as was the detail resolution. Response was top to bottom full-bodied... it was fast, ultra-clean and spectacularly transparent.

Instrumental textures and timbres were as accurate and realistically portrayed as I've ever heard them. The amount of true detail revealed in very familiar records was unprecedented...

...I went back to be sure and I'm sure. I played the rip of Richard Thompson's "The Angels Took My Racehorse Away" from Henry The Human Fly his first solo LP (UK Island) recorded by John Wood at Sound Techniques and I've never before heard the drum sound so realistic and life-like and I mean never. But that was just the start of what I heard from the many tracks I played. The rip of a lacquer of "Night in Tunisia" from the Blue Note album of the same title was downright astonishing. "Can't You Hear My Knocking" just about made me faint. I'll stop now.
http://www.analogplanet.com/content...system-you-will-ever-hear#OKojxhmWyEFzfi28.97
 

Groucho

New Member
Aug 18, 2012
680
3
0
UK
Ironic, that adding an A/D-DSP-D/A segment in the signal chain can improve the overall system sound.

Do you mean the Fremer/Kyron Audio quote?

Well an A/D-DSP-D/A segment is added but a number of very primitive filters that decouple the amps from the drivers are removed.
 

Don Hills

Well-Known Member
Jun 20, 2013
366
1
323
Wellington, New Zealand
Do you mean the Fremer/Kyron Audio quote?

No. Just an observation, contrary to the beliefs of a segment of audiophiles who won't have a bar of anything digital.

Well we add an A/D-DSP-D/A segment but remove a number of very primitive filters that decouple the amps from the drivers.

True. But they were analogue filters, not digital, and therefore must sound better. :p
 

Groucho

New Member
Aug 18, 2012
680
3
0
UK
To be very specific to this thread, if we are removing the passive crossovers then we are not really able to tell what the contribution of the DSP phase correction is, in isolation. However, DEQX can be used to correct the phase of a passive speaker, and the new Devialet SAM system does this, certainly with the bass. People seem to be impressed with the results - not rigorous listening tests, but they can, at least, flick a switch to hear the system with correction and without.

A DEQX review where a passive speaker is being corrected using DSP:

Selecting the speaker-only correction with the B&Ws made an undeniable improvement in top-to-near-bottom integration in ways I would not have expected. Bass was not much changed, but the midrange was noticeably more open from the upper bass up. Treble seemed a bit softer but, most important, it was more difficult than ever to tell that there were four separate drivers in the box. Especially for dense and complex music, such as a revelatory new recording of George Antheil's Ballet Mécanique by Gil Rose and the Boston Modern Orchestra Project (SACD/CD, BMOP 1033). The PreMate clarified and illuminated details in the stereo soundstage that were murky without it...

http://www.stereophile.com/content/...rdigital-equalizer-page-2#Hh4gEfprTmqxqEkR.97

Going active would take this significant improvement to another level beyond that.
 

dallasjustice

Member Sponsor
Apr 12, 2011
2,067
8
0
Dallas, Texas
If one can improve the speaker's time domain, the frequency domain usually also improves. DSP is great tool with very few limitations. One of those limitations is the extent to which one can acquire an accurate speaker measurement. This can be done quasi-anechoic but the resolution is limited for the home DIY'er.

For me, any manufacturer who sells a digital active speaker, must be able to demonstrate how they linearize drivers and correct time discontinuity. IOW, a real anechoic environment is needed to do higher resolution DSP linearization. If that's not done during the design phase, then the home DIY'er is in an equally good position to do it themselves.
 

Nyal Mellor

Industry Expert
Jul 14, 2010
590
4
330
SF Bay Area, CA, USA
I don't have much in the way of facts and data. I believe there have been some studies done into the audibility of group delay at midrange frequencies, and some more at low frequencies, but I'm not sure any of it is conclusive.

FWIW I've been a DEQX dealer for maybe 5 years. What I've found is the better the speaker the less the speaker correction improves things. But the improvement is still always there, if the measurements are done correctly. The improvement with the best speakers is typically small, but we found that it even improved the sound of YG Anat IIIs which AFAIK are $70k or so. I also don't hear much improvement with my ATCs, which are much lower cost. For example the sound of the SCM19v2 does not fundamentally improve by turning on the speaker correction. There is a minor firming up of imaging, maybe a little more smoothness, that is it. With poorly measuring speakers the differences can be profound, like having a different speaker entirely.

