AVS › AVS Forum › Audio › Receivers, Amps, and Processors › The Subjective and Objective Evaluation of Room Correction Products
New Posts  All Forums:Forum Nav:

The Subjective and Objective Evaluation of Room Correction Products - Page 2

post #31 of 582
11/3/09 at 4:47pm
I don't get the mono thing at all, sure the listener may be able to discern differences better that way but, and correct me if I'm wrong, aren't room EQ's like ARC & Audyssey designed to EQ multiple speakers and effectively determine the best use of filters and delays based on a summation of same? Don't they have a fuzzy logic that might allocate resources differently than if they truly did each speaker individually?
Reply
Reply

AVS Top Picks

post #32 of 582
11/3/09 at 4:49pm
Quote:
Originally Posted by sdurani View Post

The identities/order really isn't important, since the most popular may not end up being your personal favourite, any more than the most popular song on radio these days will be your current favourite too. Sure it's likely you'll prefer it, but not guaranteed. Better to listen for yourself (with your system, in your room) than rely on mass preference.

I totally disagree. music preference is subjective, you might like a song and I might not. This is a scientific study, and it's presented like "here is this incredible finding we have, some products make the sound better, some really mess it up. Sorry, but that's all we're going to tell you."

Sure, I'll go buy all 5 units (6 if I include the neptune) and run my own tests...so this article helps me how?

I totally understand the whole bad manners thing if not wanting to badmouth the competition, but since it's a scientific test with double blind listening panels, I think that's a totally different thing. It's like if consumer reports found the brakes are bad on a certain car, but said well, "We won't tell you which one, you will have to go buy them and try to find out for yourself. Happy driving, don't tailgate"

this is beyond manners, it's a scientific study, and anyone who participated in it ran the same risk.
Reply
Reply
post #33 of 582
11/3/09 at 4:58pm
Dr. Olive,

Thanks for the link.

I am under the impression that when listening at the same level as the recording environment and if the acoustics of the listening matches the recording environment, that a flat frequency response would generally be preferred.

However, rarely is it that the case. In your test do you know whether the listening tests were conducted at the same levels that the content was recorded?

If listening at lower levels in order to accommodate the decrease in the sensitivity of human hearing wouldn't it require a boost in the bass to maintain the same perception of loudness in that frequency region?

If the intent of your reference listening room was to recreate a typical home listening environment, then its fair to say that the listening room acoustic environment differed from the recording acoustic environment. In typical home listening environments it is usually necessary to roll-off the highs to tame brightness.

So, if listening at below the recording's mixing level a test listener would likely prefer a boost in the bass to make up for the decrease in hearing sensitivity, and also gentle roll-off in the treble to deal with acoustic differences between the mixing and simulated home listening environments. Perhaps this accounts for your test results of a tilted frequency response?

As you are probably aware, Audyssey has released an extension of their room correction software called Dynamic EQ. It measures the actual levels of content in real-time and automatically adjusts the equalization to compensate for the decrease in hearing sensitivity at lower levels. This feature would dynamically "tilt" the frequency response curve as needed according to the levels in each portion of the content.

Do you plan on introducing a similar feature in your Harman products, or will you be using a static tilted target curve?

Thanks.

Larry
Reply
Reply
post #34 of 582
11/3/09 at 5:03pm
Thread Starter 
Quote:
Originally Posted by apodaca View Post

Nice read. I myself would seem to 'be in agreement' with these results. Of all of the room corrections I have tried which were limited to Audessey Flat, Front as well as Yamaha YPAO Flat, Front and Natural - I much prefer the Natural setting which provides a slight rolloff of high frequencies.I also cant stand subwoofers that dont go down to 20hz which does make the sound thin in my opinion.

Having said all of this though I believe that the recordings used for evaluation should be disclosed and the study needs to show it was independent of these.

Hi Apodaca:

The recordings are disclosed in the AES preprint and the presentation PDF in slide 6:

Jennifer Warnes' "Bird on a Wire"
James Taylor "That's Why I'm Here"
Tracy Chapman "Fast Car"

I agree it would have been nice if we could have tested the room corrections with more than 3 programs but that wasn't practical. The recordings are all generally well-recorded by Grammy award winning recording engineers George Massenburg and Frank Filipetti, and contain good LF content, which was essential for these tests. We've been using these recordings for +15 years in loudspeaker tests, and they consistently reveal subjective differences in loudspeakers that correlate well with anechoic measurements. These recordings have consistently indicated that listeners prefer loudspeakers that have a flat, smooth on-axis response, extended bandwidth, and smooth off-axis response. To me, that is good confirmation that the recordings are relatively well controlled nuisance variables.

