ZOOMING In REW (and a little Dirac)

[marcl]

Feb 24, 2023 13:28:02 GMT -5 marcl said:

Topic will be basic Dirac calibration and verification with REW

I'll make it easier ... here's the Zoom login:


Join Zoom Meeting
us06web.zoom.us/j/86705395587?pwd=Q1ZEcGgzck9uOGZXOUd3YW5ZZDZodz09

Meeting ID: 867 0539 5587
Passcode: 039651

We'll each do a live Volume calibration to demonstrate any differences between systems. Then we can load some previous measurement sets, tak about target curves, and do some REW post-calibration measurements.

Starts in 25min!

[marcl]

Recording of this evenings discussion can be found here. Thanks to Lane, Whit, Forrest and Scott for participating ... and especially Scott for the live Dirac demo.


us06web.zoom.us/rec/share/xVtqhB4cLu2tWck-HjwRXE_GH9EKT_8diGFQQlS0hTAtORm5fzieh6Evi2uxk-PI.xmfCRwQdL0iiraZ3
Passcode: 9p+qy7Wc

[marcl]

We discussed our various room and speaker configurations during the call.  Mine is unusual, to say the least.  So here are two diagrams that tell the story as it stands now.  This was an evolution of course, with the most "unconventional" stuff happening over the past five years.

fbczar  denali  ttocs  lhracing




Also as we were discussing center channel Clarity ... here's my current measurement with Dirac.  The CCR crosses to the two DWMs at 200Hz and then they cross to subs at 50Hz.  C50 speech Clarity, C80 music Clarity




And here's my overall system measurement with Variable Smoothing.

[marcl]

From time to time I try to get more specific information on what constitutes a good range for Clarity values C50, C80 and D50. I almost always get frustrated and give up having found nothing but circular discussions with no specifics. So today I asked ChatGPT specific questions based on the values I measured for my center channel.  The responses were consistent with what I know about the definitions of these parameters, AND it gave straight answers regarding my typical values!  

Is a value of 20db for C50 considered good?

A C50 value of 20 dB would be considered very good, as it indicates a significant level difference between the direct sound and the reflected sound arriving at the listener's ears 50 milliseconds after the direct sound.
A C50 value of 20 dB or higher is typically associated with excellent speech intelligibility, even in large or noisy rooms. However, it's important to note that the specific application and size of the room may affect what constitutes a "good" C50 value. Additionally, other factors, such as the frequency response of the room and the background noise level, may also play a role in determining overall speech intelligibility.

Is a value of 35db for c80 considered good?

A C80 value of 35 dB would be considered excellent, as it indicates a significant level difference between the direct sound and the sound reflected off the surfaces of the room 80 milliseconds after the direct sound.
A C80 value of 35 dB or higher is typically associated with very good speech intelligibility and a high-quality acoustic environment, even in larger or more reverberant spaces.

Is a value of 100% for d50 considered good for speech and music?

A D50 value of 100% would indicate that all of the direct sound and early reflections are reaching the listener's ear within 50 milliseconds, which would be ideal for speech and music intelligibility.
However, it's important to note that a D50 value of 100% may not be attainable in many real-world acoustic environments due to the presence of background noise, room dimensions, and other factors. In general, a D50 value of 0.4 or higher is considered very good for speech and music intelligibility, but the specific requirements may vary depending on the application and room size.

[marcl]

Also related to last night's discussion ... answer from ChatGPT:


What is the difference between minimum phase and non minimum phase?

Minimum phase and non-minimum phase are two types of filter responses that describe how a filter affects the phase of the signal passing through it.

A minimum phase filter is a type of filter that has a phase response that can be derived from its magnitude response alone. In other words, the phase response of a minimum phase filter is uniquely determined by its amplitude response. This means that a minimum phase filter has a smooth and predictable phase response that is free from phase distortions. Examples of minimum phase filters include most passive electrical filters, such as RC, RL, and RLC circuits.

A non-minimum phase filter, on the other hand, has a phase response that cannot be derived from its amplitude response alone. The phase response of a non-minimum phase filter contains additional phase shifts that are not accounted for by its amplitude response. These phase shifts can cause distortion and smearing of the signal passing through the filter. Examples of non-minimum phase filters include some types of digital filters, such as linear phase filters, and some types of equalizers, such as graphic equalizers.

In summary, the main difference between minimum phase and non-minimum phase filters is that minimum phase filters have a predictable and smooth phase response that is free from distortion, while non-minimum phase filters have a more complex phase response that can cause distortion and smearing of the signal passing through the filter.

