Interesting article by Mitchco.

[dallasjustice]

I'll try to put words in Mitch's mouth as well as my own. It's been my experience that time domain made closer to linear phase does improve the soundstage clarity. The spatial relationships between different instruments or vocals become better defined.

But how can soundstage clarity be studied? Are there any studies which have really investigated the above described phenomenon by comparing digitally corrected time domain versus normal passive speakers?

I have spoken to Kevin Voeks (Dr. Product development at Harman/Revel) why they don't they compensate for time differential. He pointed to Dr. Vanderkooy work and conversations with him and that showed the differential simply is not audible with music in listening rooms (it can be to some extent with headphones or in anechoic chamber). He said by making them time aligned they would have to use lower order crossovers which would create other problems that were definitely audible.

I have not spoken to Dr. Toole about it but here are some quotes from his book that says the same thing:

"This [results of blind listening tests] suggests that we like flat amplitude
spectra and we don’t like resonances, but we tolerate general phase shift, meaning
that waveform fidelity [both amplitude and phase] is not a requirement.

[...]
Loudspeaker transducers, woofers, midranges, and tweeters behave as
minimum-phase devices within their operating frequency ranges (i.e., the phase
response is calculable from the amplitude response). This means that if the
frequency response is smooth, so is the phase response, and as a result, the
impulse response is unblemished by ringing. When multiple transducers are
combined into a system, the correspondence between amplitude and phase is
modified in the crossover frequency ranges because the transducers are at different
points in space. There are propagation path-length differences to different
measuring/listening points. Delays are non-minimum-phase phenomena. In the
crossover regions, where multiple transducers are radiating, the outputs can
combine in many different ways depending on the orientation of the microphone
or listener to the loudspeaker.

The result is that if one chooses to design a loudspeaker system that
has linear phase, there will be only a very limited range of positions in space
over which it will apply. This constraint can be accommodated for the
direct sound from a loudspeaker, but even a single reflection destroys the relationship.

As has been seen throughout Part One of this book, in all circumstances,
from concert halls to sound reproduction in homes, listeners at best
like or at worst are not deterred by normal refl ections in small rooms. Therefore,
it seems that (1) because of reflections in the recording environment there
is little possibility of phase integrity in the recorded signal, (2) there are challenges
in designing loudspeakers that can deliver a signal with phase integrity
over a large angular range, and (3) there is no hope of it reaching a listener in
a normally reflective room. All is not lost, though, because two ears and a brain
seem not to care.

Many investigators over many years have attempted to determine whether
phase shift mattered to sound quality (e.g., Greenfield and Hawksford, 1990;
Hansen and Madsen, 1974a, 1974b; Lipshitz et al., 1982; Van Keulen, 1991).
In every case, it has been shown that if it is audible, it is a subtle effect,
most easily heard through headphones or in an anechoic chamber, using carefully
chosen or contrived signals. There is quite general agreement that with
music reproduced through loudspeakers in normally reflective rooms, phase
shift is substantially or completely inaudible. When it has been audible as a
difference, when it is switched in and out, it is not clear that listeners had a
preference.

Others looked at the audibility of group delay (Bilsen and Kievits, 1989; Deer
et al., 1985; Flanagan et al., 2005; Krauss, 1990) and found that the detection
threshold is in the range 1.6 to 2 ms, and more in reflective spaces.
Lipshitz et al. (1982) conclude, “All of the effects described can reasonably
be classified as subtle. We are not, in our present state of knowledge, advocating
that phase linear transducers are a requirement for high-quality sound reproduction.”
Greenfield and Hawksford (1990) observe that phase effects in rooms are
“very subtle effects indeed,” and seem mostly to be spatial rather than timbral.
As to whether phase corrections are needed, without a phase correct recording
process, any listener opinions are of personal preference, not the recognition of
“accurate” reproduction.

In the design of loudspeaker systems, knowing the phase behavior of transducers
is critical to the successful merging of acoustical outputs from multiple
drivers in the crossover regions. Beyond that, it appears to be unimportant.

