PSYCHOACOUSTICS AND I

[Blumlein 88]

I sound like an subjectivist saying this, but I have used Tact gear and Dirac embedded in gear which allow instant switching for comparison. With neither has anyone said, "man it sounds better if you leave that room correction off". That does include using it just for 500 hz and below vs letting it correct full range. When the DSP goes on everyone I have seen has no problem deciding it is a good step toward better sound. I do spend time customizing target curves which are sometimes small adjustments toward someone's preference. Whether that someone is me or friends.

Audyssey on the other hand, always in earlier versions people said, "I think I want that off". I haven't had hands on with the very latest versions of it. You could hear things it did improve, but as many it seemed to make worse. With other options I didn't spend much time determining how it was different than what Tact or Dirac were doing.

PEQ's for just the low end is an easy DSP helper. Other EQ even full range is helpful if you have some ability to measure it. Though I can't say it works as well as something like Dirac.

 

[Brad]

The more sophisticated room correction does more than apply EQ.
Drivers are also time aligned and crossover phase problems corrected (by improving the impulse response).

I use acourate and have, until recently, always preferred the full room correction.

I use synergy type horns and recently performed an outdoor measurement to apply minimum phase EQ to correct them to a flat anechoic response. With that correction AND time aligned drivers AND linear phase crossovers, they sound better without the extra mixed phase room correction.

This is consistent with what Toole says.
So I would suggest that full room correction can provide a benefit, depending on what the biggest fault is. Note that unless you listen at the designed listening distance, most speakers can benefit from DSP time alignment.

 

[Cosmik]

Interestingly, Genelec has a BYPASS button for you to click on in order to A-B test room correction in real-time. If room correction were flawed, why let their users A-B their audio setup with and without DSP based room correction. I guess DSP based room correction works in many cases.

I believe you may wish to look at a phenomenon called expectation bias..!

The typical view of correction is this:

If your listening environment is typical then there will be nulls at particular frequencies, and a peaks at frequencies, a correction system will adjust the gain of the these various frequencies to give a flat response at the measuring/listening position – all good.

(just picked at random).
Note that he says "all good" as though what he has said is self-evidently true. But what Toole and others are hinting at is that our hearing works in both the time and frequency domains so that if delayed reflections cause a null, we don't hear the null at all. This would appear to be true: if we walk around our listening room, any audio source does not seem to change its character - even though a spectrum analyser would show it to be all over the place as we walked. There are very good reasons why this would be an advantage to a human or other animal - we move through the world able to recognise and locate sources of sound as a stable 'scene' just as we do with vision.

A delay - as in a delayed reflection - is a very specific thing. It cannot be removed by tweaking a graphic equaliser at the source. And if the human brain is already compensating for it (as it seems to), then 'correction' will sound wrong.

 

[svart-hvitt]

I believe you may wish to look at a phenomenon called expectation bias..!

The typical view of correction is this:

(just picked at random).
Note that he says "all good" as though what he has said is self-evidently true. But what Toole and others are hinting at is that our hearing works in both the time and frequency domains so that if delayed reflections cause a null, we don't hear the null at all. This would appear to be true: if we walk around our listening room, any audio source does not seem to change its character - even though a spectrum analyser would show it to be all over the place as we walked. There are very good reasons why this would be an advantage to a human or other animal - we move through the world able to recognise and locate sources of sound as a stable 'scene' just as we do with vision.

A delay - as in a delayed reflection - is a very specific thing. It cannot be removed by tweaking a graphic equaliser at the source. And if the human brain is already compensating for it (as it seems to), then 'correction' will sound wrong.

I other words, you claim users of room correction software are victims of expectation bias, even among some of the world’s most experienced listeners (like award-winning mastering engineers).

That’s a bold claim.

I would have liked to see peer-reviewed research to support that conclusion. Toole is an experienced researcher but hardly a guru whose insights are the final answer.

In all research areas there are different views. I’d be surprised if Toole has the final answer in acoustics.

Do not hesitate to come up with links and files with peer-reviewed research to support your conclusion that DSP based room correction is a dead end.

 

[Cosmik]

I other words, you claim users of room correction software are victims of expectation bias, even among some of the world’s most experienced listeners (like award-winning mastering engineers).

I can supply an example of an award winning producer and a hugely experienced sound engineer falling prey to expectation bias!
http://www.bbc.co.uk/programmes/p01tds0d

It's fairly painful to listen to, but Steve Levine ("Steve has collected a number of prestigious awards including BPI Producer of the Year...") and Stephen Rinker (vastly experienced sound engineer) wax lyrical about what they are hearing when presented with high res versus CD quality music (including a track that Rinker himself engineered). But... inevitably when presented with them in the wrong order, they get it wrong, thinking the CD quality is high res because it is so much more detailed, etc. with smoother reverb tails, etc. etc..

