Acoustic Weirdness..

[Glenn Kuras]

"Finally, as I've said numerous times before - and I realize you agree (I'm just reiterating for anyone joining us late ) - active correction is no substitute for a well-treated room. It is simply an additional tool that designers and home theater enthusiasts can employ to address problems with the loudspeaker-room response."

I am glad you reiterated. I was almost thinking you have changed our stands on treatment.
Great thread BTW.
Jeff would you mind posting the key things from
> I would suggest references [63]-[65] and [70]-[71] from the Toole paper discussed earlier in this thread. < this? This is something I have always wanted to try to learn more about.
Jeff in no way am I trying to bait or question you, just want to learn more about it and how it works.

Glenn

 

[Savant]

I just checked my library and I have none of those. Gee, what a surprise. So do me a favor and tell me the punch line. Do they show the reduction in ringing at multiple locations in an EQ'd room? Can you post a graph or three? Not that I don't trust you or anything...

I will try not to infer anything from your use of the phrase "punch line" - I am hardly joking about anything.

That aside, the papers in question show and/or discuss the changes in time reponse for the experiments they conducted. I would love to post the contents of the papers in question, but that would be an illegal use of copyrighted material. If I were to weigh the amount of time and effort it would take me to get permission from the AES to use images from five different papers against what it costs you to purchase them ($25 if you're and AES member), I would vote for the latter. I realize that may seem harsh, but I hope it makes sense.

We've gone down this road before and I am hesitant to dwell on it further. Gene and I (and others here, I'm sure) are well-versed in this matter. I spent several hours myself (re)reviewing the five papers above just to be sure they contain information I thought would be helpful to you. From our perspective (and I hope I'm not typing out of turn, Gene! ), the work has been done, the conclusions have been drawn, and our opinions are expressed based on our knowledge of this work and - in Gene's case - his observations from his own implementation of the technology. If you would like to come up to speed on the same and continue the discussion, that would be swell. If not, that would be swell, too!!!

 

[Savant]

Jeff would you mind posting the key things from
> I would suggest references [63]-[65] and [70]-[71] from the Toole paper discussed earlier in this thread. < this? This is something I have always wanted to try to learn more about.
Jeff in no way am I trying to bait or question you, just want to learn more about it and how it works.

I hope I don't sound like a broken record, but if anyone here - Ethan, Glenn, others - wants to learn more about the subject of active control, I would highly recommend the papers above, as well as some others.

The gist of them is what Gene and I - mostly Gene - have been writing about in this thread, and in others. Perhaps Gene can help with some links to some other threads?

The key thing (singular) is that active control can act as a form of damping for certain low frequency room problems. Will it work for everyone? Doubtful. But neither will any one solution when it comes to room acoustics.

 

[Savant]

Actually, the standard 1/12 octave frequencies fall almost perfectly between each musical note. I have charts of both here if you'd like me to post some screen caps.

No, that won't be necessary. I had a look back at this and you're right. That this does in fact occur is rather frustrating - the 1/12th center frequency thing; not your being right.

I referred to the ANSI standard and calculated where the musical notes fall in relation to the edge frequencies. For my own peace of mind, I also calculated whether the notes would fall within the frequency error range for 1/12th octave band analyzers, which they most certainly do not. In other words, sounding a single note will likely result in exciting two bands on most 1/12th octave band analyzers due to the allowable error. Which really kind of stinks. (Now that I think about it, I think there was a Yahoo! thread about this several years ago. If I find it, I'll refer others...)

Anyway, seeing as it was a completely misguided statement on my part, I deleted it above - fyi.

Thanks for pointing this out.

 

[Ethan Winer]

Jeff,

> I will try not to infer anything from your use of the phrase "punch line" - I am hardly joking about anything. <

No need to infer at all. If you've learned anything about me by now, you should know that I never take myself too seriously, and I always keep a light-hearted attitude about everything no matter how serious the subject. All I meant was to ask if they show the practical results I have asked for repeatedly. Not a theoretical discussion of how the technology works and what might be possible. So here it is again, a simple Yes/No question:

Do those papers present hard data showing a single EQ correction reducing ringing by a meaningful amount at not just one location but several?

