• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

DRC - Digital Room Correction - Time Alignment? Is this a bleeding edge technology solution looking for a problem (and profit for its evangelists)?

OK1

Member
Joined
Apr 1, 2019
Messages
18
Likes
6
Over the last week, I've rekindled an interest in DRC, purely as a further exploration, to see what the current state of play is.

Making an effort to modify the frequency response, i.e amplitude of frequencies, to meet a target curve, with a threshold, I have no issues with this aspect of speaker/room correction. There is pretty much consensus, that this is a good objective. And every single process or software or device that aims at some kind of room or speaker correction, of some kind, includes this effort, to correct the amplitude of frequencies. Some in software, some in the speakers, some in external devices like DEQX, or from MiniDSP, or via the custom Linux server in solutions like Trinnov.

Listening to this video - see link or video embed below, and reading through comments across related threads on this and other online forums, however, I feel a certain discomfort, with the alchemy of emphasis on time alignment, not because there is not a case for this, but some of the exponents as in the video who make a whole talk about this time alignment, without addressing an elephant in the room.

Assuming that time alignment is critical, and important to the end result, it begs the question, which time alignment is really important? Cos if we wish to align time of arrival, we have several places to deal with this, not one.

1. Aligning the time of arrival between the high frequencies and the low frequencies, from all the drivers in each speaker., in a scenario where you have more than one driver, with each driver taking care of a different frequency band. Some speakers already attempt to achieve this using their own internal method, such as physically aligning the center of the tweeter driver to the centre of the woofer driver, with the tweeter set further back into the cabinet, for example. But of course depending on the orientation of these drivers, unless you have a concentric or co-axial speaker like the Genelec 8531, as you move your head, especially in the nearfield, the possibility that you will hear one driver's output arrive ahead of another is increased, and this of course also alters the amplitude of the contribution of one driver over another, changing the tone of the overall sound, especially the nearer you are to the speaker. Other options would be to use a single driver. like the Avantone Mixcubes.

So the real solution to allow you reduce the difference in arrival time between different drivers, would be careful positioning, and angling, and more likely sitting at a minimum distance from the speaker to reduce this differential in time arrival. Assuming we have two drivers, on one speaker and draw a triangle between the two drivers and the listening position, the greater the distance from the speakers, the higher the probability that (for a human head which is not actually static when sitting upright to listen to speakers), the angles of the triangle at the speaker drivers, will be more equal, and this lead to an equivalent arrival time at the listener.

My point here being, the greatest differential in the time arrival of audio from a tweeter and a woofer to a listener, is influenced by the position of the listener, more than any other, and the best solution to solving this important differential is not an electronic one, or digital one, but simple clever placement of the speaker, and listener, and angling the speaker to achieve as much as possible an almost identical time arrival at the listeners ear.

So either you go concentric/coaxial to make this pretty much a non issue, or you use sensible placement. Unless of course, the time arrival of bass and treble frequencies in the same speaker are so far apart due to delays caused by digital or analog crossovers, which in theory make the woofer to appear to be coming from a location which is 10's of feet behind the tweeter i.e the bass is delayed by 10's of milliseconds i.e. approx. 1 foot per millisecond - as per the speed of sound in air.

But it then begs the question, would not any decent speaker manufacturer have sorted this out, in their speaker design, if this delay would affect the time arrival so significantly, between the woofer and the tweeter. i.e. the solution being also - buy a decent, well designed speaker, that by design reduces this difference in arrival time.

2. Next we have the need to ensure that there is not too much variance in the arrival time of audio at the listening position, from one speaker to another.

In a simple stereo setup, the ideal solution for this is not an electronic or digital one, but a simple physical one - sit as much as possible, in the center at an equal distance between the speakers. Yet it will never be exactly equidistant cos our heads are not static and rather than aim to stick our heads in one perfect spot, the better solution is to use speakers with a good dispersion, so that we have more latitude to move around from left to right and still hear each speaker well enough. i,e. buy speakers with this desired horizontal wide-ish enough directivity.

And if our subwoofer is brought into the picture, then time align, by delaying either the main speakers or the subs, depending on which is nearer to the listening position. But this does not need any esoteric DRC, a simple measure of the distances with a measuring tape, should give us enough info to dial in the delay close enough. Kind of similar what happens in PA systems, which have a subwoofer tucked away somewhere in the hall. Not much voodoo needed.

More to the point, with the exception of possibly Trinnov, and I think Sonarworks, have not heard much about other tools, addressing time alignment between the different speakers, at least not in their main marketing info - if it was that important a feature, it should have been front and center in their publicity info. Furthermore, without a special 3D microphone like Trinnov uses, or the triangulation that Sonarworks by sending the same audio to both speakers, which for example is not measured in REW, how does one using a DRC technology approach, fix any timing differences in arrival between speakers?

Point of all this is - there seems to be so much hot debate and mention of time alignment, when some of these DRC solutions that are well touted, have no solution to detect or fix variations in timing, thus their exponents, in my view focus on the minutiae, and not much evidence of the import of their "time alignment efforts", when the simple things, like placement, can be achieved with a inexpensive tape measure, a string, or a laser based distance measuring device, not 300 dollars worth of complex software, and 500 dollars of consulting to hand hold the end user, to use the software properly.