So I say that the research is pretty conclusive and there is really no "problem" here to fix. It is the case of chasing measurements that please the eye, instead of the ears. The impulse measurements are devoid of room reflections and only use one microphone instead of two ears and a brain. Big difference in acoustics.

 

[Cosmik]

Listening tests in peer reviewed papers have all the details one would want. Not having read them is no excuse to say they may not be so!

That aside, it is easy to intuit why the conclusions are what they are. First, the test actually show that linear phase delays *are* audible in headphone listening where the room is eliminated. When we add the room, we by definition introduce linear phase delay. A high frequency with shorter wavelength will experience far more of a phase distortion than a mid frequency with longer wavelength. These phase "errors" exist by the same reason in a live session, in the recording/mastering room, and your own room. In other words, you have three sets of them at pretty high level. Speaker only contributes one component to an existing mix of unknown phase delay (which is different from track to track in your music!). That is one of the reasons the speaker created phase delay gets lost when music is playing in a room as opposed to headphones or anechoic chamber.

The other reason is what I explain here: http://audiosciencereview.com/forum/index.php?threads/perceptual-effects-of-room-reflections.13/

In a nutshell, you have two ears, not one. Instrumentation showing delays from speaker drivers represents one ear, not two. Your ears are at different distances to the speaker and hence experience different levels of phase shift. Your brain is constantly experiencing this in everyday life. Evolutionary traits have caused our brain to set aside this blurring of timing to use your word, and get to the meat of the matter which is what the source of sound is. Without this, you would constantly hear echos as the two ears hear signals at different timing, causing us to go mad!

As humans we are trained to think of things like timing in the form of precision/clock, etc. So it is natural to put huge importance on them. Our hearing and brain though are not clocks and don't work the way we think they do. So it is important to put aside our intuition and go by what the research says. And the research says achieving phase accuracy is not important and it is a futile goal anyway as nothing in your recording is phase accurate. If the talent didn't hear it in phase accurate manner, why would you insist that you do???

The issue, for me, is that the arguments above start from the justification of the limitations of archaic technology, rather than starting from the premise that a speaker should (as much as is possible) reproduce the recorded waveform. It is the assumption that we know how the brain works - and it is trivial. People assume that really, we can do what we like with the signal because a room just looks like mush to an FFT, so it must also look like mush to the brain. We can spend a lot of time and energy doing experiments to show how we think the brain works (it's just an FFT spectrum analyser isn't it?), but some of us would prefer to bypass that, and just fix the speakers anyway. We don't know how the brain works so it is folly to try to second guess it.

This means that people like me end up with the most objectively-correct systems, while the supposed audio scientists end up with passive speakers. That is irony!

 

[dallasjustice]

Delphine Devallez presented this paper to the AES in Paris last month. It seeks to investigate how time and phase differences effect soundstage in cars. The dummy "Sandy" made several pink noise binaural recordings and then various time domain metrics were analyzed and a hypothesis is presented. I don't view this study as concrete proof that improving the IACC, ITD or other metrics using DSP means listeners will statisticslly prefer improved time domain. But it seems to be a worthy investigation and a good read. (The lead author is is gorgeous too!)
http://www.aes.org/tmpFiles/elib/20160724/18234.pdf

 

[hvbias]

I'll try to put words in Mitch's mouth as well as my own. It's been my experience that time domain made closer to linear phase does improve the soundstage clarity. The spatial relationships between different instruments or vocals become better defined.

But how can soundstage clarity be studied? Are there any studies which have really investigated the above described phenomenon by comparing digitally corrected time domain versus normal passive speakers?

Can you correct time domain response in Acourate without using it for all the crossovers? This is what I gathered from Mitch's post in the paragraph that starts with "However, by disconnecting the passive XO...". It sounded like he needed to use Acourate for all his active crossovers in order to reach that near perfect time coherence.

Which seems contrary to what I have read from Uli, since Acourate is taking time and frequency response measurements.

