Measurements, EQ, and what we actually hear

[James]

Following a suggestion from curtis, I'm starting a new thread to discuss this topic.

I use a series of 1/12th octave test tones when trying different speaker placements/positioning, and it's very correlated with what I hear with music. I'll usually listen to music for a while, and see what sounds different, then listen to the test tones after that.

The clear exception is with room modes - they don't show up on the test tones, but show up on plucked double bass. For those, I rely on a room mode calculator and our piano. I can generally find the note that's more resonant than it should be on the piano.

That's why I'm leaning towards getting a precise parametric equalizer like the Behringer, fine tuning the response a bit, so that the frequency response is extended and even more neutral than it is now. I'm not imagining I'd change a lot - just boost the low bass a bit, tame a room mode or two, and bring down the 500-1000Hz rise a little bit.

I lean towards less intervention rather than more.

[James]

But then I've come across the REW, which seems much more complex and sophisticated than what I'm doing, but seemingly less correlated to what people actually hear.

So I'm interested in anything people want to share about using it. Especially if/when the measurements don't correlate to what you hear - how do you make sense of that?

[sonic icons]

"REW is free software for room acoustic measurement, loudspeaker measurement and audio device measurement. The audio measurement and analysis features of REW help you optimise the acoustics of your listening room, studio or home theater and find the best locations for your speakers, subwoofers and listening position."

https://www.roomeqwizard.com/

I have used REW. In general, ALL audio measurements (frequency response, harmonic distortion, impulse response, waterfall plots, etc. etc.) are difficult to correlate with "what people actually hear". But you can learn a lot from REW measurements of your audio system, and how your system interacts with your listening environment, that you can't learn from "just listening".

I think REW provides an amazing range of audio measurement capabilities considering that it's freeware, although it does of course require some hardware to work, such as a USB measurement microphone and a Windows laptop computer with an HDMI connector (if your AVR or audio processor has HDMI inputs).

[James]

Yes, it looks like very impressive software, and I'm also kind of amazed that it's offered for free. But I'm puzzled by the differences between measurements and listening.

For example, merrymaid520 posted some REW measurements (I believe they're the ones with the adjustments made to more accurately reflect the way people hear) on page 52 of the LX thread.

They show what look like large deviations from neutrality to me, and yet merrymaid says they sound neutral.

Would you expect to hear those deviations in your system? And/or would you expect those measurements to correlate with what you hear with test tones?

I'm not trying to pick on merrymaid - those are just the easiest REW measurements for me to find.

[curtis]

Yes, it looks like very impressive software, and I'm also kind of amazed that it's offered for free. But I'm puzzled by the differences between measurements and listening.

For example, merrymaid520 posted some REW measurements (I believe they're the ones with the adjustments made to more accurately reflect the way people hear) on page 52 of the LX thread.

They show what look like large deviations from neutrality to me, and yet merrymaid says they sound neutral.

Would you expect to hear those deviations in your system? And/or would you expect those measurements to correlate with what you hear with test tones?

I'm not trying to pick on merrymaid - those are just the easiest REW measurements for me to find.
A few things:
The room has a lot to do with the FR measurements. FR is not the only thing that determines how a speaker sounds. Trying to correlate what one hears in their room to what someone else will hear in their room is a lost cause.
Those deviations are not that big. You don't listen to test tones when listening to music as you are listening to harmonics(which is what gives you the difference in sounds of instruments or voices) as well.
Lastly, people have to understand the deficiencies in their own hearing, as well as understand measuring equipment is much more sensitive to our hearing.
I am very skeptical when someone can be specific to what they are hearing before they look at an FR graph or any other measurement for that matter. But hearing something, and then going back at looking at measurements to verify, that is more credible.

[James]

Yes, rooms are very important - I get that part. I didn't mean would you hear the same thing as merrymaid - I meant if you had a set of REW measurements like that, would you expect to hear deviations from neutrality.

What qualifies as a big deviation from neutrality? I see what looks like about a 15db difference between 100Hz and a bit over 200Hz. Will most people not hear that?

Yes, with music we hear many frequencies at the same time.

Would you expect that the REW measurements correlate with listening to test tones? Or would they be different, and if different, how would you explain that difference?

[curtis]

Yes, rooms are very important - I get that part. I didn't mean would you hear the same thing as merrymaid - I meant if you had a set of REW measurements like that, would you expect to hear deviations from neutrality.

What qualifies as a big deviation from neutrality? I see what looks like about a 15db difference between 100Hz and a bit over 200Hz. Will most people not hear that?

Yes, with music we hear many frequencies at the same time.

Would you expect that the REW measurements correlate with listening to test tones? Or would they be different, and if different, how would you explain that difference?
First, define what you mean by neutrality.

That 100hz to 200hz difference is caused by the room, and it is high peak to low peak, and those peaks are narrow. I think one would be hard press to hear the difference with a sliding sine wave tone.

I do think REW would give a good general idea of how a speaker sounds in-room, but not as specific as you make it out to be.

[James]

Well, absolute neutrality would mean a completely flat frequency response, I would think.

I picked that one because it's the biggest - there are also a lot of other smaller ones as well. Would you expect the approximately 8db difference between 90 and 100Hz to be audible?

What about the 1/12 octave test tones that I use? They're not a sine wave sweep.

[davef]

Well, absolute neutrality would mean a completely flat frequency response, I would think.

That is an incorrect statement.





What about the 1/12 octave test tones that I use? They're not a sine wave sweep.

Test tones are, in fact, sine-waves, what frequency test-tones are you using?

[davef]

Yes, it looks like very impressive software, and I'm also kind of amazed that it's offered for free. But I'm puzzled by the differences between measurements and listening.

For example, merrymaid520 posted some REW measurements (I believe they're the ones with the adjustments made to more accurately reflect the way people hear) on page 52 of the LX thread.

They show what look like large deviations from neutrality to me, and yet merrymaid says they sound neutral.

Would you expect to hear those deviations in your system? And/or would you expect those measurements to correlate with what you hear with test tones?

I'm not trying to pick on merrymaid - those are just the easiest REW measurements for me to find.

James,

The background of my entire career in this industry is to try and correlate what we hear with measurements, and I am still not there - nor do I suspect this industry will ever be. We all hear differently and the biggest mistake one can make is to trust one's own hearing.

As measurement technology improves, so does our ability to at least make an estimate as to what we can expect to hear. Your method of using sine-wave test tones combined with an SPL meter to try and determine how a speaker measures (ie sounds like) is very flawed, and a method that was thrown away over 30 years ago.

First off, sine wave test tones excite only a specific frequency, so unless you are able to run 1/12th octave test tones (12 test tones per each octave), your data is basically not usable.

Second, sine wave test tones are steady state signals, they are not music and can actually be quite damaging to a speaker (especially a ribbon tweeter)

Third, with using your SPL meter to measure ( I assume this is actually a calibrated spl meter, hopefully not an app on a mobile device) - the only thing you are somewhat accurately measuring is the acoustics of your room (and even that is questionable), not the speakers. (this is why this method was thrown in the dumpster so long ago)

And finally, and I must stress this - please do not trust your hearing to even attempt to estimate an actual frequency response measurement. There is nothing more inaccurate than your own hearing and it will take you many, many years of critical listening experience combined with understanding how to take accurate measurements and understanding of the differences between an in-room response and an anechoic response.