Glad for all those who are making a nice tidy income from aiming and claiming to solve "timing problems" via confusing, difficult to use DRC software, when basic diligence with speaker placement would have achieved a whole lot more, and there is far more value in buying speakers that have been designed properly in the 1st place. Rather than trying to fix "time alignment" problems that should not have existed in the 1st place due to poor speaker design or wrong speaker placement and positioning.

Now that frequency/amplitude correction is pretty de rigueur, today, i.e. nothing new here, is there an emergence of "solutions" and consultants, claiming to add value through esoteric "time alignment", aiming to use technology to solve timing problems that should not have needed esoteric analysis and solutions, in the 1st place. Just good speaker design, and common sense needed.

What makes this whole issue even worse, is that no matter how well we time align drivers within a speaker cabinet, or arrival time from different cabinets, to a single spot in space, no human head is static, so all this herculean accuracy, will be overridden. minute by minute, by small head movements, anyway (unless of course you have used excellent design and consumer intelligence to buy a coaxial speaker, then all this time alignment effort is a total waste of time - pun intended).

And even with a coaxial set of speakers, due to head movements, the coincidence from one speaker to another in time arrival at the listening position, will never be identical. i.e. there is a threshold beyond which all further efforts to time align, are simply chasing ghosts, that add no value, but its ok to make claims and offer services for delivering improvements that are at best hard to prove, in a real world where heads and ears are not static, and these minute timing improvements are possibly irrelevant.

 
Last edited:

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
2,246
Likes
1,622
Location
Canada
Over the last week, I've rekindled an interest in DRC, purely as a further exploration, to see what the current state of play is.

Making an effort to modify the frequency response, i.e amplitude of frequencies, to meet a target curve, with a threshold, I have no issues with this aspect of speaker/room correction. There is pretty much consensus, that this is a good objective. And every single process or software or device that aims at some kind of room or speaker correction, of some kind, includes this effort, to correct the amplitude of frequencies. Some in software, some in the speakers, some in external devices like DEQX, or from MiniDSP, or via the custom Linux server in solutions like Trinnov.

Listening to this video - see link or video embed below, and reading through comments across related threads on this and other online forums, however, I feel a certain discomfort, with the alchemy of emphasis on time alignment, not because there is not a case for this, but some of the exponents as in the video who make a whole talk about this time alignment, without addressing an elephant in the room.

Assuming that time alignment is critical, and important to the end result, it begs the question, which time alignment is really important? Cos if we wish to align time of arrival, we have several places to deal with this, not one.

1. Aligning the time of arrival between the high frequencies and the low frequencies, from all the drivers in each speaker., in a scenario where you have more than one driver, with each driver taking care of a different frequency band. Some speakers already attempt to achieve this using their own internal method, such as physically aligning the center of the tweeter driver to the centre of the woofer driver, with the tweeter set further back into the cabinet, for example. But of course depending on the orientation of these drivers, unless you have a concentric or co-axial speaker like the Genelec 8531, as you move your head, especially in the nearfield, the possibility that you will hear one driver's output arrive ahead of another is increased, and this of course also alters the amplitude of the contribution of one driver over another, changing the tone of the overall sound, especially the nearer you are to the speaker. Other options would be to use a single driver. like the Avantone Mixcubes.

So the real solution to allow you reduce the difference in arrival time between different drivers, would be careful positioning, and angling, and more likely sitting at a minimum distance from the speaker to reduce this differential in time arrival. Assuming we have two drivers, on one speaker and draw a triangle between the two drivers and the listening position, the greater the distance from the speakers, the higher the probability that (for a human head which is not actually static when sitting upright to listen to speakers), the angles of the triangle at the speaker drivers, will be more equal, and this lead to an equivalent arrival time at the listener.

My point here being, the greatest differential in the time arrival of audio from a tweeter and a woofer to a listener, is influenced by the position of the listener, more than any other, and the best solution to solving this important differential is not an electronic one, or digital one, but simple clever placement of the speaker, and listener, and angling the speaker to achieve as much as possible an almost identical time arrival at the listeners ear.

So either you go concentric/coaxial to make this pretty much a non issue, or you use sensible placement. Unless of course, the time arrival of bass and treble frequencies in the same speaker are so far apart due to delays caused by digital or analog crossovers, which in theory make the woofer to appear to be coming from a location which is 10's of feet behind the tweeter i.e the bass is delayed by 10's of milliseconds i.e. approx. 1 foot per millisecond - as per the speed of sound in air.

But it then begs the question, would not any decent speaker manufacturer have sorted this out, in their speaker design, if this delay would affect the time arrival so significantly, between the woofer and the tweeter. i.e. the solution being also - buy a decent, well designed speaker, that by design reduces this difference in arrival time.

2. Next we have the need to ensure that there is not too much variance in the arrival time of audio at the listening position, from one speaker to another.