Going back to excellent time domain response with passive crossovers, in my experience working around the problems with first order electrical crossovers causes more headaches than necessary. Maybe it works in some full range driver crossing to a tweeter, but those systems also have tons of compromises.

 

[dallasjustice]

I'm not sure what you are asking. You can read about Mitch's system here.
http://www.computeraudiophile.com/c...e-alignment-driver-linearization-walkthrough/

He time aligns his drivers, linearizes each driver and applies a final correction from seated position. Acourate does time domain correction anyway in his final correction from seated position. But it can't correct time domain as precisely as individual driver correction in a digitally active system.

The real question, IMO, is to what extent does various time domain correction improve soundstage clarity. IME it's a big improvement but there's no solid scientific investigation to address this phenomenon.

Can you correct time domain response in Acourate without using it for all the crossovers? This is what I gathered from Mitch's post in the paragraph that starts with "However, by disconnecting the passive XO...". It sounded like he needed to use Acourate for all his active crossovers in order to reach that near perfect time coherence.

Which seems contrary to what I have read from Uli, since Acourate is taking time and frequency response measurements.

Going back to excellent time domain response with passive crossovers, in my experience working around the problems with first order electrical crossovers causes more headaches than necessary. Maybe it works in some full range driver crossing to a tweeter, but those systems also have tons of compromises.

 

[hvbias]

He time aligns his drivers, linearizes each driver and applies a final correction from seated position. Acourate does time domain correction anyway in his final correction from seated position. But it can't correct time domain as precisely as individual driver correction in a digitally active system.

This clarifies it, thanks. My previous post should have been better worded as "I was wondering if you had to use all Acourate active crossovers in order to reach exceptional time domain measurements."

The real question, IMO, is to what extent does various time domain correction improve soundstage clarity.

Agreed.

 

[Purité Audio]

Active DSP crossovers in both Grimm LS1 and Kii THREEs, the Kiis in particular image incredibly , very much like the Beolabs.
Keith

 

[dallasjustice]

I think my opinion about time domain is expressed as my own.

I think the ear/brain is more sensitive to it at certain frequencies. When it comes to low frequencies, the ear is thankfully not very sensitive to it except where time misbehavior effects the frequency domain. This may be where my opinion differs from Mitch's. My JBL 4367 two way speakers have a passive crossover but I also use a digital FIR filter to crossver to a pair of mono subs. IME, there is no such thing as "perfect" subwoofer time alignment. When it comes to getting tight and smooth bass using subs, it's much more important to study how low frequencies interact with the room and with the different transducers. This is where I think the use of a step response is much less valuable. I hope that makes some sense.

This clarifies it, thanks. My previous post should have been better worded as "I was wondering if you had to use all Acourate active crossovers in order to reach exceptional time domain measurements."



Agreed.

 

[Thomas savage]

[QUOTbruno's té Audio, post: 20271, member: 13"]Active DSP crossovers in both Grimm LS1 and Kii THREEs, the Kiis in particular image incredibly , very much like the Beolabs.
Keith[/QUOTE]
Its bruno's amps in the grimm too?

 

[Purité Audio]

Bruno was one of the designers of the LS1's and NCore amps are used in both LS1 and Kii THREE.
Keith

 

[Thomas savage]

Man crush

 

[Phelonious Ponk]

The issue, for me, is that the arguments above start from the justification of the limitations of archaic technology, rather than starting from the premise that a speaker should (as much as is possible) reproduce the recorded waveform. It is the assumption that we know how the brain works - and it is trivial. People assume that really, we can do what we like with the signal because a room just looks like mush to an FFT, so it must also look like mush to the brain. We can spend a lot of time and energy doing experiments to show how we think the brain works (it's just an FFT spectrum analyser isn't it?), but some of us would prefer to bypass that, and just fix the speakers anyway. We don't know how the brain works so it is folly to try to second guess it.

This means that people like me end up with the most objectively-correct systems, while the supposed audio scientists end up with passive speakers. That is irony!