[marcl]

I mentioned in the Saturday call that I have found an issue regarding sending Bass Management to Large Fronts with Dirac 3.4.4.  This would be the case with Fronts=Large and Center Sub=LFE or None.

Here is the simple proof.  This is very similar to the circumstance I documented in April last year when Dirac 3.3.3 was released.  Except, last year the symptom was +5db added to the BM when it was sent to Large Fronts.  This time, there is NO BM sent to Large Fronts.

The condition only occurs when measurements are made with DL 3.4.4.  If measurements are made with DL 3.3.3 there is no issue, even if filters are recalculated in DL 3.4.4.  I expect from last year's experience that it is not possible to test measurements/projects taken in DL 3.4.4 using DL 3.3.3 ... it won't load them.

[marcl]

fbczar denali ttocs lhracing

I did a Volume Calibration and one measurement today just to compare to what we saw last night.

Noise floor was around 43-45dbC; Sub sweep to get to -17.7db was 77dbC;  pink noise to get other channels to -24db was 62dbC.

When I did actual measurements all the speakers fell in the range of 72-78db, depending on how far they were from my desk where I was measuring.  I was closest to the right surround and since it's full range it peaked at 84db.  None of the measurements were anywhere near sounding loud.

Very interesting that given all the differences in speakers, amps, efficiencies, etc. .... we started with the identical Volume Calibration and then Dirac ended up measuring more than 12db louder in ttocs system.

[KeithL]

I think it's worth pointing out why you're not going to find any specific answers for this...
It's also well worth starting out by understanding that all of those values vary with frequency...
(And, when you're discussing "intelligibility", the values that matter are going to be those for the frequency range of human voice.)

The reason you can't get a single answer is simply that "optimum clarity" and "most pleasing sound quality" don't necessarily coincide.
To pick the most extreme example - a totally dead room, with no reflections at all, will have excellent "clarity and intelligibility"...
And a room where it's "easy to make out announcements on the PA" often sounds horrible for music.
But virtually everyone agrees that a totally dead room (an anechoic chamber) sounds really unpleasant. 
And it also turns out that the way our brains handle delayed sound is nowhere near linear.
We hear delayed sounds that we hear within about 50 milliseconds as "liveness"...
And we hear sounds that we hear after around 100 milliseconds as "echoes"...
And, while "liveness" can actually improve "perceived clarity and intelligibility", "echoes" generally tend to degrade it...
Except that, after some longer delay, our brains are able to actively ignore echoes, and prevent them from interfering with the original signal...
And ALL OF THAT varies with frequency, and level, and with the amount and type of background noise present.
And it also varies from person to person.
And with different content (which has the most energy in different frequency ranges).
So all of that is just a bit subjective.

Feb 26, 2023 11:25:07 GMT -5 marcl said:

From time to time I try to get more specific information on what constitutes a good range for Clarity values C50, C80 and D50. I almost always get frustrated and give up having found nothing but circular discussions with no specifics. So today I asked ChatGPT specific questions based on the values I measured for my center channel.  The responses were consistent with what I know about the definitions of these parameters, AND it gave straight answers regarding my typical values!  

Is a value of 20db for C50 considered good?

A C50 value of 20 dB would be considered very good, as it indicates a significant level difference between the direct sound and the reflected sound arriving at the listener's ears 50 milliseconds after the direct sound.
A C50 value of 20 dB or higher is typically associated with excellent speech intelligibility, even in large or noisy rooms. However, it's important to note that the specific application and size of the room may affect what constitutes a "good" C50 value. Additionally, other factors, such as the frequency response of the room and the background noise level, may also play a role in determining overall speech intelligibility.

Is a value of 35db for c80 considered good?

A C80 value of 35 dB would be considered excellent, as it indicates a significant level difference between the direct sound and the sound reflected off the surfaces of the room 80 milliseconds after the direct sound.
A C80 value of 35 dB or higher is typically associated with very good speech intelligibility and a high-quality acoustic environment, even in larger or more reverberant spaces.

Is a value of 100% for d50 considered good for speech and music?

A D50 value of 100% would indicate that all of the direct sound and early reflections are reaching the listener's ear within 50 milliseconds, which would be ideal for speech and music intelligibility.
However, it's important to note that a D50 value of 100% may not be attainable in many real-world acoustic environments due to the presence of background noise, room dimensions, and other factors. In general, a D50 value of 0.4 or higher is considered very good for speech and music intelligibility, but the specific requirements may vary depending on the application and room size.