In a simple stereo setup, the ideal solution for this is not an electronic or digital one, but a simple physical one - sit as much as possible, in the center at an equal distance between the speakers. Yet it will never be exactly equidistant cos our heads are not static and rather than aim to stick our heads in one perfect spot, the better solution is to use speakers with a good dispersion, so that we have more latitude to move around from left to right and still hear each speaker well enough. i,e. buy speakers with this desired horizontal wide-ish enough directivity.

And if our subwoofer is brought into the picture, then time align, by delaying either the main speakers or the subs, depending on which is nearer to the listening position. But this does not need any esoteric DRC, a simple measure of the distances with a measuring tape, should give us enough info to dial in the delay close enough. Kind of similar what happens in PA systems, which have a subwoofer tucked away somewhere in the hall. Not much voodoo needed.

More to the point, with the exception of possibly Trinnov, and I think Sonarworks, have not heard much about other tools, addressing time alignment between the different speakers, at least not in their main marketing info - if it was that important a feature, it should have been front and center in their publicity info. Furthermore, without a special 3D microphone like Trinnov uses, or the triangulation that Sonarworks by sending the same audio to both speakers, which for example is not measured in REW, how does one using a DRC technology approach, fix any timing differences in arrival between speakers?

Point of all this is - there seems to be so much hot debate and mention of time alignment, when some of these DRC solutions that are well touted, have no solution to detect or fix variations in timing, thus their exponents, in my view focus on the minutiae, and not much evidence of the import of their "time alignment efforts", when the simple things, like placement, can be achieved with a inexpensive tape measure, a string, or a laser based distance measuring device, not 300 dollars worth of complex software, and 500 dollars of consulting to hand hold the end user, to use the software properly.

Glad for all those who are making a nice tidy income from aiming and claiming to solve "timing problems" via confusing, difficult to use DRC software, when basic diligence with speaker placement would have achieved a whole lot more, and there is far more value in buying speakers that have been designed properly in the 1st place. Rather than trying to fix "time alignment" problems that should not have existed in the 1st place due to poor speaker design or wrong speaker placement and positioning.

Now that frequency/amplitude correction is pretty de rigueur, today, i.e. nothing new here, is there an emergence of "solutions" and consultants, claiming to add value through esoteric "time alignment", aiming to use technology to solve timing problems that should not have needed esoteric analysis and solutions, in the 1st place. Just good speaker design, and common sense needed.

What makes this whole issue even worse, is that no matter how well we time align drivers within a speaker cabinet, or arrival time from different cabinets, to a single spot in space, no human head is static, so all this herculean accuracy, will be overridden. minute by minute, by small head movements, anyway (unless of course you have used excellent design and consumer intelligence to buy a coaxial speaker, then all this time alignment effort is a total waste of time - pun intended).

And even with a coaxial set of speakers, due to head movements, the coincidence from one speaker to another in time arrival at the listening position, will never be identical. i.e. there is a threshold beyond which all further efforts to time align, are simply chasing ghosts, that add no value, but its ok to make claims and offer services for delivering improvements that are at best hard to prove, in a real world where heads and ears are not static, and these minute timing improvements are possibly irrelevant.




Time alignment is important to think about and consider if you are designing a multi-way speaker system. Does it mean that if one is setting up an audio reproduction system at home that now everything has to be perfectly “zero phase” as it arrives at your ears? Else you would definitely be able to notice that something is wrong… well, no, not necessarily… most humans seem able to adjust through all kinds of audio anomalies/deficiencies just fine.

But, practically speaking, the time component part does still affect the overall magnitude frequency response when summing sound from multiple sources the lower in frequency you measure — more importantly, if it is coming from multiple speaker and sub units playing simultaneously together at a close enough SPL. If you have set your goal to have as accurate a representation of the original signal as possible (at least in the main listening position) with as little deviation as possible for the broadest of circumstances, you might as well try to get the timing part as close as you can get to reasonably “zero” or at least “similar enough” in the time component aspect. I mean, if you are able to do an adequate job through simple positioning and minimum phase filters, why not?

Because some people mix and match different speakers in a system — e.g. different non-matching active DSP (with differing fixed internal processing delays) or passive speakers and subs — your simple tape measuring technique may prove impossible to get right without actual confirmation using real acoustic measurements with a properly set up timing reference.

I do think that one should at least endeavour to get the phase profiles of the speakers below the transition zone in a multichannel system employing varying different speaker designs/models — from full-range mains to smaller bass-limited satellites — to cohere more rather than less. Above 500 to 700 Hz it doesn’t bother me as long as the main front channels are the same speakers or are similar enough.

Something more obvious one can easily notice and experiment for fun would be gradually adding delay to one speaker in a perfectly set up stereo pair via DSP — i.e. adding just a simple time delay. More advanced, I suppose, would be to artificially induce a crossover phase rotation with increasing steepness, say, via a simple all pass filter.

If you simultaneously alter both channels rather than just a single one the likelihood of noticing something different (even with a relatively steep all pass filter) is probably close to nil for the vast majority.



Now, is this really all worth the cost of an expensive new Trinnov processor? And hiring an experienced professional/consultant to set things up for me? Eh, personally, I rather do things my way because even if I get things wrong, well, at least I will be learning from it. Or at least that’s the hope.