This sounds about right. Most of the discussion here and on audiophile forums, regarding perception, ASA, how the brain processes audio, etc., seems to want to use the mystery of perception as a reason to believe in things that can't otherwise be verified. But the bottom line is, if you want your brain to process something close to reality, and the only reality your reproduction system knows is the recording, you need to deliver to your ears the most accurate reproduction of that recording that you can. Everything else is just another variation of "I hear unicorn sparkles and they don't show up in testing because they aren't measurable with todays technology." They're mistaking imagination for perception.

Tim

 

[mitchco]

Thanks for discussing my article, very interesting. With respect to phase and time coherence, Nikhil says in one of the CA comments: “Time coherence is distinct from Phase coherence though. You can have a speaker that is phase coherent but not time coherent. A Time Coherent speaker will be both time and phase coherent.”

Nikhil is correct and this is the confusion for some – it confused me for some time as well. Ever played with rePhase? https://sourceforge.net/projects/rephase/ rePhase can independently adjust the phase without changing the amplitude. This is a key point to understand. Have a read and then look at example 1: https://www.minidsp.com/applications/advanced-tools/rephase-fir-tool Again, note one can adjust amplitude and phase independently of each other. I encourage folks to try it out for themselves.

When I tried this on my passive XO system, I could not hear a difference in several listening tests I performed. Why? Even though the phase response is smooth or flat, the fact of the matter is that the acoustic centers of each of the drivers, in a multiway system, still have not changed and therefore, not time aligned. This is what is meant by time coherence or time alignment. One way to achieve this is to physically align the multi-way speaker’s acoustic centers, similar to: http://www.surrountec.com/surrounTec-prinzip.html It is not totally technically correct, but close enough for rock and roll - Nice step response!

However, with the advent of modern DSP, one can time align the acoustic centers, to a degree of accuracy that was simply not possible ten years ago. At a 48 kHz sampling rate, one can adjust the acoustic centers of drivers in 1 sample increments or 0.3 of an inch. And before anyone leaps to the conclusion about time alignment is only good at one listening spot, in my eBook, I prove that wrong by showing 14 step response measurements of my speakers across a 6ft x 2ft listening area where my couch would be and the time alignment remains exactly the same throughout.

To my ears, I agree with most of the research regarding "phase" coherence, I can’t hear a difference. But make no mistake, most of the research is about phase coherence, not time coherence. And that’s my point. To my ears, when the drivers are time aligned, that is when I hear the difference. Before anyone quotes old scripture, make sure it is about "time" coherence and I would be especially interested in the test conditions to see if it is even a valid test scenario.

As far as what our ears care about, it is the first sound arrival timbre along with the first 20ms of sound arrival that contains location cues in the recording that our ears care about most, along with inter-channel frequency response, gain and delay. I feel JJ’s presentation is as good as it gets for understanding what our ears care about in small room acoustics: http://www.aes-media.org/sections/pnw/pnwrecaps/2008/jj_jan08/

I hope you are all enjoying the music!

 

[fas42]

This sounds about right. Most of the discussion here and on audiophile forums, regarding perception, ASA, how the brain processes audio, etc., seems to want to use the mystery of perception as a reason to believe in things that can't otherwise be verified. But the bottom line is, if you want your brain to process something close to reality, and the only reality your reproduction system knows is the recording, you need to deliver to your ears the most accurate reproduction of that recording that you can. Everything else is just another variation of "I hear unicorn sparkles and they don't show up in testing because they aren't measurable with todays technology." They're mistaking imagination for perception.

Tim

Tim, perception may be a mystery in some areas, but the research has got a pretty decent handle on it now. They've known for a long time that the visual sense is capable of adjusting what the eye receives, so that the "best" message is delivered - a classic are the upside down glasses, inverting the image as perceived; the brain digests it for a while, then "gets it", and pops the picture around to the right way. And the same mechanism operates for the auditory sensing - if enough of the right clues are in the sound mix as picked up by the ears, then a recreation of the recorded acoustic event occurs in the space.