My advice to you, throw out all of your current assumptions regarding audio measurements, read as much as you can about using REW (it is a good place to start) and most importantly - about how to interpret those measurements.

Hope this helps!

[James]

That is an incorrect statement.




Test tones are, in fact, sine-waves, what frequency test-tones are you using?
Please elaborate on the first statement, if you don't mind.

Yes, but it's not a sweep - the test tones go from a low c (32.7 Hz), in 1/12 octave steps, up to the c at 4186.01 Hz. So it's a chromatic scale that goes up to and a bit beyond the top note on a violin.

It leaves out the upper octaves, but I don't hear much in the top octave or two.

[James]

James,

The background of my entire career in this industry is to try and correlate what we hear with measurements, and I am still not there - nor do I suspect this industry will ever be. We all hear differently and the biggest mistake one can make is to trust one's own hearing.

As measurement technology improves, so does our ability to at least make an estimate as to what we can expect to hear. Your method of using sine-wave test tones combined with an SPL meter to try and determine how a speaker measures (ie sounds like) is very flawed, and a method that was thrown away over 30 years ago.

First off, sine wave test tones excite only a specific frequency, so unless you are able to run 1/12th octave test tones (12 test tones per each octave), your data is basically not usable.

Second, sine wave test tones are steady state signals, they are not music and can actually be quite damaging to a speaker (especially a ribbon tweeter)

Third, with using your SPL meter to measure ( I assume this is actually a calibrated spl meter, hopefully not an app on a mobile device) - the only thing you are somewhat accurately measuring is the acoustics of your room (and even that is questionable), not the speakers. (this is why this method was thrown in the dumpster so long ago)

And finally, and I must stress this - please do not trust your hearing to even attempt to estimate an actual frequency response measurement. There is nothing more inaccurate than your own hearing and it will take you many, many years of critical listening experience combined with understanding how to take accurate measurements and understanding of the differences between an in-room response and an anechoic response.

My advice to you, throw out all of your current assumptions regarding audio measurements, read as much as you can about using REW (it is a good place to start) and most importantly - about how to interpret those measurements.

Hope this helps!
Well, then I'm not off-base with my questions/uncertainty about the topic - that's good to know.

I do use 1/12 octave steps, and I don't use an SPL - I just listen to the tones.

The idea that I shouldn't trust my own hearing is hard for me to accept somehow. I mean, I understand from hearing tests that I've lost some in the top region, and accept that - I wouldn't claim that I hear what's going on up there.

And when my wife hears more than I do down in the 30-40 Hz region, I believe her. But thinking that within my limits, I'm not hearing what's there is harder for me.

[davef]

Please elaborate on the first statement, if you don't mind.

Yes, but it's not a sweep - the test tones go from a low c (32.7 Hz), in 1/12 octave steps, up to the c at 4186.01 Hz. So it's a chromatic scale that goes up to and a bit beyond the top note on a violin.

It leaves out the upper octaves, but I don't hear much in the top octave or two.

James, a sine wave is not a sweep. A single frequency, steady state sound-wave, by definition, is a sine wave.

James, please don't take this as anything but constructive, I can see you have tremendous interest in this subject, but you do have much to learn, I suggest you purchase some books on audio measurements to get the basics.

[davef]

The idea that I shouldn't trust my own hearing is hard for me to accept somehow. I mean, I understand from hearing tests that I've lost some in the top region, and accept that - I wouldn't claim that I hear what's going on up there.

Are you assuming that you hear every frequency evenly? Are you assuming you have absolutely perfect hearing? I don't mean able to hear from 20Hz to 20kHz, but that you hear each frequency the same as another frequency? You are doing yourself a tremendous disservice to make that assumption.

The shape of your ears, the shape of your ear canals, including a dozen other factors unique to you determine the frequency response of your own hearing.

In addition, human hearing also changes with volume... Please research the Fletcher Munson curve.

Trying to determine the frequency response of a speaker, any speaker, by simply listening to steady state waves is no different than trying to determine how something tastes by simply looking at the ingredients. Even trained critical listeners and/or experienced musicians could not even attempt to do so.

[Pogre]

James,

The background of my entire career in this industry is to try and correlate what we hear with measurements, and I am still not there - nor do I suspect this industry will ever be. We all hear differently and the biggest mistake one can make is to trust one's own hearing.

As measurement technology improves, so does our ability to at least make an estimate as to what we can expect to hear. Your method of using sine-wave test tones combined with an SPL meter to try and determine how a speaker measures (ie sounds like) is very flawed, and a method that was thrown away over 30 years ago.

First off, sine wave test tones excite only a specific frequency, so unless you are able to run 1/12th octave test tones (12 test tones per each octave), your data is basically not usable.

Second, sine wave test tones are steady state signals, they are not music and can actually be quite damaging to a speaker (especially a ribbon tweeter)

Third, with using your SPL meter to measure ( I assume this is actually a calibrated spl meter, hopefully not an app on a mobile device) - the only thing you are somewhat accurately measuring is the acoustics of your room (and even that is questionable), not the speakers. (this is why this method was thrown in the dumpster so long ago)

And finally, and I must stress this - please do not trust your hearing to even attempt to estimate an actual frequency response measurement. There is nothing more inaccurate than your own hearing and it will take you many, many years of critical listening experience combined with understanding how to take accurate measurements and understanding of the differences between an in-room response and an anechoic response.

My advice to you, throw out all of your current assumptions regarding audio measurements, read as much as you can about using REW (it is a good place to start) and most importantly - about how to interpret those measurements.

Hope this helps!

Absolutely so well put Dave! Every time I see someone talk about testing their speakers with test tones and their ears, in room, I die a little bit inside... lol.

[James]

Well, I did a bunch of thinking and researching on this before looking at the thread again, and it's clearly a very complex subject. And I often find that with interesting subjects, the more you learn, the more questions you have.

If it were possible, I'd buy an equalizer, mic and REW software, learn how to use it, and see what results I get. Unfortunately, that's not possible right now. Also, it turns out that REW software doesn't work with an I-Pad, which is the only laptop we have.

As soon as it's feasible, I'll try to pick up the Behringer - that can be plugged in and used right away. And I can manually adjust things until I like the results, which should be gratifying even if they're not perfect/correct.

Of course, I'll be extremely curious about how measurements would look, and how that would compare with what I hear. At some point, I may be able to justify more of an investment. One nice thing about the Behringer equalizer is that it can work with REW - of course there would be a lot to learn about that.

I'm still interested in anything people want to share about their experiences with REW, and how what they see with the measurements does or doesn't correlate with what they hear.

[James]

If anybody's interested, I have found and used a very cool site where you can test various things, including different aspects of your hearing:

Free Online Audio Tests, Test Tones and Tone Generators (audiocheck.net) (https://www.audiocheck.net/index.php)

If you like this sort of thing (and have time), it's a lot of fun. And I find that doing the tests can improve one's ability to hear different things, like timing differences.