The statistical analysis (ANOVA) indicates that program had no significant effect on the results, nor did it interact with the room corrections. In order words, the differences among the room corrections were sufficiently large that program was not the determining factor.
Reply
Reply
post #35 of 582
11/3/09 at 5:22pm
Quote:
Originally Posted by Tonmeister2008 View Post

I just revealed which products were ours in the test.

So your products in your listening room came out best. Hmmm. I have an open mind and like the tests but this to me is a point that can't be ignored.

Do you think knowing how to best configure/setup your products vs the competitors could have skewed results? ie you got optimum performance from yours but maybe not from the competitors' products.
Reply
Reply
post #36 of 582
11/3/09 at 5:24pm
Quote:
Originally Posted by rnrgagne View Post

I don't get the mono thing at all, sure the listener may be able to discern differences better that way but, and correct me if I'm wrong, aren't room EQ's like ARC & Audyssey designed to EQ multiple speakers and effectively determine the best use of filters and delays based on a summation of same? Don't they have a fuzzy logic that might allocate resources differently than if they truly did each speaker individually?

Hi,

I agree that one of Audyssey's strong points is its ability to timbre match all satellite speakers.

However, to do that requires individually equalizing each speaker to the same target curve. The fuzzy logic part of the equation is used to develop and weigh more heavily equalization filters that consider common acoustic problems in different locations. Nevertheless, each speaker is still individually equalized.

Larry
Reply
Reply
post #37 of 582
11/3/09 at 5:45pm
Quote:
Originally Posted by Tonmeister2008 View Post

The recordings are disclosed in the AES preprint and the presentation PDF in slide 6:

Jennifer Warnes' "Bird on a Wire"
James Taylor "That's Why I'm Here"
Tracy Chapman "Fast Car"

I agree it would have been nice if we could have tested the room corrections with more than 3 programs but that wasn't practical. The recordings are all generally well-recorded by Grammy award winning recording engineers George Massenburg and Frank Filipetti, and contain good LF content, which was essential for these tests. We've been using these recordings for +15 years in loudspeaker tests, and they consistently reveal subjective differences in loudspeakers that correlate well with anechoic measurements. These recordings have consistently indicated that listeners prefer loudspeakers that have a flat, smooth on-axis response, extended bandwidth, and smooth off-axis response. To me, that is good confirmation that the recordings are relatively well controlled nuisance variables.

Well, this indicates an internal consistency but I am not so sure it permits more than a reliable consensus. I am pretty familiar with the first and third samples and have heard them with a number of systems and under a number of conditions but, for example, I cannot say how prominent the kick-drum should be the first. One can get a palpable KICK out of it under some conditions (one can make that happen). That might impress most of us but how much is enough or too much. Same for the lower bass on the other sample.

Of course, this opens the can of worms: How can we possibly know what it was supposed to sound like when we compare a sample with our internal concept of it should sound like. Frankly, I'd be hard put to judge those samples by anything except preference and, mebbe, that's enough. Mebbe we need to revive the old AR live-vs.recorded setup. Is there room behind the curtain for a small combo?
Reply
Reply
post #38 of 582
11/3/09 at 5:59pm
Thread Starter 
Quote:
Originally Posted by Anomad View Post

I totally disagree. music preference is subjective, you might like a song and I might not. This is a scientific study, and it's presented like "here is this incredible finding we have, some products make the sound better, some really mess it up. Sorry, but that's all we're going to tell you."

Sure, I'll go buy all 5 units (6 if I include the neptune) and run my own tests...so this article helps me how?

I totally understand the whole bad manners thing if not wanting to badmouth the competition, but since it's a scientific test with double blind listening panels, I think that's a totally different thing. It's like if consumer reports found the brakes are bad on a certain car, but said well, "We won't tell you which one, you will have to go buy them and try to find out for yourself. Happy driving, don't tailgate"

this is beyond manners, it's a scientific study, and anyone who participated in it ran the same risk.