[marcl]

Feb 27, 2023 10:59:03 GMT -5 KeithL said:

I think it's worth pointing out why you're not going to find any specific answers for this...
It's also well worth starting out by understanding that all of those values vary with frequency...
(And, when you're discussing "intelligibility", the values that matter are going to be those for the frequency range of human voice.)

The reason you can't get a single answer is simply that "optimum clarity" and "most pleasing sound quality" don't necessarily coincide.
To pick the most extreme example - a totally dead room, with no reflections at all, will have excellent "clarity and intelligibility"...
And a room where it's "easy to make out announcements on the PA" often sounds horrible for music.
But virtually everyone agrees that a totally dead room (an anechoic chamber) sounds really unpleasant. 
And it also turns out that the way our brains handle delayed sound is nowhere near linear.
We hear delayed sounds that we hear within about 50 milliseconds as "liveness"...
And we hear sounds that we hear after around 100 milliseconds as "echoes"...
And, while "liveness" can actually improve "perceived clarity and intelligibility", "echoes" generally tend to degrade it...
Except that, after some longer delay, our brains are able to actively ignore echoes, and prevent them from interfering with the original signal...
And ALL OF THAT varies with frequency, and level, and with the amount and type of background noise present.
And it also varies from person to person.
And with different content (which has the most energy in different frequency ranges).
So all of that is just a bit subjective.

Feb 26, 2023 11:25:07 GMT -5 marcl said:

From time to time I try to get more specific information on what constitutes a good range for Clarity values C50, C80 and D50. I almost always get frustrated and give up having found nothing but circular discussions with no specifics. So today I asked ChatGPT specific questions based on the values I measured for my center channel.  The responses were consistent with what I know about the definitions of these parameters, AND it gave straight answers regarding my typical values!  

Is a value of 20db for C50 considered good?

A C50 value of 20 dB would be considered very good, as it indicates a significant level difference between the direct sound and the reflected sound arriving at the listener's ears 50 milliseconds after the direct sound.
A C50 value of 20 dB or higher is typically associated with excellent speech intelligibility, even in large or noisy rooms. However, it's important to note that the specific application and size of the room may affect what constitutes a "good" C50 value. Additionally, other factors, such as the frequency response of the room and the background noise level, may also play a role in determining overall speech intelligibility.

Is a value of 35db for c80 considered good?

A C80 value of 35 dB would be considered excellent, as it indicates a significant level difference between the direct sound and the sound reflected off the surfaces of the room 80 milliseconds after the direct sound.
A C80 value of 35 dB or higher is typically associated with very good speech intelligibility and a high-quality acoustic environment, even in larger or more reverberant spaces.

Is a value of 100% for d50 considered good for speech and music?

A D50 value of 100% would indicate that all of the direct sound and early reflections are reaching the listener's ear within 50 milliseconds, which would be ideal for speech and music intelligibility.
However, it's important to note that a D50 value of 100% may not be attainable in many real-world acoustic environments due to the presence of background noise, room dimensions, and other factors. In general, a D50 value of 0.4 or higher is considered very good for speech and music intelligibility, but the specific requirements may vary depending on the application and room size.

Of course ... but just because there is no one absolute objective number, does not mean there isn't a range over which many people find music and voice clarity to be subjectively pleasing.  That's what I have been looking for, and so far have found no studies that correlate those three measurements, frequency ranges and subjective clarity.  I think it is possible to give a range ... some guidance ...

The best I have found is somewhere someone said values above 20db are generally good.  The reason I asked about the specific values above is that they are the rough values that I measure for my center channel above 200Hz.  I would say my subjective view is in my room speech is VERY clear and music sounds amazing ... especially with this new CCR center speaker.  I have about 180sqft of 1D and 2D diffusion and very little broadband absorption.  So I'll toss out there that a Clarity measurement like this would probably sound good to most people.

[KeithL]

I think a large part of the reason why you don't see numbers quoted for this sort of thing is that a few numbers generally don't suffice to tell the entire picture.
(And, while many "good rooms" have "certain numbers", attempting to "design by the numbers" won't necessarily result in "a good room".)

For example, if we're talking about dialog intelligibility, we're concerned mostly with the range of frequencies for human voice (say 500 hz to 2 kHz).
However, while the overall level of early reflections occurring matters, the direction they're coming from is probably more critical.
(So having a single measurement of how quickly the level of those frequencies decays at the listening position isn't going to tell us enough.)
However you WILL find plenty of advice about what speaker arrangements and room treatments tend to produce a good result.