[James]

Are you assuming that you hear every frequency evenly? Are you assuming you have absolutely perfect hearing? I don't mean able to hear from 20Hz to 20kHz, but that you hear each frequency the same as another frequency? You are doing yourself a tremendous disservice to make that assumption.

The shape of your ears, the shape of your ear canals, including a dozen other factors unique to you determine the frequency response of your own hearing.

In addition, human hearing also changes with volume... Please research the Fletcher Munson curve.

Trying to determine the frequency response of a speaker, any speaker, by simply listening to steady state waves is no different than trying to determine how something tastes by simply looking at the ingredients. Even trained critical listeners and/or experienced musicians could not even attempt to do so.
Thanks for the response - those are good points.

I knew, but forgot about the Fletcher Munson curve - that definitely would point towards boosting the bass a bit, especially at lower volumes.

[curtis]

James,

RE: "I'd buy an equalizer"

I highly suggest before you do that, you understand what is going on in your room and your ears.

Also, all your equipment is full analog, all that Behringer stuff runs in the digital domain, so it sends the signal through a ADA cycle, which can also harm the sound. It is the reason why most people use the Behringers to EQ just their subs/bass.

I suggest going out and listening to different setups, speakers, rooms, etc and what your perceptions are for everyone one of them. Get some experience just listening.

BTW, I have heard a speaker, with the same enclosure and drivers, but with two different crossovers. Their FR graphs was very similar, but they sounded very different. Very eye (ear) opening. It cracks me up when people ask about a speaker's internal crossover point.

[curtis]

If anybody's interested, I have found and used a very cool site where you can test various things, including different aspects of your hearing:

Free Online Audio Tests, Test Tones and Tone Generators (audiocheck.net) (https://www.audiocheck.net/index.php)

If you like this sort of thing (and have time), it's a lot of fun. And I find that doing the tests can improve one's ability to hear different things, like timing differences.
The problem with that is you assume the playback equipment you are using has accurate reproduction and is calibrated.

[James]

Curtis,

Thanks for the suggestions.

I had that concern about the Behringer - unfortunately I haven't been able to find any analog parametric equalizers. And graphic equalizers don't allow for such precise adjustments, so I'm not sure that I'd be happy with results from those. Have you used any analog graphic equalizers? If so, what's your experience been like with them?

This isn't my first system - I've gone through a number of different speakers. And I used to regularly go to our local bricks and mortar audio dealer to listen to things there. Sadly they went out of business - hard to compete with the internet I would guess.

Also, when traveling on vacation once, I was able to listen to the Wilson Audio Watt/Puppy speakers at a dealer in Ohio - that was a treat! The whole system cost about $75,000, and that was a long time ago.

I grew up listening to music, both live and recorded, and it's always been important to me. About 20 years ago, I picked up a Stereophile magazine, and got interested in audio stuff. Some of what I learned helped to explain things I'd heard years earlier, but without an understanding or vocabulary to describe them.

That makes sense to me about the crossover, and it's why I'm interested in Vandersteen speakers and how they compare to Ascend. When I listened to the 1C at a dealer with my test tones, I could hear some lumpiness right around the crossover region. I don't hear that with my Towers - I figure that's because of the higher-order crossovers.

It would be great if I could easily switch out the Towers for the Vandersteens, and listen to them at home for a while with my equipment/room/music, but that's not really possible. So if anybody here has experience with Vandersteen speakers, I'd be grateful if they would share it.

[James]

I don't understand that comment - what do you mean by "calibrated"?

[James]

In retrospect, even though they were very impressive, I'm not sure I'd like to have the Wilson speakers. They really brought the music into the room and up close to the listener, and I prefer to have the music across the room from me - less immersive.

And I'm not sure that they didn't exaggerate the soundstaging a bit.

At the time, I didn't have my test CD, so I just listened to some music - Oscar Peterson on Marian McPartland's Piano Jazz show and a couple of other things.

[curtis]

I don't understand that comment - what do you mean by "calibrated"?
Just accuracy...that it is accurately reproducing a tone at a given level.

How do your PC speakers sound to you...or headphones that you use?

[James]

Just accuracy...that it is accurately reproducing a tone at a given level.

How do your PC speakers sound to you...or headphones that you use?
They sound limited of course, especially in the bass region.

But they seem fine for those tests I mentioned, especially tests on how small a volume difference or timing difference you can hear.

I wouldn't expect to hear a full 20-20kHz sweep on them, by any means.

We could use the I-Pad and do the tests while listening to the Towers, which would be interesting - if I can convince my wife on that one :-)

[curtis]

They sound limited of course, especially in the bass region.

But they seem fine for those tests I mentioned, especially tests on how small a volume difference or timing difference you can hear.

I wouldn't expect to hear a full 20-20kHz sweep on them, by any means.

We could use the I-Pad and do the tests while listening to the Towers, which would be interesting - if I can convince my wife on that one :-)
Wouldn't you be interested in the frequency response of those speakers before you did the test?

Before any test like that, wouldn't you want to know the playback equipment was accurate?

You want to correlate what you see FR graph to what you hear, but you take a test on your hearing without knowing the FR graph of the output device?

I thought you sent test tones to your Towers already. How did you do that?

[James]

Wouldn't you be interested in the frequency response of those speakers before you did the test?

Before any test like that, wouldn't you want to know the playback equipment was accurate?

You want to correlate what you see FR graph to what you hear, but you take a test on your hearing without knowing the FR graph of the output device?

I thought you sent test tones to your Towers already. How did you do that?
It doesn't seem necessary, especially for the tests I mentioned. The one testing audibility of volume changes uses an a440 tone, which is clearly played by my computer speakers. I'm not testing my hearing with it the way that an audiologist does - I go to one of those every year for that. I guess we have to trust that their system is accurate.

You should really check out the website - you might find it interesting. And you seem to have plenty of knowledge about your system to feel confident it's reproducing the various tests accurately.

I have the Stereophile Editor's Choice CD, and it has the Chromatic Scale test on it. This streaming stuff through an I-pad is new to me - I haven't had that capacity for very long. Over the last few years of owning the Towers, I've moved them around a lot, changed toe-in, listened with and without the grills. Every time I make a change, I listen to music for a while to see if I hear any changes, then play the Chromatic Scale test.

[James]

It's kind of a microcosm of the larger issue, isn't it? How do you know that your system is accurately reproducing a 1db volume level test?

Get a mic and the REW software, and you can learn a lot, for sure. But how do you know if the mic is accurate?

It seems like there's a strange sort of problem there - you can only gauge accuracy by measuring, but how do you know the tools you're using to measure are accurate? By using other tools, but then...

The Chromatic Scale test on my CD is supposed to be a set of evenly spaced 1/12 octave tones at the same volume level, but I just sort of trust that's true - I don't know for sure there aren't some errors in it, either as far as tonality or volume.

[curtis]

It's kind of a microcosm of the larger issue, isn't it? How do you know that your system is accurately reproducing a 1db volume level test?

Get a mic and the REW software, and you can learn a lot, for sure. But how do you know if the mic is accurate?
That is why you get a calibrated mic.