If I can do a good enough job playing around with this whole sound system “correction” or optimization manually by myself (based on my own informed or maybe only “half-informed”standards) then might as well. *Of course, I know not everyone is in the same situation and mental predisposition as me.
 
OP
O

OK1

Member
Joined
Apr 1, 2019
Messages
18
Likes
6
Time alignment is important to think about and consider if you are designing a multi-way speaker system. Does it mean that if one is setting up an audio reproduction system at home that now everything has to be perfectly “zero phase” as it arrives at your ears? Else you would definitely be able to notice that something is wrong… well, no, not necessarily… most humans seem able to adjust through all kinds of audio anomalies/deficiencies just fine.

But, practically speaking, the time component part does still affect the overall magnitude frequency response when summing sound from multiple sources the lower in frequency you measure — more importantly, if it is coming from multiple speaker and sub units playing simultaneously together at a close enough SPL. If you have set your goal to have as accurate a representation of the original signal as possible (at least in the main listening position) with as little deviation as possible for the broadest of circumstances, you might as well try to get the timing part as close as you can get to reasonably “zero” or at least “similar enough” in the time component aspect. I mean, if you are able to do an adequate job through simple positioning and minimum phase filters, why not?

Because some people mix and match different speakers in a system — e.g. different non-matching active DSP (with differing fixed internal processing delays) or passive speakers and subs — your simple tape measuring technique may prove impossible to get right without actual confirmation using real acoustic measurements with a properly set up timing reference.

I do think that one should at least endeavour to get the phase profiles of the speakers below the transition zone in a multichannel system employing varying different speaker designs/models — from full-range mains to smaller bass-limited satellites — to cohere more rather than less. Above 500 to 700 Hz it doesn’t bother me as long as the main front channels are the same speakers or are similar enough.

Something more obvious one can easily notice and experiment for fun would be gradually adding delay to one speaker in a perfectly set up stereo pair via DSP — i.e. adding just a simple time delay. More advanced, I suppose, would be to artificially induce a crossover phase rotation with increasing steepness, say, via a simple all pass filter.

If you simultaneously alter both channels rather than just a single one the likelihood of noticing something different (even with a relatively steep all pass filter) is probably close to nil for the vast majority.



Now, is this really all worth the cost of an expensive new Trinnov processor? And hiring an experienced professional/consultant to set things up for me? Eh, personally, I rather do things my way because even if I get things wrong, well, at least I will be learning from it. Or at least that’s the hope.

If I can do a good enough job playing around with this whole sound system “correction” or optimization manually by myself (based on my own informed or maybe only “half-informed”standards) then might as well. *Of course, I know not everyone is in the same situation and mental predisposition as me.
Thanks.

A lot of the value of forums like audiosciencereview and associated youtube channels, and many other fact and logic based meeting places on the web, is simply the value of making us think about what we are doing, hearing and investing, and asking the questions why?

If there is a reason why we need to spend 10,000 dollars on a pair of speakers, to achieve a certain quality of listening, at least we should know why. Example being the Kii Three or the Genelec 8351's, I'm more on the studio speaker side of things, which is where I also do most of my listening, in a bedroom studio, in near/mid field distance from the speakers.

The need for reasonableness, has not stopped the hi-fidelity vendors from touting claims about the benefits of certain esoteric cables to audio, even where this forum has debunked the myths around some of these claims, or all kind of other claims with little or no relevant evidence.

I was reading a review on 6 moons earlier today, this one here, a review of a DRC solution :


And it included the following conclusion "Switching to bypass to compare the familiar sound but now using the PreMate’s DAC showed further skills. The DAC excelled at presenting a detailed, powerful, full-bodied and dynamic sound with the kind of sonic energy that will instil a sense of life and enthusiastic pace into any system."

As the Americans would say - waffle.

And while they do make an effort to be logical with a modicum of rational data presented, it has been a relief to read some of the recent reviews (over the last year or so) of speakers, on ASR, which use the "Spinorama" measurements to explain what the reviewer hears., devoid of the kind of amorphous lingo one has become accustomed to elsewhere.

This was what got my back up, when I saw the video on the other thread on ASR (https://www.audiosciencereview.com/...e-of-the-art-of-digital-room-correction.27591), which is now closed so I could not air my response there, hence the need to start this thread. Well, we cannot expect everything on ASR to have the same objectivity as Amir's reviews., and it would be wonderful on ASR, to bring some reasonableness back to the subject of DRC, i.e. what's true, what's really the key benefits that can be achieved, and how is this done. Which tools work well, or not.

And to also be clear about what aspects of the various claims by DRC exponents, including by those who are now deemed to be experts in the subject, albeit without any peer verified endorsement, remain in the realm of possibilities and options that we should not be investing too much concern about.

From a logical standpoint, not being an expert in acoustics or audio engineering, one thing I would want the exponents of time alignment to tell me would be - how many milliseconds of delay, is introduced in the differential arrival of different frequencies from a single speaker, such as a two way speaker that is properly designed. What's the maximum value of this delay of certain frequencies, due to changes in phase by analog or digital crossovers, in milliseconds, then I can make an informed decision, for myself, is it worth my while chasing down this anomaly.