> If I were to weigh the amount of time and effort it would take me to get permission from the AES to use images <

I'd never ask you for that. But if there were a single graph showing what I'm asking, I can't imagine anyone would mind. Heck, that would then cause me and probably several others here to go buy the paper(s). I can afford a few hundred bucks. But when this has come up in the past, and I've pressed the point, the answer was always that the papers explain the theory but don't actually show real-world measurements at multiple locations proving ringing is reduced globally.

> not your being right. <

ROF,L.

Actually, the much more important thing I was right about, which you did not acknowledge, is that 1/12 octave analysis misses the true extent of nulls.

--Ethan

 

[Savant]

Click HERE to see the graphs.

It's clear that the nulls are much worse than a 1/12 octave display indicates, and the total span from peak to adjacent null, which is probably the main issue, is also much worse than 1/12 octave shows. Look especially at the peak/null pair at 155 and 163 Hz. Those two frequencies are also less than 1/12 octave apart.

Playing devil's advocate a little...

Using this example of two different resolutions, I'm trying to imagine how I would interpret them. First, take the problem at 112 Hz. Regardless of the resolution used, I would likely conclude that there's a problem at 112 Hz. Either graph would tell me I should figure out the source of the problem - modal or otherwise - at 112 Hz and fix it. Whether the null is 13 dB low or 24 dB low or something else, it's still a problem. So I'd be OK with the 1/12th octave resolution in this case.

Next, the problem at 163 Hz. This is much more interesting; 25 dB down versus 6 dB down. If I were to only see the 1/12th octave response, I would probably conclude that there is no problem here. In fact, I might not even pay attention to 163 Hz to begin with. However, if I were only to see the lower graph with the better resolution, it might initially be cause for concern. The source of this problem is most likely either a boundary interference, or a mode. If it is a boundary interference, it will only be present at the one point where this measurement was taken. Additional measurements taken at other points might show dips at other frequencies, in which case I can perform other calcs to figure out which boundary needs some attention. If, on the other hand, the problem is modal, the bandwidth is likely in the range of ±2-3 Hz. From the lower graph, I would determine that excitation of this mode may not result in an audible dip on the order of 25 dB. I can make this assumption due to the (relative) narrowness of the dip at 163 Hz on the lower graph. Thus, I would probably still focus my efforts on other problems. (This is not uncommon - problems above about 100-150 Hz generally require less attention than lower ones provided good broadband room treatments are being implemented.)

So, going back to the upper graph. What if I only saw the upper graph? IMO, it would probably be a more accurate representation of my observations of the room response. I believe the problem at 112 Hz would be more audible than the problem at 163 Hz. My 1/12th octave response would show me this relative difference - even though I might not even be paying attention to 163 Hz - and I would development room treatments accordingly.

Additionally, if 163 Hz is a boundary interference, the null would become more pronounced elsewhere, even on a 1/12th octave analyzer. Or not - it could just as likely be narrow enough to not pop out as a significant problem anywhere.

This is just thinking out loud. Again - more devil's advocate than anything else. I have used 1/12th octave analyzers a lot for low frequency assessments. They tend to work out very well. And I think your example lends itself to that judgment. But it is a judgment call. If it turned out that the problem at 163 Hz was very obvious, but not showing up on any measurements, that would probably lead me in a totally different direction - i.e., perform a more detailed measurement a la the lower graph.

I guess my main point is that there is more to a room than an analyzer and a microphone. Just because we see a null doesn't mean we should immediately jump all over it as "major problem." More information is needed and the observations one makes whilst analyzing carry much more weight than any graph.

 

[krabapple]

The "optimum room ratios" produce inconsistent results; several papers in the last 10-20 years have addressed their over(mis)use. Many designers are abandoning them in favor of the better tools that are now available.

Also, FYI, the best placement for four subs, according to the paper Dr. Toole cites, is one sub at the midpoint of each wall:

Note that Toole et al, in all their subwoofer work, also quite rightly says this applies only for a closed rectangular room. In other words, a model system...a good place to start analytically, but for the great many people whose listening rooms do not fit that description, it's not necessarily the best advice...one might say all bets are off, if the room deviates really radically (e.g., one 'wall' totally open to another room, a very common situation)

He's also apparently saying his new reflection findings apply when using the sorts of speakers his (and Sean Olive's) previous work indicates 'sound best' to listener -- good off axis performance and good directivity. NRC stuff.
I know Axiom designs their speakers according to NRC/Toole findings, I've been trying to find out which of the JBL line actually does too. Anyone know?