 

[svart-hvitt]

I can supply an example of an award winning producer and a hugely experienced sound engineer falling prey to expectation bias!
http://www.bbc.co.uk/programmes/p01tds0d

It's fairly painful to listen to, but Steve Levine ("Steve has collected a number of prestigious awards including BPI Producer of the Year, Musicweek Top Singles Producer and a Grammy for his work with Deniece Williams. And a SONY Radio Award. ") and Stephen Rinker (vastly experienced sound engineer) wax lyrical about what they are hearing when presented with high res versus CD quality music (including a track that Rinker himself engineered). But... inevitably when presented with them in the wrong order, they get it wrong, thinking the CD quality is high res because it is so much more detailed, etc. with smoother reverb tails, etc. etc..

I do not question that expectations bias exists among the most experienced, competent professionals. FWIW they cannot always distinguish mono from stereo (thinking that mono gives more punch etc. ...).

But there’s a difference between what a professional does on an anecdotal basis and how a professional streamlines his daily routine workflow.

You seem to use expectation bias as evidence that DSP based room correction is more heat than light. I guess you see the rhetorical fallacy therein?

 

[Cosmik]

I do not question that expectations bias exists among the most experienced, competent professionals. FWIW they cannot always distinguish mono from stereo (thinking that mono gives more punch etc. ...).

But there’s a difference between what a professional does on an anecdotal basis and how a professional streamlines his daily routine workflow.

You seem to use expectation bias as evidence that DSP based room correction is more heat than light. I guess you see the rhetorical fallacy therein?

I am arguing for zero 'room correction' (we must be careful not to use the phrase as though it is a given that room correction exists and works) on the basis of a rational argument - it was not I who introduced the idea of the A-B button and asking award winning producers to say which they like better as evidence of whether 'room correction' 'works'. I think that was you...

In any situation, switching in a parametric equaliser may give a temporary "Wow! I can now hear the bow scraping on the strings so much more clearly - as though a veil has been lifted!" effect. For a while, music producers in their streamlined daily workflow couldn't imagine life without gated reverb; recently they couldn't imagine life without extreme compression and/or clipping for extra punch and loudness. I am not in awe of music 'engineers' and producers...

 

[Burning Sounds]

I use Acourate filters (X-over and EQ) in JRiver and can switch very quickly (2 mouse clicks) between them and stock IIR X-over/eq filters for my Linkwitx LX521s. Subjectively I prefer the Acourate filters and measurements show not only a better FR, but also a better step response and group delay. I don't fill a large 100Hz dip that is probably caused by the large sofa.

I have a hunch that the better the polar response of a loudspeaker (without room correction) the better it will respond when well thought out room correction is applied. I think this is especially true off axis. As far as I know Siegfried Linkwitz is not an advocate of room correction. So, is it all "good" when room correction is applied? No, like anything in audio there are always trade-offs. A good recording seems to benefit from room correction, such as improved low level detail and separation, but likewise low level sibilance on a poor recording can become more instrusive. Overall for me the positives outweigh the negatives.

The only recording where I switch off Acourate is Roger Waters' Amused To Death as the room correction kills the effects of this Q Sound recording. Anyone else using room correction, whether Dirac or Acourate, experience this?

 

[Fitzcaraldo215]

I am arguing for zero 'room correction' (we must be careful not to use the phrase as though it is a given that room correction exists and works) on the basis of a rational argument - it was not I who introduced the idea of the A-B button and asking award winning producers to say which they like better as evidence of whether 'room correction' 'works'. I think that was you...

In any situation, switching in a parametric equaliser may give a temporary "Wow! I can now hear the bow scraping on the strings so much more clearly - as though a veil has been lifted!" effect. For a while, music producers in their streamlined daily workflow couldn't imagine life without gated reverb; recently they couldn't imagine life without extreme compression and/or clipping for extra punch and loudness. I am not in awe of music 'engineers' and producers...

You raise some valid points, but it seems you are an outsider looking in - a non-user without much relevant experience with the technology. And, we understand your continuing objections to listening comparisons involving human subjects, even scientifically conducted, double blind, and controlled ones.

Be that as it may, here is an interesting older paper by Sean Olive, et al that investigates the subject and which others might find useful:

https://drive.google.com/file/d/0B9...cyLWEzZTAtMGJiODQ1ZTUxMGQ4/view?ddrp=1&hl=en#

Theoretical arguments against the technology aside, it seems that in these tests many subjects convincingly preferred some of the "room correction" products over no correction in the test room. Other products did not fare so well. But, all products used a similar methodology of calculating EQ filters based on mike measurements of the speaker/room and a comparison of those against the product's specific target frequency response curve. Target curves varied, as did the internal methods - filter types, etc. - for each product.

Unlike you, I think listening tests conducted scientifically that reveal significant preferences in a statistically significant way over many tests and test subjects are quite useful in that they avoid the gee whiz factor on one particular music recording that you object to.

 

[Cosmik]

it seems you are an outsider looking in - a non-user without much relevant experience with the technology.

Not quite the case.

My system is full DSP/active/three-way with phase & amplitude driver correction, time alignment *and* EQ curves for speaker correction. In other words, it has all the components of a system with DSP 'room correction', and I am quite at home waving a measurement microphone around. The only difference is where and how I use the mic, and the resulting derivation of the filters' impulse responses.

I have been running this system for about four years - I wrote all the software including the DSP 'convolution engine'.