A popular one is the UMIK-1 which comes with a calibration file. They are each tested before shipped to the retailer and the corresponding calibration file is available online. That calibration file can be loaded into REW. It can also be used with Dirac.

I believe the dB meter we used during our demo session, Dave sends out once in a while to get tested/calibrated.

You have to have the right equipment to do the right job.

[James]

Well, yes, that's the one I was looking at - it's very reasonably priced, and can be used easily with the REW software.

Can you post a picture of the frequency response of the one you have?

[curtis]

Well, yes, that's the one I was looking at - it's very reasonably priced, and can be used easily with the REW software.

Can you post a picture of the frequency response of the one you have?
Of my mic?

I use Anthem's ARC/Genesis, and their own mic comes with the receiver/pre-pro. The software takes the serial number and downloads the correction file.

Have you looked up measurements of the Wilson Audio WATT/Puppy?
You can find a lot of measurements here:
https://www.soundstagenetwork.com/index.php?option=com_content&view=article&id=16&Itemid=140

Here is Stereophile's for the WATT/Puppies
https://www.stereophile.com/content/wilson-audio-wattpuppy-system-5-loudspeaker-measurements
https://www.stereophile.com/content/wilson-audio-specialties-wattpuppy-system-8-loudspeaker-measurements

Do they sound like the graphs you see?

Stereophile also has Vandersteens.

[James]

It's been a number of years since I heard the Wilson's, and I don't remember which version of the W/P I heard, or anything very specific about the frequency response.

The main thing that I do remember is the "larger than life" soundstaging, which I believe may have been a bit exaggerated.

Oscar Peterson's piano sounded huge, and the imaging on a folk recording seemed funny somehow.

It was impressive, no question about that, but I didn't like it that much. And, of course, they're ridiculously expensive for normal folks like us.

Stereophile does have reviews of Vandersteen speakers with measurements, but not of the ones I'd be interested in. And there was an issue about the magazine not measuring one of the Vandersteens on the optimal axis - Richard Vandersteen sent his measurements with a comment, and his were much more neutral looking. If I remember right, he claimed something like +/- 2db over a broad frequency range (that's a tighter spec than the usual +/- 3db).

[James]

I actually went to the store, a smallish dealer in Columbus OH, to listen to the Thiel 1.6 speakers, which I'd read about in Stereophile. But when I got there, a fellow in the store asked if I wanted to hear their top "flagship" system - of course I said yes please.

So I went up to the top floor, and listened to the Wilson's with some Spectral monoblock amps, if I remember right. It was very exciting to listen to an expensive system that I'd never buy, and I was very impressed with it.

After that, the Thiels sounded small and just sort of ok.

It was only later while reflecting on what I'd heard that I came to the conclusion I wouldn't want the Wilson system in my house.

I didn't have my test cd yet - it would have been very interesting to run that on the system and see what I heard with it.

[James]

We all have our preferences - in the end, for me a perfect system would be one that I didn't hear at all, and that just reproduced exactly what's on the recordings I play through it - no colorations, nothing added and nothing subtracted.

[curtis]

We all have our preferences - in the end, for me a perfect system would be one that I didn't hear at all, and that just reproduced exactly what's on the recordings I play through it - no colorations, nothing added and nothing subtracted.
The big question is then how do you know you are hearing what the recording engineer heard? That person is in a different room with different equipment.

[James]

I don't know if you need to know that.

They may be using equipment that's not neutral to highlight aspects of the mix and make their job easier. There was a Yamaha monitor, I think it was the NS-10, that was widely used in recording studios, and it was widely understood that it was far from a neutral speaker.

But how do you know you're hearing exactly what was recorded is a good, valid question.

[curtis]

I don't know if you need to know that.

They may be using equipment that's not neutral to highlight aspects of the mix and make their job easier. There was a Yamaha monitor, I think it was the NS-10, that was widely used in recording studios, and it was widely understood that it was far from a neutral speaker.

But how do you know you're hearing exactly what was recorded is a good, valid question.

The bottom line is there is a lot of information you need to know to make that determination. What about soundstage and imaging?

[James]

Right, that's a tough one to know.

If you had the recording engineer who recorded what you're listening to, you could ask them if it matched how they recorded it, but that's unlikely to happen.

But it might be easier to decide that a system is inaccurate. If I listened to 20 CDs from different genres and different times, and they all had what sounded like an overly big/overly "3D" presentation, it might be reasonable to conclude that the system wasn't just reproducing the original recordings.

Because it's unlikely they all would have had that sort of imaging/soundstaging.

Jeez, I forgot about this - I've been so focused on frequency response and room modes with the Towers, but that Editor's Choice CD from Stereophile has a number of music tracks, all recorded by John Atkinson, and he describes a number of things about the recordings, including imaging/soundstaging.

So you could listen to those tracks, note what you heard, and then read his descriptions. That should give you a lot of information.

Stereophile Editor's Choice - Test CD | Shop Music Direct (https://www.musicdirect.com/accessories/stereophile-editors-choice-test-cd)

[curtis]

Coming back full circle, you can't know really what a speaker sounds like by looking at the FR graph, just maybe a general idea.

[racrawford65]

Measurements are nice / interesting, but I agree with Curtis' statement. In the end, it's what sounds best to you in your room, regardless of measurements, specs, etc., IMO.

[James]

This is funny - everything has shifted 180 degrees.

I'm the one who trusted my hearing, and I was told that I shouldn't, and that measurements like the REW software offers are more reliable as far as what's really happening.

Now you're saying what you hear is more important, regardless of measurements.

Then why get the REW software and use it to measure and equalize stuff? You can just as easily just get an equalizer and do it manually, the way I'm probably going to do it.

Personally, I'm very interested in the correlation (or lack of it) between measurements and what I hear. If/when my test tones correlate with what I hear in music it makes me happy. If I got the REW software and it showed a vastly different response than what I hear, it would bother me.

As an aside, the REW software looks very complex to me, and like it would take quite a while to learn how to use it correctly and/or well for one's particular situation.

[curtis]

This is funny - everything has shifted 180 degrees.

I'm the one who trusted my hearing, and I was told that I shouldn't, and that measurements like the REW software offers are more reliable as far as what's really happening.

Now you're saying what you hear is more important, regardless of measurements.

Then why get the REW software and use it to measure and equalize stuff? You can just as easily just get an equalizer and do it manually, the way I'm probably going to do it.

Personally, I'm very interested in the correlation (or lack of it) between measurements and what I hear. If/when my test tones correlate with what I hear in music it makes me happy. If I got the REW software and it showed a vastly different response than what I hear, it would bother me.

As an aside, the REW software looks very complex to me, and like it would take quite a while to learn how to use it correctly and/or well for one's particular situation.
What I have been saying is you can not reliably say how a speaker will sound by looking at measurements.

What is not bad idea is to try and correlate what (an anomaly) you hear with measurements, not the converse.

The use of REW is to work on issues with the room, not the speakers. You buy speakers for how they sound, the room affects that sound. That is why it is called Room EQ Wizard.

BTW...the Ascend demo room uses NO EQ. It is treated though, and doesn't have any noticeable echo.

[James]

All of the equalization that REW applies are to the speakers, if I understand it right. It's not doing anything to your room.