The point of this test was not to determine which product is the best. The point was to do a controlled listening test on different room corrections (including our own) to determine how different they are, in what way are they different, and whether the subjective results correlate with objective measurements. I've shown some experimental data based on listening tests that room correction when executed properly can improve the sound significantly. Prior to that, we mostly had to trust the marketing and sales literature that the products did what they were supposed to.

To my knowledge, such a test has never been reported before. It's a small step, but a first one. I'm hoping that more researchers and room correction manufacturers take up the challenge, and start doing similar tests like these and publishing the results.
Reply
Reply
post #39 of 582
11/3/09 at 6:35pm
"I totally understand the whole bad manners thing if not wanting to badmouth the competition, but since it's a scientific test with double blind listening panels, I think that's a totally different thing."

That would be the case if the test wasn't performed by one of the competitors.

"As you are probably aware, Audyssey has released an extension of their room correction software called Dynamic EQ...Do you plan on introducing a similar feature in your Harman products, or will you be using a static tilted target curve?"

Doesn't Dolby volume, which the HK 7500 receiver has, do that?

"The point was to do a controlled listening test on different room corrections (including our own)"

I sur would have liked to see how EZSet/EQ did.
Reply
Reply
post #40 of 582
11/3/09 at 6:54pm
Quote:
Originally Posted by Tonmeister2008 View Post

Here is what I can tell you: the Harman room correction algorithms tested in this paper (RC1 and RC2) are the same ones available in the new JBL Synthesis HATS/SDEC systems. The HATS/SDEC system will also be available for Revel loudspeaker systems with room correction available up to the point where the room modal/boundary effects are no longer an issue (400-500 Hz). Beyond that frequency, you are equalizing the loudspeaker, which is not necessary with these loudspeakers.

The same room correction technology will find its way into other Harman products in the near future.

Wondering how the Lexicon MC12 EQ would have performed. Anyone have any experience on the Lex as compared to the other systems? Isn't the JBL Synthesis based on the Lexicon EQ system?
Reply
Reply
post #41 of 582
11/3/09 at 8:14pm
Thread Starter 
Quote:
Originally Posted by Frank D View Post

Wondering how the Lexicon MC12 EQ would have performed. Anyone have any experience on the Lex as compared to the other systems? Isn't the JBL Synthesis based on the Lexicon EQ system?

The Lexicon MC12 EQ was an optional feature in the MC12 but not the same room correction used in JBL Synthesis.
Reply
Reply
post #42 of 582
11/3/09 at 9:13pm
Quote:
Originally Posted by primetimeguy View Post

So your products in your listening room came out best. Hmmm. I have an open mind and like the tests but this to me is a point that can't be ignored.

I thought the same thing.
Reply
Reply
post #43 of 582
11/3/09 at 9:51pm
Quote:
Originally Posted by InPhase View Post

Sound Field Management is the technology used in the Bassq. I don't know if this is usable above the transition frequency. Toole only refers to SFM in his book in reference to bass. It seems that Sean is referring to a new Harman eq that no information has been released on yet.

Terrific article and thread in the old school AVS tradition.

Is one or both of the Harman EQs in the article not the Synthesis SDEC-3000 or 4500? Sure looks like it. http://jblsynthesis.com/products/pro...=US&Region=USA

One of the great things about the Synthesis SDEC has always been that the EQ was designed for a limited group of speakers for which the response and characteristics were well known, and for which a unique target curve was specified, all of which are part of the data base in the equalizer. During installation, the model of each speaker in the system was selected, which specified the appropriate target curve. The "problem" for other room correction systems is that they must deal with all speakers, not really having a known starting point as does the Synthesis system. It would appear that these Harman prototypes joined the rest of the correction systems in beginning from scratch, so to speak.

Sean, you mentioned that brands of processors were not meaningful to the article/test, but do these different products not utilize significant differences in technology to reach their objectives? It would seem that, if not by name, then by technology, the processors would certainly play an important role in the evaluations and conclusions. Was that not addressed, recognized, or evaluated in any way?

"....and save the environment, at the same time."
We can't escape that even here. :-D
Reply
Reply
post #44 of 582
11/3/09 at 9:53pm
Quote:
Originally Posted by Cam Man View Post

Terrific article and thread in the old school AVS tradition.