I am more familiar with the pro-audio world, and have never owned a single piece of gear that could be called hi-fi, and in that world, 1st thing you look at are the specs, they do not tell you everything, but at least they give you an idea of where to ask further questions and redirect your acquisition interest to.

Come to think of it, when I listened to that video, much of it was the same waffle, no specifics, how do we measure the improvement that DRC introduces. e.g, in practice is there an improvement in the frequency evenness between the high and low frequencies, at the listening position, and if so what is the quantity of this improvement.

If a "science" of DRC, that is based on measurement, at the starting point of its workflow, is not able to express its benefit in numbers, that we can interpret, then we clearly have issues.

I remember when once upon a time professional studio monitors (speakers typically active ones) would be advertised as +/- so much dB variation from X to Y frequencies, as measured in an anechoic chamber or on axis, with the tweeter, etc, etc, e.g +/-3dB between 50 and 18Khz, and this would mean something to us even if we acknowledge that each manufacturer may not have identical methods for arriving at their speaker quality claims. At least we got some empirical statements as a starting point.

This way when I read that the KH 80 DSP speaker has specs described here with these words "Putting some numbers on it, the published frequency response charts are flat within ±0.6dB across the entire working range, and any two speakers are matched within 0.25dB of each other — which is just astonishing!" I know what to expect. Reference for this claim is here


But with DRC, we get nothing to quantify the improvement that this "scientific" process introduces. If we can measure the flaws, and quantify them, should we also not be able to express the improvement with some numbers, rather than the subjective waffle that seems to be the norm.

I would not bother to vent on any other forum, but hopefully this is audiosciencereview, where demands for empirical proof and quantifiable justification of claims, in this esoteric aspect of time alignment, should find a safe haven of open expression., and hopefully we can all get to the bottom of the alchemy, at least those who really want to know what is true, and also what is irrelevant or bogus, about the various time alignment opportunities, which are claimed to improve audio accuracy in our listening.

Where is the threshold beyond which, we need not bother.

I give an example, for many years I was a supporter of the high sample rate recording mantra. But over time, it has dawned on me from reading the opinions of others, and Dan Worrall in particular - who has excellent video debunks of professional audio topics on Youtube under his channel or published in the Fabfilter channel, that high sample rate recording has its own issues, which have to be understood. Many years ago via a simple easy test using an analyzer - Voxengo SPAN, it was easily demonstrable that higher sample rates introduce more noise (negligibly so, but this is true). So its up to me to decide, where the diminishing returns are no longer acceptable, for me, and I can choose the sample rate that is optimal, for me based on knowledge. Now I record at 48k, which should be ample, if I have decent converters.( a condition which is true for almost all audio interfaces including the budget ones). I have three converters now, each with Analog to Digital conversion between 113dBA and 120dBA dynamic range. Yes there are now converters with over 125dBA dynamic range, but proper understanding of my own needs where I rarely record more than 10 tracks from external audio sources, informs me that this is NOT a reason to upgrade. Yes it may measure better, but it does not make any difference to keep chasing esoteric measurements in newer converters, at least not because of their improved dynamic range, so I know where to stop investing, and focus energies and money on other improvements.

DRC needs a similar expose, sanity check, so we know how far each of us wishes to go, and can decide where to stop throwing further effort at imperceptible or non measurable improvements.
 
Last edited:

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
2,246
Likes
1,622
Location
Canada
Where is the threshold beyond which, we need not bother.

DRC needs a similar expose, sanity check, so we know how far each of us wishes to go, and can decide where to stop throwing further effort at imperceptible or non measurable improvements.

If you are altering the original signal digitally, I don't see why the changes will not be reflected in the measurements. If the change is significant enough, then it should be visible even in a messy room measurement.

There are several confounding factors that cloud any post-analysis when it comes to the positive or negative effects of any kind of "digital room correction". It would be better to isolate these effects individually and study the collected measurements in detail (and how it correlates with one's subjective perception) before making any broad conclusions that the particular DRC applied to a system is good/bad. In many cases, one might only be getting "variations" of some qualitatively mild to moderate perceptual differences or improvements from the baseline. Depending on source material, listening levels, and reflection-decay profile of the room, you may need to apply more or less adjustements to suit one's particular situation and personal taste better. However, if your baseline starting point is egregiously "non-flat" (see below examples), then the effect of any DRC may be both measurably and subjectively enormous -- this is often the case with subwoofers -- hence also possibly why naive audio enthusiast neophytes often avoid/shy away from subs -- because they don't know how to implement DRC effectively. Few have the experience and know-how in how to integrate them optimally from the start. Moreover, you have to "learn" the room itself... as not everyone's room situation and arrangement constraints are the same. Even if you have an anechoically flat speaker and 120dB SINAD DAC, the response is often drastically altered the moment you measure the transfer function (as it's also being convolved dynamically as you move around) inside a real and highly compromised, non-anechoic room environment.

sub couch.png sub desk.png
*partially corrected linear phase HPF (to attain zero lip-sync audible delay in a HTPC setup) combined with good-old regular minimum phase EQ.
**optimal EQ is not the same for both listening positions.