I'm going to get Toole's paper this weekend. I always learn from his papers.

 

[krabapple]

"Finally, as I've said numerous times before - and I realize you agree (I'm just reiterating for anyone joining us late ) - active correction is no substitute for a well-treated room. It is simply an additional tool that designers and home theater enthusiasts can employ to address problems with the loudspeaker-room response."

I am glad you reiterated. I was almost thinking you have changed our stands on treatment.
Great thread BTW.
Jeff would you mind posting the key things from
> I would suggest references [63]-[65] and [70]-[71] from the Toole paper discussed earlier in this thread. < this? This is something I have always wanted to try to learn more about.
Jeff in no way am I trying to bait or question you, just want to learn more about it and how it works.

Glenn

I'm a little bit shocked that neither of two makers of room treatments here seems to be aware of this literature on control of room acoustics.

 

[krabapple]

Playing devil's advocate a little...

Using this example of two different resolutions, I'm trying to imagine how I would interpret them. First, take the problem at 112 Hz. Regardless of the resolution used, I would likely conclude that there's a problem at 112 Hz. Either graph would tell me I should figure out the source of the problem - modal or otherwise - at 112 Hz and fix it. Whether the null is 13 dB low or 24 dB low or something else, it's still a problem. So I'd be OK with the 1/12th octave resolution in this case.

But something seems curious to me. Suppose you stuck with 1/12 octave resolution, and applied 'correction' to the 112 Hz null to make the region 'flat' (let's assume for now that there is a way to do this without causing other problems). Then re-measure it using the finer-resolution method. Would you not still find a null of perhaps as much as 12 dB at 112Hz, given the disparity between the measurements of the pre-correction response?


.

 

[Savant]

All I meant was to ask if they show the practical results I have asked for repeatedly. Not a theoretical discussion of how the technology works and what might be possible.

IMO, the papers in question start by addressing the theoretical - as any good technical paper will - and then attempt to apply the theory to practical applications, albeit in controlled settings. Such is the nature of research. The results, discussions, and conclusions presented show what is possible.

So here it is again, a simple Yes/No question:

Do those papers present hard data showing a single EQ correction reducing ringing by a meaningful amount at not just one location but several?

IMO, they do. But as with any technical paper, results, discussion, and conclusions are subject to interpretation and usually lead to further discourse. I do not believe they establish any absolutes. But that is generally the case with acoustics - a subjective science.

> If I were to weigh the amount of time and effort it would take me to get permission from the AES to use images <

I'd never ask you for that. But if there were a single graph showing what I'm asking, I can't imagine anyone would mind. Heck, that would then cause me and probably several others here to go buy the paper(s). I can afford a few hundred bucks. But when this has come up in the past, and I've pressed the point, the answer was always that the papers explain the theory but don't actually show real-world measurements at multiple locations proving ringing is reduced globally.

Upon reflection (no pun intended), I believe you would probably be disappointed since the "multiple locations" aspect is likely not addressed to your satisfaction. Most of the papers - if not all of them; I'd have to go back and check - use the technique of spatial averaging when attempting to verify their work. This makes sense since they are ultimately looking to address room problems as a whole and not point-in-space problems. Correct me if I'm misinterpreting, but you seem to have an aversion to this approach. Thus, you probably won't find what you're looking for. The techniques and results are perfectly satisfactory to me. But I have a different perspective, I guess...

Actually, the much more important thing I was right about, which you did not acknowledge, is that 1/12 octave analysis misses the true extent of nulls.

Yes, I did. See above. You posted while I was typing.

 

[Savant]

But something seems curious to me. Suppose you stuck with 1/12 octave resolution, and applied 'correction' to the 112 Hz null to make the region 'flat' (let's assume for now that there is a way to do this without causing other problems). Then re-measure it using the finer-resolution method. Would you not still find a null of perhaps as much as 12 dB at 112Hz, given the disparity between the measurements of the pre-correction response?

Here are three (of more/many?) possibilities:

• The correction would be significant enough to make an acceptable audible difference. While the dip might still be significant on the constant-bandwidth measurement, the 1/12th might correlate better to the improvement heard.

• The response does look "flatter" in the 1/12th octave graph, but the problem, while improved, is still audibly present and less than ideal. As mentioned above, this might lead to additional measurements to better pinpoint the problem.