Room treatments treat the room.

There's been some debate about whether or not equalizers work well for room issues, with some people (Richard Hardesty, for one) advocating room treatment instead (and buying speakers with neutral anechoic measurements to start).

Two of the things I'd plan on using an equalizer for seem like it should work very well - raising the bottom end a bit to compensate for Fletcher Munson, and taming a slight rise in the speaker response. I'm not as sure that it is the best solution for room modes.

But I don't have the space or budget for more room treatment - we already have two GIK Tri-traps in a corner behind the speakers, but there aren't really any other good places to put things like that (it's our living room).

[racrawford65]

Further to Curtis' comments on REW, it can be useful to help with speaker/subwoofer positioning. For that, Quick Measure in ARC Genesis also works.

[curtis]

All of the equalization that REW applies are to the speakers, if I understand it right. It's not doing anything to your room.
You adjust the speakers to compensate for the room.

Right from the first paragraph on the Room EQ Wizard page:
"The audio measurement and analysis features of REW help you optimise the acoustics of your listening room, studio or home theater and find the best locations for your speakers, subwoofers and listening position."

[James]

Yes, I understand that.

As I said, there's some debate about how well that approach works.

I've used the test cd I mentioned to help with finding the best place and orientation of our speakers. It's been very helpful for that.

If you can't hear the problems, but find them with measuring tools, how likely is it that you'll hear any potential improvements? And if you do hear the problems, then you can use an equalizer to improve things and you should hear those improvements.

I think you're trying to make an extremely complex subject seem simple, and it's not simple.

[curtis]

Further to Curtis' comments on REW, it can be useful to help with speaker/subwoofer positioning. For that, Quick Measure in ARC Genesis also works.
You beat me too it.

On another note, I re-ran ARC Genesis a couple days ago. My room had gone through some changes over the years, but nothing that I thought warranted re-measuring. I was wrong, it made a difference.

[James]

Had you heard anything different before you measured it again?

[curtis]

I think you're trying to make an extremely complex subject seem simple, and it's not simple.
LOL....that is what I think you are trying to do.

[curtis]

Had you heard anything different before you measured it again?
I had always known that if you make a change to the room, you should run the routine over again, but I was lazy.

All the talk on the forum about measuring, and I also thought I noticed my bass/mid-bass seemed a little "sloppy" in the last few months, but I wasn't sure why. Combined to the two, I decided to break out the mic, mic stand, and long USB cable and do it again.

This morning, I realized the coffee table is not where I want it to be. I will leave it for now, but move it later, and then re-measure. I haven't put the mic/stand away yet, but the last two nights I have been having a good time just listening to music (after the Dodger game).

[davef]

This is funny - everything has shifted 180 degrees.

I'm the one who trusted my hearing, and I was told that I shouldn't, and that measurements like the REW software offers are more reliable as far as what's really happening.

Now you're saying what you hear is more important, regardless of measurements.

No, this is a completely wrong conclusion and I believe you know this. You can trust your ears to determine what sounds good to you, that is it. No different than trusting your taste buds to determine if you will like the taste of that chocolate bar you bit into.

Measurements determine the actual performance and accuracy of a speaker. They are an excellent visual representation of speaker performance. No different in needing the actual ingredients and percentages of the ingredients to determine if that chocolate bar is healthy for you.

You have been trying to trust your own hearing with sine-wave test tones to determine the measurements of a speaker.


Interestingly, I find the towers response in my living room doesn't seem to follow the PIR that you posted - it's much more even than that.

Would you trust your taste buds to determine the ingredients and the amount of those ingredients? Of course not, not even professional chefs can do this.

When you have accurate measurements of a speaker that you like, if a different speaker measures very similarly, you can make the assumption you will like that speaker too. If there are aspects of a speaker you don't like - after properly measuring the speaker and playing with equalization, you can then take steps to improve what sounds better to you. Or, if you are after absolute accuracy, you can use the measurements to apply EQ to help tame room issues or correct problematic response issue with the speakers.

Do not assume accuracy and what sounds good to you are one in the same. (they are never related)




Then why get the REW software and use it to measure and equalize stuff? You can just as easily just get an equalizer and do it manually, the way I'm probably going to do it.

Because then you haven't actually learned anything and have no baseline of determining what frequency response sounds good to you.


Personally, I'm very interested in the correlation (or lack of it) between measurements and what I hear. If/when my test tones correlate with what I hear in music it makes me happy. If I got the REW software and it showed a vastly different response than what I hear, it would bother me.

Trust me when I say this, the measurements you will see will indeed show a VASTLY different response than what you think you hear. You have already proved this in a previous statement you made, which I quote above. You won't be happy, but you will be on your way to learning more.

[James]

Ok, I can get that idea - it's a combination of measurements for accuracy and your hearing for your tastes.

I do think that some people have more refined palates than others and they can tell more of what's in a dish than others. Professional chefs, for one. And just like you can damage your hearing by listening to very loud sound for too long, you can probably dull your palate by eating a lot of salty, spicy food.

I'm not really trying to determine the REW measurements of our speakers by listening to them. It's more like eating something and thinking there's not much salt and then finding out there's a ton of salt in it. While I don't think I could tell exactly what ingredients are in a dish in exactly what proportion, I'd like to think I can taste that there's a lot of salt in it.

I know from personal experience with reducing my salt intake that I now can more accurately taste when there's a lot of salt in a dish. And I value that, as I value testing my hearing and being able to hear small volume or timing differences.

That's interesting - is it really true that nobody likes accuracy? I believe that I want an accurate system - that's one reason I bought the Ascend speakers, which show good accuracy in the measurements section. And it's why I bought my other components, which had reviews showing decent neutrality.

Sorry, I may have misunderstood that part - why would accuracy and what sounds good to me (or anybody else) never be related?

[James]

I'm really torn about the REW.

On the one hand, it appeals to my perfectionism and wanting to hear an absolutely neutral/accurate reproduction (even if it turns out I prefer some imperfections).

But on the other, if there are massive imperfections in my current set-up, then they'd require massive corrections, and I also have some purist tendencies.

And, from what I've read, it's a very complex program and I think it would take a long time and a lot of effort to be sure I'm using it correctly.

The worst of both worlds would be to find out there are massive imperfections, but then not be able to fix them.

[James]

Also, just intellectually, I find it hard to reconcile that there might be large deviations from neutrality that I don't hear.

I can hear 1db volume changes on a test (nothing smaller).

On the test tones, I don't hear huge changes in volume. If they're there, why wouldn't I be able to hear them?

[davef]

And what exactly do you consider to be a neutral in-room response?

[James]

I feel like that's a trick question somehow :-)

Generally, I consider neutrality to be an accurate representation of the source signal. So if the test tones are recorded at the same volume level, I should hear them that way.

Is there another definition of neutrality that would apply to this question?

[davef]

I feel like that's a trick question somehow :-)

Generally, I consider neutrality to be an accurate representation of the source signal. So if the test tones are recorded at the same volume level, I should hear them that way.

Is there another definition of neutrality that would apply to this question?

So, in other words, you are stating that you consider a neutral in-room response to be listening to your test tones at your main listening position such that you hear all of them at the same amplitude, correct?