Even the above room measurements are somewhat "lucky" or favorable as I've already found out the best positions for the subwoofer and main listening seats in this dedicated listening room of mine. In a previous home, it was very difficult to get a decent subwoofer frequency response as my available options for positioning and single sub only setup were far more limited.
 
Last edited:

Hipper

Addicted to Fun and Learning
Joined
Jun 16, 2019
Messages
740
Likes
602
Location
Herts., England
Is that video really talking about time alignment as you describe it - drivers, speakers - or rather decay times of the sound after it enters your room?

I assume time alignment of drivers and speakers should be sorted by the speaker manufacturers and your positioning of them in the room.

DRC/DSP/EQ and above all room treatment (bass traps etc.) can help with decay times. Decay times have a big effect on how you hear music in your room. I found that as I smoothed the frequency response, mostly with room treatment but also with a bit of EQ, the decay times came down from around 400ms to 200ms and became smoother over the whole spectrum. This is very noticeable in what you hear in the form of more precise notes and generally greater clarity of the music. See the spectograms in my post #60:

https://www.audiosciencereview.com/...in-room-measurements.13540/page-3#post-411614
 

dominikz

Addicted to Fun and Learning
Forum Donor
Joined
Oct 10, 2020
Messages
529
Likes
1,834
I find this article from AudioXpress is a really nice overview of the topic.
Personally (after doing extensive testing of various DRCs) I'm still quite skeptical of time/phase correction importance, in otherwise well-designed audio systems.
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
2,246
Likes
1,622
Location
Canada
One more excellent resource article on Time and Phase Alignment written by Bob McCarthy:


There's a very good reason why professional sound designers/engineers do not just rely on the frequency magnitude graphs when setting up more complex combined loudspeaker systems.
 
  • Like
Reactions: OK1
OP
O

OK1

Member
Joined
Apr 1, 2019
Messages
18
Likes
6
Is that video really talking about time alignment as you describe it - drivers, speakers - or rather decay times of the sound after it enters your room?

I assume time alignment of drivers and speakers should be sorted by the speaker manufacturers and your positioning of them in the room.

DRC/DSP/EQ and above all room treatment (bass traps etc.) can help with decay times. Decay times have a big effect on how you hear music in your room. I found that as I smoothed the frequency response, mostly with room treatment but also with a bit of EQ, the decay times came down from around 400ms to 200ms and became smoother over the whole spectrum. This is very noticeable in what you hear in the form of more precise notes and generally greater clarity of the music. See the spectograms in my post #60:

https://www.audiosciencereview.com/...in-room-measurements.13540/page-3#post-411614
Unless I am totally mistaken, my recall, and anyone can independently verify, by listening to the video again, the DSP expert was talking about aligning the time arrival of audio, at the listening position.

Absolutely concur with you, I expect time alignment of drivers in the speaker should be something that should be taken care of by the speaker manufacturer. Ideally in the speaker itself, but I have learned of one case - the HEDD speakers (active) typically aimed for professional studios, which also provides a software plugin, to be used in a DAW(Digital Audio Workstation) such as Cubase, Pro Tools, Reaper or Logic, to achieve some kind of alignment, of frequencies.

Links to the description of this plugin and other time alignment details from HEDD below. Not saying they are the only ones, but just using them as an example where the manufacturer takes the pains to sort this out at the speaker, so we do not have to think of fixing such mis-alignments anywhere else. IN this case HEDD also addresses time alignment where subs are not equidistant to the listening position. Interesting. !

The point being, as an end user we are perfectly in our rights to decide where we wish to invest our money, but I think its a lot more contained if the speaker and the speaker manufacturer can take care of all alignments between the drivers in a speaker, such as HEDD has done, so we do not have to achieve this independently. They know their speakers best.



 
OP
O

OK1

Member
Joined
Apr 1, 2019
Messages
18
Likes
6
I find this article from AudioXpress is a really nice overview of the topic.
Personally (after doing extensive testing of various DRCs) I'm still quite skeptical of time/phase correction importance, in otherwise well-designed audio systems.
Thanks ever so much. It was an outstanding read. Just what I needed. Opinions by at least three of the leading lights in the field, Dr Floyd Toole, Dr. Wolfgang Klippel, and Andrew Jones, names which I am familiar with, whose contributions to audio reproduction, and measurement, are about as good as one can get.

And the conclusions(and things I deduce from the article, and other things I have learned) :

1. That as far as we know, there are a few special cases where phase alignment has any consistent audible benefit, and even in these cases, the primary benefit was the enhancement of amplitude, i.e. correction of amplitude losses due to phase cancellation.

2. Unless you are using concentric/coaxial speakers, the benefit of any phase alignment, occurs only on one listening axis., and could be a disbenefit on any other axis.

3. Most of the evidence of phase benefits is limited to listening on headphones or in an anechoic chamber, and as soon as you reproduce sound in a room, all manner of phase modifying interaction from reflections in the room. Trinnov is the only product I know of at this time that appears to do something to align and improve the room induced phase anomalies - if I understand the article properly.