• If there is no perceptible change, but the response does appear "flatter," I would not only want to perform additional measurements (probably using constant bandwidth filter settings)—I would also want to figure out whether the problem is actually a culmination of two or more problems. It's possible that what was determined to be the driving problem - and may certainly be the driving problem according to the measurements - was not actually the problem being heard. This is more rare, but I've had it happen to me before!

Great thoughts, krab!!!

 

[Glenn Kuras]

I'm a little bit shocked that neither of two makers of room treatments here seems to be aware of this literature on control of room acoustics.

Trust me I am well aware of this, but want to learn more about it. IMO it sounds more like theory then practical use, but I may be wrong. To date I have never seen a water fall plot where ringing has been fixed by any thing else but acoustic treatment. Even if I am right I still would like to learn about the subject as much as possible.
BTW I have a ton of respect for Jeff and Gene, so these are only questions.

Glenn

 

[Savant]

> If I were to weigh the amount of time and effort it would take me to get permission from the AES to use images <

I'd never ask you for that. But if there were a single graph showing what I'm asking, I can't imagine anyone would mind.

Upon even further reflection, it should be pointed out that it's not a question of whether anyone would "mind." It's simply illegal.

Heck, that would then cause me and probably several others here to go buy the paper(s). I can afford a few hundred bucks.

And if the AES were a for-profit organization, I'm sure they'd agree. But they're not, so it's really moot whether an illegal use of their material would generate revenue for them.

Sorry for my bluntness. I'm not trying to be a poop. It's just that this is quite a sensitive issue these days.

 

[mtrycrafts]

I'm going to get Toole's paper this weekend. I always learn from his papers.

Do you have a proper citation for it? Searching AES papers by Toole shows very few, and not that one.

 

[Savant]

Do you have a proper citation for it? Searching AES papers by Toole shows very few, and not that one.

It just came out this year:
Loudspeakers and Rooms for Sound Reproduction—A Scientific Review by Floyd E. Toole, J. Audio Eng. Soc., Vol. 54, No. 6, 2006 June, pp. 451-476

 

[gene]

There's no need for that sort of personal attack. I haven't once mentioned my company or my products in this thread, nor do I usually in other threads discussing the science of acoustics. This is your forum so you can do what you want, but from my perspective that was an undeserved low blow having nothing to do with the subject at hand.

You deserve to be made aware of your obvious biases and agenda to promote the only solution you know which you incidentally sell on your website. The problem is you promote a lot of misinformation which puts me in a predicament on whether or not allow you to spread misinformation on our forums while we are trying to educate our readers on legitimate methods of achieving good bass response. The issue is I don’t have the time or desire to continually debate you. This reminds me of the issues we had years ago with Jon Risch on exotic cables though he eventually went away. You seem to take his persistence to a new level.

That's better than third octave, but still is low enough resolution to hide the true extent of the peaks and nulls. Especially the nulls.

I measure with much higher precision but then apply 1/12th octave smoothing since it better represents how we hear bass frequencies.

I understand that very well. But no room is a single pole filter! This is why I ask again and again for proof that ringing can be improved in practice. I'm still waiting. As soon as someone shows evidence that EQ can improve ringing for an area large enough to include both ears at the same time, and not substantially worsen the response or ringing elsewhere, I will change my opinion in a heartbeat.

Bass frequencies don’t vary much at all from one ear to the other and our ears aren’t sensitive enough to pick up that variation. Our ears are most sensitive to the wavelength of the distance between them which is in the midrange frequencies. Check out Robinson and Dadison curves which well establish this point.

There is tons of evidence supporting active equalization benefits for subwoofers. If you reduce the modal energy at its source, it will improve frequency and transient response at the listening area. Dr. Toole has proven this and I believe he sent you data on it. THX has proven this. Anthoni Grimani has proven this. WE have proven this in our articles such as:

http://www.audioholics.com/techtips/setup/loudspeakers/SubwooferplacementP1.php

So please prove me wrong!

Proving you wrong is about as useless as proving to the Black Knight in Monty Python that he was defeated. Its merely a flesh wound.

 

[Vaughan Odendaa]

Forgive me if this hasn't been asked, but where can I download this PDF document of Dr Tooles ?

Thanks.