Is one or both of the Harman EQs in the article not the Synthesis SDEC-3000 or 4500? Sure looks like it.

He's not tellin'
Reply
Reply
post #45 of 582
11/3/09 at 10:05pm
Quote:
Originally Posted by Anomad View Post

He's not tellin'

It's certainly a photo double.

Another interesting thing that comes up in this is the sound power dip at about 2KHz at the crossover of the tower. This is the subject of discusssion of a feature of Audyssey; the Mid-range Correction Filter. More food for thought.
Reply
Reply
post #46 of 582
11/3/09 at 10:23pm
Thread Starter 
Quote:
Originally Posted by Kal Rubinson View Post

Well, this indicates an internal consistency but I am not so sure it permits more than a reliable consensus. I am pretty familiar with the first and third samples and have heard them with a number of systems and under a number of conditions but, for example, I cannot say how prominent the kick-drum should be the first. One can get a palpable KICK out of it under some conditions (one can make that happen). That might impress most of us but how much is enough or too much. Same for the lower bass on the other sample.

Of course, this opens the can of worms: How can we possibly know what it was supposed to sound like when we compare a sample with our internal concept of it should sound like. Frankly, I'd be hard put to judge those samples by anything except preference and, mebbe, that's enough. Mebbe we need to revive the old AR live-vs.recorded setup. Is there room behind the curtain for a small combo?

Hi Kal,

I agree with you 100% that recordings and the "circle of confusion" could influence the outcome of these tests, and this disclaimer is clearly stated in the conclusion of our paper.

As you point out, we are measuring preference - not reference - since listeners cannot know for sure what the artist intended regarding the sound of the kick drum for example. That is true for 99% of all scientific listening tests ever done on audio equipment, and most audio equipment reviews. Does that mean they are invalid or should be discounted?

The lack of a reference does not seem to deter trained listeners from reliably formulating a preference among loudspeakers, or in this case, room corrections. Of course, they have much more information available than just a kick drum, and some of the room corrections (RC4 and RC6) we tested have serious colorations that tend to swamp out any effects related to the "circle of confusion."

You mention the old AR live-vs.-recorded, which I consider nothing more than a deceptive parlor trick that only proves how fallible humans are. Case in point: Edison used live versus recorded demonstrations in 1918 to "prove" that his phonograph recordings were indistinguishable from a live performance (see Figure 2.1 in Floyd Toole's book).

I've already argued with AR live-vs-recorded advocates (and there are many still alive) why this method is seriously flawed, but perhaps I will revisit this in an upcoming article in my blog.
Reply
Reply
post #47 of 582
11/3/09 at 11:01pm
I don't know that the sample size of both material and listeners is big enough to consider this a "scientific" test, and since 40% of the EQ's were Harman, it seems weighted to some degree.
Reply
Reply
post #48 of 582
11/3/09 at 11:07pm
Another interesting bit of info on the SYNTHESIS® Harman Audio Test System. http://jblsynthesis.com/products/pro...=US&Region=USA
Reply
Reply
post #49 of 582
11/4/09 at 5:46am
Wondering how the Harmen would sound in a room that had too much, room gain, in the 35Hz and below? Would it take that gain down, leave it (this would not be good) or make it sound worse?
Reply
Reply
post #50 of 582
11/4/09 at 5:58am
Thread Starter 
Quote:
Originally Posted by Cam Man View Post

Terrific article and thread in the old school AVS tradition.

Is one or both of the Harman EQs in the article not the Synthesis SDEC-3000 or 4500? Sure looks like it. http://jblsynthesis.com/products/pro...=US&Region=USA

One of the great things about the Synthesis SDEC has always been that the EQ was designed for a limited group of speakers for which the response and characteristics were well known, and for which a unique target curve was specified, all of which are part of the data base in the equalizer. During installation, the model of each speaker in the system was selected, which specified the appropriate target curve. The "problem" for other room correction systems is that they must deal with all speakers, not really having a known starting point as does the Synthesis system. It would appear that these Harman prototypes joined the rest of the correction systems in beginning from scratch, so to speak.