4. From other information I have read/studied, especially from the work of Dr Toole, it is of advantage to have a speaker that has even dispersion in the horizontal and vertical axis, so that reflections from the room surfaces are also uniform (subject of course to variations in absorption at different frequencies). It definitely helps to start of with good directivity at source. Two schools of thought here some prefer narrow directivity, some prefer wider directivity, Pick your poison. Speakers with uniform directivity, and excellent flatter frequency response under anechoic conditions are an excellent starting point, in audio propagation in a room.

5. My own further actions will be to focus on these things :

a. Choose good speakers, that measure well. When I can afford this, get one that is concentric like a Genelec 8531. I accept that its better to invest in a good speaker than attempting to turn a lousy speaker into a prince, using DSP.

b. DSP will only add the icing on the cake to an already decent speaker. Focus on amplitude correction of frequencies, and only bother with phase alignment (or zero phase) as an interesting "experiment".

c. Treat the room acoustically sympathetically(do not go overboard, attempting to recreate an anechoic chamber), do the main reflection points, ceiling baffle if I can afford that, some furniture, some carpeting, soffit where possible, some bass trapping if possible.

d. Small rooms will always have worse issues, so aim for the largest room I can use, to delay the arrival of reflections, and keep them distinct, avoiding any smearing they may have on the direct sound. Either soffit the speakers - in wall to avoid reflections from the front wall, or keep the speakers well away from walls - to keep the reflections delayed as much as possible from the direct sound.

e. In my case, I'm producing music also so will focus on nearfield placement, to reduce the ratio of reflected to direct sound.

f. Speakers are far from perfect, our best measured speakers(in anechoic measurements), are still presented with deviations of +/- XdB(usually in the 2 or 3 dB range and still considered good enough), in manufacturers documentation, with 1/3 octave smoothing, which is whole lot of approximation. Either the measurements are unable to be accurate without smoothing, or the speakers themselves cannot be presented as accurate without the caveat of smoothing. Our ears must be really forgiving, of such inaccuracies, or drivers/transducers are really not that accurate, no matter how good our current knowledge and manufacturing and material science is. i.e. compared to when we measure amplifiers and converters, with absolutely no smoothing whatsoever, and these can now be really flat frequency across the spectrum, with imperceptible distortion levels, speakers are still far from the ideal, no matter how good the are. Therefore will be realistic in my ambitions. If I want to hear pin sharp, as best as any transducer can be - just get a pair of headphones, which by implication cancels out any room issues, and apply cross feed to this. Even that will not be perfect, cos we are not yet at the point where transducers are accurate. Definitely will dial down expectations, cos no DSP or technology available today, will reproduce anything near as accurate as the sound of real audio, it will for the foreseeable future remain an approximation, due to limitations in the technology of speakers. Speakers are getting better e.g Kii Three which measures as accurate as +/- 0.5 dB. but this also used DSP to achieve this.

Microphones already introduce their own phase anomalies, so much of this phase alignment mantra, is somewhat futile, cos our ear is already used to listening to all recorded audio, with phase inaccuracies, right from the 1st audio recordings ever made, and that has not changed.

The issue about the accumulation of phase changes at low frequencies from multiple high pass filters in the chain was most revealing.

Highly appreciate your contribution, to helping me understand where to focus my energies and investments. Thank you.
 
Last edited:

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
3,385
Likes
1,959
Over the last week, I've rekindled an interest in DRC, purely as a further exploration, to see what the current state of play is.

Making an effort to modify the frequency response, i.e amplitude of frequencies, to meet a target curve, with a threshold, I have no issues with this aspect of speaker/room correction. There is pretty much consensus, that this is a good objective. And every single process or software or device that aims at some kind of room or speaker correction, of some kind, includes this effort, to correct the amplitude of frequencies. Some in software, some in the speakers, some in external devices like DEQX, or from MiniDSP, or via the custom Linux server in solutions like Trinnov.

Listening to this video - see link or video embed below, and reading through comments across related threads on this and other online forums, however, I feel a certain discomfort, with the alchemy of emphasis on time alignment, not because there is not a case for this, but some of the exponents as in the video who make a whole talk about this time alignment, without addressing an elephant in the room.

Assuming that time alignment is critical, and important to the end result, it begs the question, which time alignment is really important? Cos if we wish to align time of arrival, we have several places to deal with this, not one.

1. Aligning the time of arrival between the high frequencies and the low frequencies, from all the drivers in each speaker., in a scenario where you have more than one driver, with each driver taking care of a different frequency band. Some speakers already attempt to achieve this using their own internal method, such as physically aligning the center of the tweeter driver to the centre of the woofer driver, with the tweeter set further back into the cabinet, for example. But of course depending on the orientation of these drivers, unless you have a concentric or co-axial speaker like the Genelec 8531, as you move your head, especially in the nearfield, the possibility that you will hear one driver's output arrive ahead of another is increased, and this of course also alters the amplitude of the contribution of one driver over another, changing the tone of the overall sound, especially the nearer you are to the speaker. Other options would be to use a single driver. like the Avantone Mixcubes.