 

[mitchco]

@Cosmik cool! Tell us more about your custom DSP solution. Have you posted measurements?

 

[Fitzcaraldo215]

Not quite the case.

My system is full DSP/active/three-way with phase & amplitude driver correction, time alignment *and* EQ curves for speaker correction. In other words, it has all the components of a system with DSP 'room correction', and I am quite at home waving a measurement microphone around. The only difference is where and how I use the mic, and the resulting derivation of the filters' impulse responses.

I have been running this system for about four years - I wrote all the software including the DSP 'convolution engine'.

My mistake. But, if you are using DSP for EQ curves, then I am not clear on your objections. Do you use it to correct for room induced modal peaks, at least?

 

[Cosmik]

@Cosmik cool! Tell us more about your custom DSP solution. Have you posted measurements?

Hi Mitchco. Slightly off-topic, it's a Linux-based system that is slaved to the DAC's sample rate. The incoming 16-bit data is processed with three FFT-based filters and sent to the DACs as 24-bit.

The filters effectively overlay the following functions:

• crossover filters
• driver correction
• EQ (for various reasons - but not 'room correction' per se!)

The aim is for linear phase and time alignment, with minimal EQ effectively for baffle step compensation - depth set by ear. The bass can be EQ'ed and the roll-off tailored as required.

Different configs can be saved and switched in instantaneously. Large changes in crossover frequencies and slopes result in virtually no discernible change in the sound (although obviously huge changes if listening to just one driver in isolation) - sighted test caveat - so I think it is an indication it is working reasonably well.

Measurements-wise, I am only aiming to correct the drivers, so I am not making any in-the-listening room measurements. I used to have a great listening room where I felt able to eschew any 'correction' at all, but have recently transferred to a room where I reluctantly may have to do some slight work on the bass... :-(

 

[Cosmik]

My mistake. But, if you are using DSP for EQ curves, then I am not clear on your objections. Do you use it to correct for room induced modal peaks, at least?

Hi Fitzcaraldo. See above.

 

[amirm]

Audyssey on the other hand, always in earlier versions people said, "I think I want that off". I haven't had hands on with the very latest versions of it. You could hear things it did improve, but as many it seemed to make worse.

My experience exactly.

 

[amirm]

I know of only one published controlled blind test of Room EQ products:

The Subjective and Objective Evaluation of
Room Correction Products
Sean E. Olive1, John Jackson2, Allan Devantier3, David Hunt4 and Sean M. Hess5
Harman International, Northridge, CA, 91329, USA

This was the outcome:

I have circled the no-EQ, control rating. As we see, three other schemes (two from Harman) did better. Two did the same or worse.

Expert listeners were used. Here is the reason they voted the way they did with respect to spectral qualities:

And subjective results:

On target curves:

The smoothest, falling curve won. Losing systems for example had deficient bass response.

And the same in words:

 

[Blumlein 88]

snip.........

The only recording where I switch off Acourate is Roger Waters' Amused To Death as the room correction kills the effects of this Q Sound recording. Anyone else using room correction, whether Dirac or Acourate, experience this?

I have listened to Amused to Death many times on a system using the Tact room correction. It didn't seem to interfere with the Q Sound process at all. This with panel speakers which don't reflect much at all off the side walls. Lots of reflection off the rear wall 62 inches behind the speakers.

 

[March Audio]

The only recording where I switch off Acourate is Roger Waters' Amused To Death as the room correction kills the effects of this Q Sound recording. Anyone else using room correction, whether Dirac or Acourate, experience this?

I havent listened to this (will look on Roon Tidal when I get home) however one thing that struck me about the Acourate and DSP XO i implemented was how much better spatial effects work. So your different findings are interesting.

 

[Cosmik]

I know of only one published controlled blind test of Room EQ products:

The Subjective and Objective Evaluation of
Room Correction Products

Sean E. Olive1, John Jackson2, Allan Devantier3, David Hunt4 and Sean M. Hess5
Harman International, Northridge, CA, 91329, USA

This was the outcome:

I think we have to be careful about what these tests mean. If the speaker used (a B&W?) was passive and with its own foibles, then the 'room correction' may also be performing speaker correction. In the audiophile world, the assumption is that a "high quality" speaker such as a B&W is inherently good, but it would be interesting* to do a similar comparison using a modern DSP-based active speaker as the 'mule'.

*but not necessarily scientific

 

[Burning Sounds]

I have listened to Amused to Death many times on a system using the Tact room correction. It didn't seem to interfere with the Q Sound process at all. This with panel speakers which don't reflect much at all off the side walls. Lots of reflection off the rear wall 62 inches behind the speakers.

I havent listened to this (will look on Roon Tidal when I get home) however one thing that struck me about the Acourate and DSP XO i implemented was how much better spatial effects work. So your different findings are interesting.

Just had another listen - and of course the Q Sound effects are all where they should be - the old soldier way outside and forward of the left speaker and the dog over my right shoulder - I made the statement based on my previous set of Acourate filters (the ones I wasn't entirely happy with in a previous thread on room correction). I hadn't listed to Amused to Death since I redid the crossovers and eq in Acourate a few months ago. Thanks guys!