Sean, you mentioned that brands of processors were not meaningful to the article/test, but do these different products not utilize significant differences in technology to reach their objectives? It would seem that, if not by name, then by technology, the processors would certainly play an important role in the evaluations and conclusions. Was that not addressed, recognized, or evaluated in any way?

"....and save the environment, at the same time."
We can't escape that even here. :-D

Hi Cam Man,

I'm not quite sure what you mean by your last question regarding how did we account for differences in the technology of these room corrections. It is exactly those differences that we were interested in testing.

In one important way, these room corrections products are quite similar: they are all multi-seat room correction devices that take a measurement at several seats and equalize their response to a target defined by the manufacturer.

The Lyngdorf is perhaps most different from the others in terms of where the microphones are placed for the measurements. It takes a measurement at the primary listening seat, as well as several measurements away from the seat to try and capture the sound power response of the loudspeaker/room and its LF behavior. One of our room corrections (RC1) was also focused on the primary listening seat but didn't move the microphones far out into the room but rather kept them within 1-2 m from the listening seat. This produced the best objective measurements and the highest subjective ratings, yet they were not significantly higher than our 6-seat average correction (RC2).

The other room corrections (Audyssey, Anthem and Harman RC2) placed the microphones at the 6 listening seats in the exact same position. The fact that they produced much different in-room measurements and subjective ratings after correction, suggests that the microphone position was not the determining factor here, but rather how the different room corrections choose a target curve and then design and optimize their correction filter to fit the target. The Audyssey is different from the Harman, Lyngdorf, and Anthem in that it implements the correction filter using a FIR filter rather than an IIR filter. The benefits of FIR room/car correction over IIR room/car correction have never been adequately explained or demonstrated to me, and the results from these listening test have not changed that. To date, the benefits of FIR filters are mostly that they cost more to implement (if it costs more, it has to sound better right?), which means that marketing departments have something they can talk about. The Audyssey is also different in that it tries to second guess the sound power response of the loudspeaker and inserts a dip in the target at around 2 kHz by default. If you happen to own well-designed loudspeakers it will actually compromise their sound quality, unless you go in and modify the default target curve and remove the dip.

The differences I describe are what makes these products perform and sound different and were part of the variable "room correction" that we tested.
Reply
Reply
post #51 of 582
11/4/09 at 6:16am
I stand corrected on my previous statements. These were not anechoic tests, how can the 2khz hole be primarily attributed to the speaker? Am I ignorant some standard knowledge that beyond 500hz rooms are invisible?
Reply
Reply
post #52 of 582
11/4/09 at 6:53am
Quote:
Originally Posted by noah katz View Post

"As you are probably aware, Audyssey has released an extension of their room correction software called Dynamic EQ...Do you plan on introducing a similar feature in your Harman products, or will you be using a static tilted target curve?"

Doesn't Dolby volume, which the HK 7500 receiver has, do that?

Hi Noah,

Yes, Dolby Volume is similar in that it attempts to adjust the equalization based on the volume of the content. One of the differences is it uses Dynamic Range Compression. Audyssey Dynamic EQ doesn't use DRC. The user is presented the option of engaging Audyssey Dynamic Volume, which works in conjunction with Dynamic EQ, and it does use DRC.

Regardless, my questions to Dr. Olive had to do with a speculation as to why the test listeners preferred a boost in the bass gently trailing off to a cut in the treble. Without repeating my entire posting, let me summarize my points regarding the preferred bass response.

If the listening test were conducted at levels lower than what the recording engineer experienced during mixing, that might account for the preference in boosting the bass. If that were the case, depending on the difference in levels, a room correction technique that happened to have the right boost for the selected content might have an advantage in the testing. Since Audyssey's Reference target curves are designed to be flat out to 4 kHz this would put them at a disadvantage if the content were played back below the mixing levels.

On the other hand for another selection of content, particularly content mixed to film industry standards, Audyssey MultEQ with Dynamic EQ engaged might have an advantage over competing room correction techniques, if the competing technique used a fixed bass boost or "tilt" in the frequency response regardless of listening levels.

If the Harman room correction techniques use a fixed "tilt" in the frequency response, it is doubtful that overlaying Dolby Volume processing into the mix would improve preference testing, particularly if it adds another bass boost and engages Dynamic Range Compression.