You're correct that the most accurate reproduction is needed; the argument is really about what needs to be accurate enough ... my experiences and experiments have constantly reinforced that it's the distortion and loss of low level information that's the killer for getting the auditory sense to unscramble the acoustic information well enough; and not the things most people usually worry about.

You are very wrong about imagination vs. perception ... once a person perceives such a convincing portrayal he then "perceives" what's wrong with normal reproduction - and no amount of imagination is going to help with that ...

 

[Blumlein 88]

The upside down glasses experiment changed processing sense of the brain. It didn't alter the perceptual acuity of the eye.

 

[Blumlein 88]

Duplicate post.

 

[fas42]

The upside down glasses experiment changed processing sense of the brain. It didn't alter the perceptual acuity of the eye.

Agreed. The model that the brain used for processing the eye data was changed, on the fly, so that the visual information made sense - this what the extension to understanding auditory behaviour that was contributed by ASA is about: that the brain uses a model of what the aural data 'means', and uses a "best fit" comparing of what's coming in, to that model. If it's a good fit then you get convincing sound; if not up to scratch then you are constantly aware that you're listening to hifi reproduction - the switch in the brain is literally as marked as that "inverting of the visual image" - for me, at least ...

 

[RayDunzl]

Ever played with rePhase?

I have, in the context of passive speakers and measurement/correction at the listening position.

It's powerful, but manual. If I remember correctly 277 sliders available for setting phase, another 277 for amplitude, and numerous other little tools.

I now have AcourateDRC, which does phase, but, maybe to a lesser degree than is possible with rePhase. Or, maybe it does it right.

---

Now I'm pondering if there is some way to combine two separately generated FIR filters (text).

 

[amirm]

The issue, for me, is that the arguments above start from the justification of the limitations of archaic technology, rather than starting from the premise that a speaker should (as much as is possible) reproduce the recorded waveform. It is the assumption that we know how the brain works - and it is trivial. People assume that really, we can do what we like with the signal because a room just looks like mush to an FFT, so it must also look like mush to the brain. We can spend a lot of time and energy doing experiments to show how we think the brain works (it's just an FFT spectrum analyser isn't it?), but some of us would prefer to bypass that, and just fix the speakers anyway. We don't know how the brain works so it is folly to try to second guess it.

The ear is not an FFT and the brain certainly is not. No reference were made to such concepts in what I post so not sure why you are speaking of them.

We don't need to understand the brain here. What we need to understand is combination of two ears and the brain together. We run narrow, controlled tests and that teaches us what is detected and what is not. This type of decomposition can then be used to expand our understanding of how to build audio equipment. Where to put effort and priority, and where not. Experiments get repeated by different researchers and conclusions are then reached out of the collective wisdom and work of everyone involved.

If I whisper a mile away from you, you won't hear it. You can't dispute this with, "we don't know how the brain works." We very well know such limitation. It is the less obvious ones to lay people that are not understood and this forum is created to bring that understanding to everyone.

As to speakers, they are compromises. The key to building the best is to know what compromises are more OK than others. This type of research grounded in proper investigation gets us there. As I explained time "coherency" gets lost in certain crossover designs. The very same designs may have very desirable characteristics. If we could have both then sure, we should. But it simply is not necessary or a loss if it is not there as the OP article tries to make it so.

Controlled testing of speakers with human listeners point to measurements that highly correlate with the human preferences. Timing is not such a thing.

 

[RayDunzl]

http://www.computeraudiophile.com/c...ides-unique-perspective-vancouver-audio-show/

Keith

Ok, so you made me look...

It mentions:

Here's mine, DSP'd for the listening position:

Upper trace, the mathematically combobulated step response (using REW) extracted from a log sweep, and bottom, a microphone recording (Audacity) of the initial rise of a square wave (a short lived (100ms) "step").

Looking at the somewhat remarkable congruence in the appearance of the two, I have to go "Hmmm...", as I didn't expect to see such a close match between the two measurement methods.