I can think of a few ways speakers are not "zero" phase:
1) driver time delay/alignment differences
2) driver magnitude and phase differences
3) crossover related magnitude and phase changes
 

dallasjustice

Member Sponsor
Apr 12, 2011
2,067
8
0
Dallas, Texas
Wouldn't you agree that one of the limitations to linearizing drivers with DSP at home is the limited resolution? In my room we've linearized down to 1khz. I've also done the same with Acourate. TBH, it's superfluous to linearize if using a good full range target at those frequencies.

I think it's exciting when a digital active speaker manufacturer can linearize down to much lower frequencies because they can measure in real large anechoic environments and get higher resolution impulses.
I don't have much in the way of facts and data. I believe there have been some studies done into the audibility of group delay at midrange frequencies, and some more at low frequencies, but I'm not sure any of it is conclusive.

FWIW I've been a DEQX dealer for maybe 5 years. What I've found is the better the speaker the less the speaker correction improves things. But the improvement is still always there, if the measurements are done correctly. The improvement with the best speakers is typically small, but we found that it even improved the sound of YG Anat IIIs which AFAIK are $70k or so. I also don't hear much improvement with my ATCs, which are much lower cost. For example the sound of the SCM19v2 does not fundamentally improve by turning on the speaker correction. There is a minor firming up of imaging, maybe a little more smoothness, that is it. With poorly measuring speakers the differences can be profound, like having a different speaker entirely.

I can think of a few ways speakers are not "zero" phase:
1) driver time delay/alignment differences
2) driver magnitude and phase differences
3) crossover related magnitude and phase changes
 

Groucho

New Member
Aug 18, 2012
680
3
0
UK
Wouldn't you agree that one of the limitations to linearizing drivers with DSP at home is the limited resolution? In my room we've linearized down to 1khz. I've also done the same with Acourate. TBH, it's superfluous to linearize if using a good full range target at those frequencies.

I think it's exciting when a digital active speaker manufacturer can linearize down to much lower frequencies because they can measure in real large anechoic environments and get higher resolution impulses.

With my system I am happy to smooth the phase correction fairly heavily - it takes care of the most significant errors. I don't think there would be much point in correcting the 'fur' because the next day the temperature and humidity would be different and it would have shifted. Or at a different replay level the driver characteristics would change slightly. I am not chasing unachievable theoretical perfection, and I would bet that I wouldn't hear the difference anyway.

In a way, I am not too concerned about the A/B audibility of phase correction, as such. I just think that the default position should be approximately flat phase response rather than what we have now, which suggests that aiming for flat phase is some sort of weird esoteric perversion. I am happy, then, that this will ultimately sound 'best' and leave the ears and brain the least work to do in attempting to piece together what they're hearing.
 

Yuri Korzunov

Member
Jul 30, 2015
138
0
16
I made experiment several years ago. I created software module that:

1. Generate several harmonics simultaneously.
2. Allow change phases (relative delay) of each of the harmonics in real time.

Changing phase knobs position change sound. It is not on ear sensitivity border.

More right experiment compare pulses with different phases between harmonics.
 

About us

  • What’s Best Forum is THE forum for high end audio, product reviews, advice and sharing experiences on the best of everything else. This is THE place where audiophiles and audio companies discuss vintage, contemporary and new audio products, music servers, music streamers, computer audio, digital-to-analog converters, turntables, phono stages, cartridges, reel-to-reel tape machines, speakers, headphones and tube and solid-state amplification. Founded in 2010 What’s Best Forum invites intelligent and courteous people of all interests and backgrounds to describe and discuss the best of everything. From beginners to life-long hobbyists to industry professionals, we enjoy learning about new things and meeting new people, and participating in spirited debates.

Quick Navigation

User Menu

Steve Williams
Site Founder | Site Owner | Administrator
Ron Resnick
Site Co-Owner | Administrator
Julian (The Fixer)
Website Build | Marketing Managersing