Larry
Reply
Reply
post #53 of 582
11/4/09 at 7:06am
Quote:
Originally Posted by rnrgagne View Post

I don't get the mono thing at all, sure the listener may be able to discern differences better that way but, and correct me if I'm wrong, aren't room EQ's like ARC & Audyssey designed to EQ multiple speakers and effectively determine the best use of filters and delays based on a summation of same? Don't they have a fuzzy logic that might allocate resources differently than if they truly did each speaker individually?

Quote:
Originally Posted by LarryChanin View Post

Hi,

I agree that one of Audyssey's strong points is its ability to timbre match all satellite speakers.

However, to do that requires individually equalizing each speaker to the same target curve. The fuzzy logic part of the equation is used to develop and weigh more heavily equalization filters that consider common acoustic problems in different locations. Nevertheless, each speaker is still individually equalized.

Larry

Hi,

I'd like to add one other point about this fuzzy logic issue. This proprietary method of unequally weighing the equalization filters at different locations in the room has value in improving the sound quality at non-primary listening locations without unduly compromising the "sweet spot" sound quality.

Since Dr. Olive's testing put the test listeners in the primary listening location, the test was not geared to determining the preferences in the "cheap seats" where Audyssey might have an advantage.

Larry
Reply
Reply
post #54 of 582
11/4/09 at 7:08am
Quote:
Originally Posted by Anomad View Post

music preference is subjective, you might like a song and I might not.

How is that different from room correction, where you might like what one RC is doing and I might not? It's not like there was some objective target to hit, since there is no reference for music mixing studios, so all they could test for was subjective preference.
Quote:
This is a scientific study, and it's presented like "here is this incredible finding we have, some products make the sound better, some really mess it up. Sorry, but that's all we're going to tell you."

No. The study shows subjective preference, that too averaged amongst all listeners. It's possible some prefered RC3 over RC2 but, since we didn't see individiual score cards, all we can tell from the presentation is which was prefered on average.
Quote:
Sure, I'll go buy all 5 units (6 if I include the neptune) and run my own tests...so this article helps me how?

The article correlates measurements with preference. So if you end up liking some of units you tried, you'll know why. When newer units come out, you'll have a better handle on which ones are worth trying out and which ones would be a waste of time for you.
Quote:
It's like if consumer reports found the brakes are bad on a certain car, but said well, "We won't tell you which one, you will have to go buy them and try to find out for yourself. Happy driving, don't tailgate"

Get a grip. Nothing so life-threatening going on here.
Reply
Reply
post #55 of 582
11/4/09 at 7:33am
Quote:
Originally Posted by Tonmeister2008 View Post

I agree with you 100% that recordings and the "circle of confusion" could influence the outcome of these tests, and this disclaimer is clearly stated in the conclusion of our paper.

As you point out, we are measuring preference - not reference - since listeners cannot know for sure what the artist intended regarding the sound of the kick drum for example. That is true for 99% of all scientific listening tests ever done on audio equipment, and most audio equipment reviews. Does that mean they are invalid or should be discounted?

Not at all. In fact, preference, assessed statistically over as broad a base as possible, is the right way to make a product that will please as many users as possible. I was merely beating the dead, but accurate, horse.

Quote:
You mention the old AR live-vs.-recorded, which I consider nothing more than a deceptive parlor trick that only proves how infallible humans are. Case in point: Edison used live versus recorded demonstrations in 1918 to "prove" that his phonograph recordings were indistinguishable from a live performance (see Figure 2.1 in Floyd Toole's book).

I know. That was a throw-away remark. Nonetheless, rather than A/B comparison of live-vs.recorded, pretraining the subject pool with live performances as close to those in the recordings would go a long way towards eliminating the issue. I guess it comes down to my general aversion to using studio-intensive pop recordings for critical assessment of anything.....except preference.

Quote:
I've already argued with AR live-vs-recorded advocates (and there are many still alive) why this method is seriously flawed, but perhaps I will revisit this in an upcoming article in my blog.

Sure. Would be interesting.
Reply
Reply
post #56 of 582
11/4/09 at 7:34am
Thread Starter 
Quote:
Originally Posted by jmichaelf View Post

I stand corrected on my previous statements. These were not anechoic tests, how can the 2khz hole be primarily attributed to the speaker? Am I ignorant some standard knowledge that beyond 500hz rooms are invisible?