[James]

That's hard to answer yes or no.

I would be hearing a neutral accurate reproduction of the source that way, I think.

It's possible because of Fletcher Munson that the measured in-room response could be tilted up in the bass but I wouldn't hear it that way.

So the measured response wouldn't be neutral, but I would hear it as neutral. Especially at lower volume levels, which is how I generally listen.

If the REW measurements showed more volume in the bass than I was hearing, that would be in-line with my understanding of the Fletcher Munson curve. It's really the 15db peak-to-trough graphs not in the bass that throw me - could I really not be hearing that sort of thing?

One of the things I'd expect to change if I had an equalizer is that I'd boost the bass levels until I could hear it as flat down to about 40Hz, to be able to fully hear the low E of the double bass.

[davef]

That's hard to answer yes or no.

I would be hearing a neutral accurate reproduction of the source that way, I think.

It's possible because of Fletcher Munson that the measured in-room response could be tilted up in the bass but I wouldn't hear it that way.

So the measured response wouldn't be neutral, but I would hear it as neutral. Especially at lower volume levels, which is how I generally listen.

I am just repeating your own statements James. It is a very simple question, for which I didn't mention anything about equal loudness curves or measurements.

Basically, you keep repeating yourself that what you consider to be a neutral response in a speaker, for your own ears, is you hearing equal amplitude in your test tones. (assuming those test tones were actually recorded at the same amplitude). James, there is nothing wrong with trying to achieve what YOU consider to be a neutral response in a speaker placed in your room. However, the definition of a neutral speaker is not based on one's hearing, and for very good reason.

Further to the point, please see the attached measurement. If listening in your room, which of these frequency response measurements would you expect to sound more neutral to your ears? Black measurement or Red measurement?

2361

[James]

What I want is to hear a neutral, accurate reproduction of the source signal. I want to hear what's been recorded, without additions or subtractions, as much as possible.

The question we're discussing is how to achieve that, isn't it? For various reasons, it seems that REW measurements show some large-ish deviations from that at a normal listening position.

I know the "right" answer to that question is the red line, from some reading. And I can understand (at lower listening levels) why the bass boost sounds neutral. But I don't really understand why the treble should be reduced like that - the Fletcher Munson curve shows decreasing sensitivity at lower levels to the treble as well as the bass.

And, I can see with some of the REW measurements that the sort of average curve is sloping downwards, which would fit the red line. But they also show peaks and dips that I would think should be audible.

[davef]

What I want is to hear a neutral, accurate reproduction of the source signal.

Then the first thing you need to do is to immediately stop using your ears to determine this. An accurate reproduction of the source material can not be determined by listening to test tones to try and determine how neutral the response of a speaker is in your room, nor can your own hearing.


The question we're discussing is how to achieve that, isn't it? For various reasons, it seems that REW measurements show some large-ish deviations from that at a normal listening position.

To achieve an actual neutral response in your room, you need measurement software James. REW shows large-ish deviations because those large-ish deviations exist, far worse than you can even imagine by looking at some measurements posted here. Doesn't matter what software or hardware you use.


I know the "right" answer to that question is the red line, from some reading. And I can understand (at lower listening levels) why the bass boost sounds neutral. But I don't really understand why the treble should be reduced like that - the Fletcher Munson curve shows decreasing sensitivity at lower levels to the treble as well as the bass.

What I posted has nothing to do with Fletcher Munson.


And, I can see with some of the REW measurements that the sort of average curve is sloping downwards, which would fit the red line. But they also show peaks and dips that I would think should be audible.

And this goes back to the very beginning of your questions.... You make wrongful assumptions regarding human hearing. It doesn't appear to me that you are actually open to learning anything and you are set in your ways, which is fine. But I am signing off on this until you actually take the proper steps to learn more. You don't even need to use REW, just read some books on the subject.

In summary, the goal of a neutral speaker is to reproduce the source material neutrally, keeping deviations from neutral linearity to a minimum. Determining the actual accuracy of a loudspeaker can only be done with proper anechoic measurements. Once you place that neutral speaker in a room, where it is subject to room modes and reflections, all bets are off. At that point, how the speaker reacts in your room is characterized by the directivity of the speaker.

To further complicate things, humans do not "hear" neutrally. When you think you hear one of your test tones at the same amplitude as another frequency, you are then taking into account the response of your room, the response of your own hearing and the directivity of the speaker. If you were to actually measure this, what you hear as being neutral, would absolutely not "look" neutral to you but would to many people on this forum due to having experience in this subject.

For example, go online and listen to some white noise. White noise is a flat frequency response, or equal energy for each frequency. It is going to sound thin and bright to you, definitely not neutral to your ears. Now go listen to some pink noise, it will sound far more neutral to you, but is anything but neutral with regard to the frequency response.

With regard to achieving your goal in your room, you already have very neutral speakers. The next step is to deal with your room, and you can't do that without measuring its acoustics. Once you have that information, you can then work on room treatments or add EQ to deal with room issues. Once measurements confirm that you are achieving a decently neutral response, you will need to see an audiologist, have your hearing tested from 20kHz to 20kHz. They will provide with you an estimated frequency response of your own hearing. You can then use the inverse of this to add that as a target curve for your equalization... At that point, you will have achieved your goal.

[racrawford65]

Well said, Dave.

[James]

Ok.

Thanks for the information in the last 4 paragraphs. I'll try to understand and digest it so I can get closer to my goal. The last step seems like it would be unfair to my wife, who also listens to a lot of music - I'm not sure I'd take that one.

I have in fact done a fair amount of research since we began the discussion in order to learn more about this topic - maybe that hasn't been apparent in my comments somehow. As you've said, it's a very complex subject. Sometimes it's hard to find the right level of information if you want to learn enough, but not have to become an expert in something complex.

The apparent mismatch between what's seen and heard is counter-intuitive to me and hard to reconcile, but I'll try to work on that.

I appreciate your taking the time to try to educate me on the topic.

[James]

Back to REW - it seems like very complex software.

For those that use it, is there a reasonably straightforward way to measure the in-room response, and correct it to a neutral one?

I've looked up a lot about it, but haven't found anything like that yet.

Something like "REW for Dummies" might be good for me.

[racrawford65]

AVS Forum - Austin Jerry has a great guide. In case you can't find it, I've copied what I think is most recent to my dropbox. There's also a great thread on REW setup & use on AVS, link below.

https://www.dropbox.com/s/88xevdv15v9ieln/REW%20101%20HTS%20Current%20Version.pdf?dl=0

https://www.avsforum.com/threads/simplified-rew-setup-and-use-usb-mic-hdmi-connection-including-measurement-techniques-and-how-to-interpret-graphs.1449924/page-1622#post-61603823

[racrawford65]

One more REW resource...

https://www.roomeqwizard.com/#

[James]

Thanks - that second one looks like it might be at a good level for me.

Do you think that using an equalizer like the Behringer would create problems, since it has AD/DA conversion and my preamp/amp are both analog?

Man, 132 pages on "Getting Started" - that's going to take a while to get through.