--Sincerely,

 

[Ethan Winer]

Jeff,

> Playing devil's advocate a little... <

That's a wonderful analysis, thanks. But my only point was in the context of "Look how flat my room is." I simply wanted to show that the peaks and nulls are far more severe than a 1/12 octave display would imply.

> The results, discussions, and conclusions presented show what is possible. <

As we both know, almost anything is possible. Using DSP that's sufficiently sophisticated you can even counter ringing and obvious echoes at higher frequencies. But as we both also (should) know, the more correction you apply, the smaller the corrected region becomes. And the more you make things worse elsewhere too.

> Most of the papers ... use the technique of spatial averaging when attempting to verify their work. This makes sense since they are ultimately looking to address room problems as a whole and not point-in-space problems. <

This does not make sense because what each listener hears is the sound at their own seat. As I said earlier, the average of a 30 dB boost and a 30 dB null is perfectly flat. This is why the only way to prove that EQ reduces ringing in practice is to show a number of single-point measurements. Only then is it proven that all seats in the house truly benefit. Anything less is rationalizing and "cooking the books" so to speak.

--Ethan

[edit to fix a typo.]

 

[Ethan Winer]

I'm a little bit shocked that neither of two makers of room treatments here seems to be aware of this literature on control of room acoustics.

Ouch!

I'm quite aware of all this stuff. As I've mentioned in the past, one of my best friends, Bill Eppler, is a "genius engineer" who codes this exact type of DSP in assembly language. We have discussed DSP room correction in depth many times, and he's explained to me what's possible and what's not and why. I'm quite sure I understand it sufficiently. If not the math, at least the concepts which is what matters.

Suppose you stuck with 1/12 octave resolution, and applied 'correction' to the 112 Hz null to make the region 'flat' (let's assume for now that there is a way to do this without causing other problems). Then re-measure it using the finer-resolution method. Would you not still find a null of perhaps as much as 12 dB at 112Hz, given the disparity between the measurements of the pre-correction response?

Exactly. 1/12 octave is simply not adequate if you want to know what's really going on. Though as Jeff said, what's audible can be different from what's measured.

--Ethan

 

[Ethan Winer]

Gene,

> The problem is you promote a lot of misinformation <

A lot of misinformation? Please name them all. Or do you mean my single objection to the claim that EQ can reduce ringing?

It has been proposed that EQ can reduce ringing. In my experience it cannot. To back up my claim I have posted graphs showing the RESULTS of a professional consultant attempting to do exactly this and failing. So I don't see why it's unreasonable for me to ask for proof that EQ can reduce ringing in practice, by a meaningful amount, and over a useful physical area. For all the research and theoretical technical papers out there, you'd think somebody would have performed the simple tests I am asking to see!

> This reminds me of the issues we had years ago with Jon Risch on exotic cables <

Are you really suggesting that bass traps and acoustic treatment are in the same snake oil category as exotic cables?

> I measure with much higher precision but then apply 1/12th octave smoothing since it better represents how we hear bass frequencies. <

Could you post the original high-resolution data? I don't see why "how we hear" bass should be glossed over with averaging. Look again at the example graph I posted earlier showing the disparity between 1/12 octave and ETF's high resolution display. If a null that aligns with the frequency of a bass note really is 25 dB deep, then that's exactly how far down it will sound when that note is played.

> Bass frequencies don’t vary much at all from one ear to the other <

Wanna bet? Ears are about six apart. See this pair of graphs measured four inches apart:

Click HERE for the response at the listening position.
Click HERE for the response four inches to the right.

Look especially at the region between the 56 Hz and 92 Hz markers. At the listening position the span from peak to null is 29 dB. Four inches to the right the span is only 18 dB. That's a difference of 11 dB over a distance of only four inches. And yes, I was surprised too, so I measured both places again just to be sure it wasn't a glitch. It was not a glitch.

> Check out Robinson and Dadison curves which well establish this point. <

This is why I'm such a strong proponent of empirical evidence rather than relying on theory alone. Since I have proven (above) beyond all doubt that the low frequency response can change substantially over very small spans, it's clear the theory needs to be modified.

> WE have proven this in our articles such as <

I didn't see anything there showing EQ reducing ringing at all let alone over a span of several seats. If I missed that please point me to the specific page and figure.

Thanks.

--Ethan