The 2 kHz hole is in the sound power response of the speaker, as shown in the anechoic measurements of the B&W802N in slide 8 of the presentation
Reply
Reply
post #57 of 582
11/4/09 at 7:38am
Quote:
Originally Posted by Tonmeister2008 View Post

The Lyngdorf is perhaps most different from the others in terms of where the microphones are placed for the measurements. It takes a measurement at the primary listening seat, as well as several measurements away from the seat to try and capture the sound power response of the loudspeaker/room and its LF behavior.

This is probably responsible for my disappointment with its implementation. Unless one sits in the the target seat (and does not move or lean much), one gets a sound that is even poorer than without Lyngdorf EQ.

Quote:
One of our room corrections (RC1) was also focused on the primary listening seat but didn't move the microphones far out into the room but rather kept them within 1-2 m from the listening seat. This produced the best objective measurements and the highest subjective ratings, yet they were not significantly higher than our 6-seat average correction (RC2).

Is that, primarily, a function of the mic positioning or of the underlying correction algorithms?
Reply
Reply
post #58 of 582
11/4/09 at 7:42am
Thread Starter 
Quote:
Originally Posted by LarryChanin View Post

Hi,

I'd like to add one other point about this fuzzy logic issue. This proprietary method of unequally weighing the equalization filters at different locations in the room has value in improving the sound quality at non-primary listening locations without unduly compromising the "sweet spot" sound quality.

Since Dr. Olive's testing put the test listeners in the primary listening location, the test was not geared to determining the preferences in the "cheap seats" where Audyssey might have an advantage.

Larry

It's true that we didn't do listening tests in the other seats. That will be another experiment.

However, I do have objective measurements for the room corrections in the other seats (which we didn't show in the paper), that give a good indication of how they sound.
Reply
Reply
post #59 of 582
11/4/09 at 8:53am
Quote:
Originally Posted by Tonmeister2008 View Post

Hi InPhase,



If you buy decent loudspeakers with a flat on-axis response and smooth sound power response, then there is little need to equalize them above the room transition frequency (300-500 Hz).

Bingo... I'd prefer EQ solutions only function below the transition frequency and leave the loudspeaker design issues to those who are designing them. If they do it poorly, then they should be judged accordingly.

What we need Sean, is for a review magazine to use your methods and room setup. Soundstage has that sort of relationship with the NRC, why not have Harmon develop that kind of relationship with a mainstream review magazine? I know cost would be an issue because these things are not free but I think it would be in the best interest of Harmon to pull the industry out of the Dark Ages.
Reply
Reply
post #60 of 582
11/4/09 at 9:26am
Quote:
Originally Posted by Tonmeister2008 View Post

It's true that we didn't do listening tests in the other seats. That will be another experiment.

However, I do have objective measurements for the room corrections in the other seats (which we didn't show in the paper), that give a good indication of how they sound.

Hi Sean,

Thanks for the response.

At the risk of stating the obvious, when setting the filters for multiple seating locations versus a single seat we are always dealing in compromises. Do we give added weight to the primary listening location, do we consider other locations at all, if so, do we give equal weight to each seat, etc?

Of course, regardless of the correction method, the user usually has the option of skewing the results as he/she sees fit by clustering or expanding the placement of measurement positions, as you have done in your experiment. As the measurements are spread further apart from the primary listening location the acoustic environment changes and the resulting correction filters must differ. If the filters appropriate to the expanded positions are equally averaged and applied to the primary location the correction there will be compromised. Generally, the more the measurements are expanded the more the primary position's correction will be effected.

In an attempt to minimize this inevitability, Audyssey groups the responses of the different measurement positions into clusters based on similarities. Then instead of weighing the filters equally it gives added weight to filters appropriate to multiple measurements (seats). Whereas the primary listening location's response is still compromised, depending on whether it shares acoustic characteristics with other measured locations, the negative effects may be somewhat mitigated.

Do the Harman prototypes also consider acoustic similarities between the measurement positions before applying the final single correction filter, or is each measurement given equal weight?

Thanks.

Larry
Reply
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Receivers, Amps, and Processors

AVS Top Picks

AVS › AVS Forum › Audio › Receivers, Amps, and Processors › The Subjective and Objective Evaluation of Room Correction Products