[racrawford65]

Don’t know, honestly. Guess it depends on quality of components in it and how well they implement. My 2 channel is essentially all analog, although I do have a SugarCube in my vinyl chain (generally direct bypass unless an album doesn’t clean up satisfactorily) and a dSpeaker in the sub line for bass EQ

[James]

That makes sense.

I'd guess that the DAC in my CA is better than the one in the Behringer, and I'd hate to mess that up. But the Behringer is a 24/96 DAC - maybe that would be good enough.

I'll have to call Behringer, ask them some questions, and maybe just try it, if there's a decent return policy in case it doesn't work out well.

After we do the next house projects, one of which is somewhat more expensive now than it would have been last year (covid/inflation).

[Mag_Neato]

There's also some very useful video tutorials on Youtube on REW, namely from Home Theater Gurus, but there are a few others if you do a search. The benefit of using REW is that once you have the room measurement it lets you create EQ curves in it, which can then be saved into a file for uploading into a DSP processor, such as the MiniDSP units. I use a MiniDSP 2x4 HD to EQ/calibrate my dual subs. Nothing above 80hz comes out of it so my mains are not getting a reconverted signal from it, just the subs.

[James]

I've looked at a lot of stuff about REW, but most of it seems very hard to understand.

Just now I found a very good basic guide on getting a minidsp UMIK set up with REW, and how to get started measuring - in case anybody else might find it helpful:

UMIK-1/2 setup with REW (minidsp.com) (https://www.minidsp.com/applications/acoustic-measurements/umik-1-setup-with-rew)

Unfortunately it stops there, and I haven't found a comparable guide on how to then use REW to generate corrective eq curves (and which curve would be the most accurate/neutral).

Since my set-up is old, simple 2-channel, I could set the corrections manually on the Behringer after that.

I found something from somebody who had a PhD in Physiological Acoustics that he had a hard time understanding and using REW - that made me feel a bit better about my trouble with that.

[James]

Our local library has a recording studio, which is pretty cool, and you can hire an engineer.

Maybe the best thing for me to do, when the time comes, is to find one with experience with REW and hire them to help me use it.

The program offers a lot for free - I wouldn't mind paying a bit to use it correctly.

[racrawford65]

It isn't really too difficult to measure with REW...plenty of folks on AVS that will answer questions...but it's your money..

[James]

Seems awfully complicated to me.

I found a local engineer who'd be willing to help me with it for a reasonable hourly rate, which is exciting. As long as it doesn't take too long, it should be fine.

The first thing would be to get the equalizer and play with it - it would be about 1/2 the financial investment of getting everything. It should be easier to connect and use than the software. And if it colored the sound in some way we don't like, we could probably return it.

Also, I'd be very interested to see what the REW software shows, if we've gotten the system equalized to where it sounds very neutral to us - how close or far we are from actual neutrality and/or the REW recommendations.

The system sounds very good to us right now - there aren't that many changes I'd plan to make with the Behringer.

The possibility of getting this stuff and improving our system should get me through a couple of big house projects - something to look forward to :-)

[James]

Well, I finally found a guide to using REW that I might be able to use on my own:

Room correction with REW using PEQ (mehlau.net) (https://mehlau.net/audio/room-correction-peq/)

It might not be the ideal way to use it - there are choices to be made along the way that I don't fully understand. But I can at least try it on my own first and see how it works.

[James]

Next step seems to be picking a "house curve".

Bob Katz suggests flat from 20-1000Hz, then a straight line down to -6db at 20kHz will sound neutral to us at the listening position.

He's a very well-respected mastering engineer, so I'll start with his suggestions.

[curtis]

In all of this, how does the system sound to you now, before you even get started?

You don’t have a baseline of your own thoughts before being influenced by anything.

REW is the first step no matter what you do afterwards.

[James]

The system sounds very good, both with test tones and with music.

It seems to be generally neutral, with a few deviations. It rolls off a bit earlier in the bass than I'd like, around 45-50Hz (I'd like it flat to 40Hz, so I can fully hear the lowest note on a double bass). There are at least a couple of room modes that create extra resonances. And the 500-1000Hz region seems a bit elevated to me.

It's the best sound we've ever had, and it sounds very good with different genres of music. We listen to mostly acoustic music, but there are synthesizers on some of the stuff we listen to as well.

[James]

I'd love to get everything right now and use it, but that's not possible.

And I want to make sure that the Behringer equalizer doesn't sound weird to us because of the internal AD/DA converters first, before getting a mic and REW.

So it'll be two steps, and I'll have to be a bit patient about it.

[curtis]

I'd love to get everything right now and use it, but that's not possible.

And I want to make sure that the Behringer equalizer doesn't sound weird to us because of the internal AD/DA converters first, before getting a mic and REW.

So it'll be two steps, and I'll have to be a bit patient about it.
What is holding you back from getting REW up and running? No PC or mic?
Before all else, you need to know what is going on in your room.

The Behringer would be least worry on my list. You could/should do things with the room/positioning, even just to experiment, before employing any kind of EQ.

[James]

Well, I'd have to buy a new laptop (REW doesn't work on an I-pad) and a mic. That would be about half of the total cost of the project.

I've spent a few years moving the speakers, and experimenting with different amounts of toe-in, and right now they sound the best they've sounded yet. I've gone from very close to the back wall to where they are now, over 3 ft. away from it (this is already further into the room than would be ideal, so I don't think we could bring them out any more).

The idea of finding out what's wrong but not being able to correct it seems like it would be frustrating to me - that's why I would want to get the equalizer first.

[curtis]

Treating with an EQ is the last thing you should do. You don't even know what is "wrong" and there may not be any real issue.

Find out what you want changed. Deal with it with placement and room treatments if you can...or even different speakers

If you still are not satisfied, then you EQ to your liking.

You are going down the EQ rabbit hole without even knowing what you want changed.

My 2 cents. Good luck.

[James]

As I said, there are a few things I hear that I'd like to improve a bit. And my wife hears at least a few of the things I hear, so she'd also hear improvements (she just doesn't care as much as I do about it).

I've exhausted the placement possibilities, and the two Tri-Traps are all we can reasonably use in our living room. They did make an audible difference - reduced some of the room mode resonances and cleaned up the sound a bit higher up as well.

After going through more speakers than I can easily list, and loving the Towers, the last thing I'd want to do is get different speakers (which would be much more expensive than what I'm considering).

Thanks for your advice and good wishes.

[Alleric]

Just bear in mind that there will always be room modes unless you have a completely absorptive anechoic chamber. Roland/Boss's room at their Japanese headquarters is apparently unnerving in how quiet it is. In a consumer space, unless you're willing to treat it to take it out of the consumer and into the professional realm, you'll always have resonances, you'll always have nulls. All you can really do is just move them in the frequency domain by moving speakers or moving you. For room treatments, the most immediate bang for the buck to help things above midbass (say, 250-300hz) are broadband absorption on the side walls to kill first reflections. Down below 250-300hz you're looking at bass traps.

Psychoacoustically, the human ear is more tolerant of nulls than it is of peaks, and the tolerance is modulated by the steepness of the Q of the null or peak.