So the real solution to allow you reduce the difference in arrival time between different drivers, would be careful positioning, and angling, and more likely sitting at a minimum distance from the speaker to reduce this differential in time arrival. Assuming we have two drivers, on one speaker and draw a triangle between the two drivers and the listening position, the greater the distance from the speakers, the higher the probability that (for a human head which is not actually static when sitting upright to listen to speakers), the angles of the triangle at the speaker drivers, will be more equal, and this lead to an equivalent arrival time at the listener.

My point here being, the greatest differential in the time arrival of audio from a tweeter and a woofer to a listener, is influenced by the position of the listener, more than any other, and the best solution to solving this important differential is not an electronic one, or digital one, but simple clever placement of the speaker, and listener, and angling the speaker to achieve as much as possible an almost identical time arrival at the listeners ear.

So either you go concentric/coaxial to make this pretty much a non issue, or you use sensible placement. Unless of course, the time arrival of bass and treble frequencies in the same speaker are so far apart due to delays caused by digital or analog crossovers, which in theory make the woofer to appear to be coming from a location which is 10's of feet behind the tweeter i.e the bass is delayed by 10's of milliseconds i.e. approx. 1 foot per millisecond - as per the speed of sound in air.

But it then begs the question, would not any decent speaker manufacturer have sorted this out, in their speaker design, if this delay would affect the time arrival so significantly, between the woofer and the tweeter. i.e. the solution being also - buy a decent, well designed speaker, that by design reduces this difference in arrival time.

2. Next we have the need to ensure that there is not too much variance in the arrival time of audio at the listening position, from one speaker to another.

In a simple stereo setup, the ideal solution for this is not an electronic or digital one, but a simple physical one - sit as much as possible, in the center at an equal distance between the speakers. Yet it will never be exactly equidistant cos our heads are not static and rather than aim to stick our heads in one perfect spot, the better solution is to use speakers with a good dispersion, so that we have more latitude to move around from left to right and still hear each speaker well enough. i,e. buy speakers with this desired horizontal wide-ish enough directivity.

And if our subwoofer is brought into the picture, then time align, by delaying either the main speakers or the subs, depending on which is nearer to the listening position. But this does not need any esoteric DRC, a simple measure of the distances with a measuring tape, should give us enough info to dial in the delay close enough. Kind of similar what happens in PA systems, which have a subwoofer tucked away somewhere in the hall. Not much voodoo needed.

More to the point, with the exception of possibly Trinnov, and I think Sonarworks, have not heard much about other tools, addressing time alignment between the different speakers, at least not in their main marketing info - if it was that important a feature, it should have been front and center in their publicity info. Furthermore, without a special 3D microphone like Trinnov uses, or the triangulation that Sonarworks by sending the same audio to both speakers, which for example is not measured in REW, how does one using a DRC technology approach, fix any timing differences in arrival between speakers?

Point of all this is - there seems to be so much hot debate and mention of time alignment, when some of these DRC solutions that are well touted, have no solution to detect or fix variations in timing, thus their exponents, in my view focus on the minutiae, and not much evidence of the import of their "time alignment efforts", when the simple things, like placement, can be achieved with a inexpensive tape measure, a string, or a laser based distance measuring device, not 300 dollars worth of complex software, and 500 dollars of consulting to hand hold the end user, to use the software properly.

Glad for all those who are making a nice tidy income from aiming and claiming to solve "timing problems" via confusing, difficult to use DRC software, when basic diligence with speaker placement would have achieved a whole lot more, and there is far more value in buying speakers that have been designed properly in the 1st place. Rather than trying to fix "time alignment" problems that should not have existed in the 1st place due to poor speaker design or wrong speaker placement and positioning.

Now that frequency/amplitude correction is pretty de rigueur, today, i.e. nothing new here, is there an emergence of "solutions" and consultants, claiming to add value through esoteric "time alignment", aiming to use technology to solve timing problems that should not have needed esoteric analysis and solutions, in the 1st place. Just good speaker design, and common sense needed.

What makes this whole issue even worse, is that no matter how well we time align drivers within a speaker cabinet, or arrival time from different cabinets, to a single spot in space, no human head is static, so all this herculean accuracy, will be overridden. minute by minute, by small head movements, anyway (unless of course you have used excellent design and consumer intelligence to buy a coaxial speaker, then all this time alignment effort is a total waste of time - pun intended).

And even with a coaxial set of speakers, due to head movements, the coincidence from one speaker to another in time arrival at the listening position, will never be identical. i.e. there is a threshold beyond which all further efforts to time align, are simply chasing ghosts, that add no value, but its ok to make claims and offer services for delivering improvements that are at best hard to prove, in a real world where heads and ears are not static, and these minute timing improvements are possibly irrelevant.


There's a very detailed discussion thread already on this topic:

 
  • Like
Reactions: OK1
OP
O

OK1

Member
Joined
Apr 1, 2019
Messages
18
Likes
6
There's a very detailed discussion thread already on this topic:

Thanks. I'll review the discussion. Appreciated.
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
1,975
Likes
1,374
Location
Sweden
Audibility of phase in traditional speakers are not there IMO as long as you don't mess around with phase shifting filters between 100-800 Hz.

 
Top Bottom