If you're trying to set up a totally locked-in stereo image, strict two-channel, the first thing I would do is solve the time domain problem: get the speakers equidistant to the main listening position first. If you have no room correction in your processing, just measure it with string and tape. Once that's done, if the speakers you're using are going to throw a nice 3d image, they should be doing it.

Once that's solved, speaker rotation for toe should be evaluated. This is mostly just to EQ tweeter behavior. Once that's solved, scan with REW and look at the "All SPL" tab. At this point you're about to open a pandora's box. You're looking behind the curtain. Some of the things you'll see there you can fix, some of them you can't. Adjust whatever EQ mechanism you have, rescan. Rinse, repeat. Welcome to tweaker's paradise. Now you can read up on RT60 measurements, how to manipulate their decays, realize you don't need a second career and just be happy getting basic time and frequency response nailed down as much as you can.

Sometimes just getting the distance right on a pair of big towers and throwing an amp at them is nirvana. :)

Anyway, changing EQ with no way to measure is configuring blind. Being able to measure with no way to change it gives visibility into the potential issue with no power to change it. I would find either of these options really annoying. :)

[James]

Hi Alleric,

Thanks - I actually went ahead and got a mic and REW. Short version - it's very interesting, many of the things it found I had already heard, but not all of them, and it improved the system in audible ways.

We had some corner bass traps before, and the speaker positioning was very good, based on years of moving/listening/using a test CD/repeat.

I used the old B&K house curve, based on theoretical and audible grounds, after a lot of experimenting with others, like Bob Katz and Harman. The simpler implementation with fewer points worked better than a more complex one:

20 Hz 0, 200 Hz -0.5, 2000 Hz -3.0, 20000 Hz -6.0

None of the corrections were very wide bandwidth and the cuts weren't very large, confirming my sense that our system and the speakers were already very very good.

There are a couple of things REW didn't measure (I'm not sure why) so I'm doing some final tinkering by ear with those.

With the Behringer equalizer on, there is a very slight decrease in sound quality generally, but it's really only noticeable in direct comparisons - kind of a more crystalline sound, a bit harder and cooler. And the improvements in frequency response outweigh that downside.

It was fascinating/fun/frustrating/confusing, and ultimately very satisfying.

[James]

I should add that the unsmoothed measurements had sharper peaks and dips than I hear - it was the psycho-acoustically smoothed graph that looked a lot like what I hear, but not completely.

I've always heard a couple of things that my wife doesn't hear, and it turns out I was right about them - that was nice.

But there are also a couple of things neither one of us heard that show up on the smoothed graph, which is interesting/puzzling.

And even the unsmoothed graph isn't bad, compared to some I've seen - the overall frequency response is about +/-7db, so no 20db peaks or dips.

[Alleric]

There are a couple of things REW didn't measure (I'm not sure why) so I'm doing some final tinkering by ear with those.


What, specifically, were the unmeasured things?



I should add that the unsmoothed measurements had sharper peaks and dips than I hear - it was the psycho-acoustically smoothed graph that looked a lot like what I hear, but not completely.


Considering that the peak and null "rhythm" accelerate massively as you go up in frequency, this is entirely normal. People often freak out when they see the initial unsmoothed REW spl curves, but this is commonly because they don't know how the math works concerning frequency and octave. Western diatonic music theory contains 12 semitones within a given octave, inclusive. Down between 10-20hz, each hertz almost contains a pitch on its own. That mess you see up north of 2k? Each of those peaks and nulls is a partial of a given pitch.

The psycho-acoustic graph is pretty much the closest representation of what your system is doing compared to how you year it. For troubleshooting and configuration of a room, though, I highly suggest using Variable smoothing. This does pretty much the same thing, but in spots where you jaggies really matter it will amplify them graphically for you. Helps to localize issues you should actually care about.





I've always heard a couple of things that my wife doesn't hear, and it turns out I was right about them - that was nice.

But there are also a couple of things neither one of us heard that show up on the smoothed graph, which is interesting/puzzling.


I can solve this one for you: your ears hear your room differently than a calibrated (or uncalibrated!) microphone. We want our rooms to play "flat" or match one of many house curves (Tool, HK, whatever). Your mic can show you all day that you hit that correctly with your config, but you will always hear it differently. Human ear, it does not have flat sensitivity to the frequency range humans can hear. What makes it worse is that the curve we actually hear varies at different SPL. This is is where the dynamic EQ built into the DSP of Denon receivers is solid gold. You tune your room at or near reference SPL, which is way louder than most people listen to things, then let Dynamic EQ be that magical DSP bridge to your human ears.



And even the unsmoothed graph isn't bad, compared to some I've seen - the overall frequency response is about +/-7db, so no 20db peaks or dips.

You're telling me your All SPL graph, unsmooth, shows no peak or null greater than 7db? Especially up past 2k? Peaks, that can happen. Nulls? If that's legit, buy a mixing desk and and charge mastering time per hour. :)

[James]

There's a room mode at about 74Hz, and another one at about double that, that the measurements didn't show. But they're definitely audible on plucked double bass and plucked lute/guitar.

Yes, I used variable smoothing for the eq filters - that's what REW recommends.

I'm not sure I understand that part yet (I understand the bit about different volume levels). If our ears hear differently from mics, which I believe, then why use the mic to set the filters? Are we over-correcting somehow for things we don't need?

Yes - I used a moving mic technique right about at our listening position, going across our couch from one side to the other, which covered both speakers to about the outside edges. I took about 160 measurements and used the average forever function. The unsmoothed graph is very smooth from 2k to 20K, with a flat to gently sloping downward curve down to about -5db at 20k.

There were no recommended eq settings above 1K, with the B&K curve.

Nice idea :-)

The biggest irregularities are in the 100-200Hz region - there's a -5db cut at 120 and a -5.5db cut at 170.

[Alleric]

I'm not sure I understand that part yet (I understand the bit about different volume levels). If our ears hear differently from mics, which I believe, then why use the mic to set the filters? Are we over-correcting somehow for things we don't need?



Because the mic has no psychology. The mic has consistency. The mic provides reliable validation of changes made. Bear in mind, there is no empirical "correct" or "incorrect" on a frequency response curve. Not everyone likes legit "flat". It can sound very thin, clinical. I happen to not mind that, but I also happen to not mind a 3db house curve. I also happen to not mind a 3db house curve with another 3db slope in EQ below 80hz.

As for over-correcting things we don't need? That's a philosophical question. I can't answer that for you.

Your nulls between 100-200? Those are likely your room jaggies. They are yours and yours alone. Love them. Embrace them. They are your friends forever (unless you treat the ever-loving crap out of your room).

[James]

Right, I understand the personal preference thing for house curves. It's the measurements of the pycho-acoustically smoothed response before any eq that show things we don't hear that are puzzling.

If we don't hear those, maybe we don't need to "correct" them with eq?

The things I mentioned were peaks, with the eq corrections. But there are also some nulls, which I sort of understand aren't correctable with eq. Fortunately we seem to hear peaks more than dips, so it's not a big problem.

It sounded great before, and sounds really great now, both on our test cd and with music - I think I can stop trying to improve things for a while (my wife will be happy about that :-)), at least once I get the room mode corrections done.

I think I'm close with those - and they're bigger corrections than any of the REW ones.