In that example...
Assuming we're looking at a really simple 3.0 channel situation with left, right, and center speakers...
We would like to see some sort of absorber on the floor about halfway between the center channel speaker and the primary listening position.
(To reduce early reflections from the floor in between.)
And we would probably like to see some absorbers on the side walls on both sides.
(Again to reduce early reflections from the center channel, to the listening position, from the side walls.)
It might be beneficial to have an absorber on the ceiling, but we can live without that, although a bit of diffusion (like a textured ceiling) would probably also be beneficial.
BUT it would be MOST important that the reflections from the two side walls were symmetrical...
(Excessive reflections will make the dialog a bit less intelligible... but asymmetrical reflections from one side will make it "pull to one side"... which is far more annoying.)

This is the reason most folks I know will be more inclined to offer suggestions about "good room layout" than about specific "target measurements"...

Feb 27, 2023 11:19:42 GMT -5 marcl said:

Feb 27, 2023 10:59:03 GMT -5 KeithL said:

I think it's worth pointing out why you're not going to find any specific answers for this...
It's also well worth starting out by understanding that all of those values vary with frequency...
(And, when you're discussing "intelligibility", the values that matter are going to be those for the frequency range of human voice.)

The reason you can't get a single answer is simply that "optimum clarity" and "most pleasing sound quality" don't necessarily coincide.
To pick the most extreme example - a totally dead room, with no reflections at all, will have excellent "clarity and intelligibility"...
And a room where it's "easy to make out announcements on the PA" often sounds horrible for music.
But virtually everyone agrees that a totally dead room (an anechoic chamber) sounds really unpleasant. 
And it also turns out that the way our brains handle delayed sound is nowhere near linear.
We hear delayed sounds that we hear within about 50 milliseconds as "liveness"...
And we hear sounds that we hear after around 100 milliseconds as "echoes"...
And, while "liveness" can actually improve "perceived clarity and intelligibility", "echoes" generally tend to degrade it...
Except that, after some longer delay, our brains are able to actively ignore echoes, and prevent them from interfering with the original signal...
And ALL OF THAT varies with frequency, and level, and with the amount and type of background noise present.
And it also varies from person to person.
And with different content (which has the most energy in different frequency ranges).
So all of that is just a bit subjective.

Of course ... but just because there is no one absolute objective number, does not mean there isn't a range over which many people find music and voice clarity to be subjectively pleasing.  That's what I have been looking for, and so far have found no studies that correlate those three measurements, frequency ranges and subjective clarity.  I think it is possible to give a range ... some guidance ...

The best I have found is somewhere someone said values above 20db are generally good.  The reason I asked about the specific values above is that they are the rough values that I measure for my center channel above 200Hz.  I would say my subjective view is in my room speech is VERY clear and music sounds amazing ... especially with this new CCR center speaker.  I have about 180sqft of 1D and 2D diffusion and very little broadband absorption.  So I'll toss out there that a Clarity measurement like this would probably sound good to most people.


View AttachmentView AttachmentView Attachment

[marcl]

Feb 27, 2023 15:23:28 GMT -5 KeithL said:

I think a large part of the reason why you don't see numbers quoted for this sort of thing is that a few numbers generally don't suffice to tell the entire picture.
(And, while many "good rooms" have "certain numbers", attempting to "design by the numbers" won't necessarily result in "a good room".)

For example, if we're talking about dialog intelligibility, we're concerned mostly with the range of frequencies for human voice (say 500 hz to 2 kHz).
However, while the overall level of early reflections occurring matters, the direction they're coming from is probably more critical.
(So having a single measurement of how quickly the level of those frequencies decays at the listening position isn't going to tell us enough.)
However you WILL find plenty of advice about what speaker arrangements and room treatments tend to produce a good result.

In that example...
Assuming we're looking at a really simple 3.0 channel situation with left, right, and center speakers...
We would like to see some sort of absorber on the floor about halfway between the center channel speaker and the primary listening position.
(To reduce early reflections from the floor in between.)
And we would probably like to see some absorbers on the side walls on both sides.
(Again to reduce early reflections from the center channel, to the listening position, from the side walls.)
It might be beneficial to have an absorber on the ceiling, but we can live without that, although a bit of diffusion (like a textured ceiling) would probably also be beneficial.
BUT it would be MOST important that the reflections from the two side walls were symmetrical...
(Excessive reflections will make the dialog a bit less intelligible... but asymmetrical reflections from one side will make it "pull to one side"... which is far more annoying.)

This is the reason most folks I know will be more inclined to offer suggestions about "good room layout" than about specific "target measurements"...

Feb 27, 2023 11:19:42 GMT -5 marcl said:

Of course ... but just because there is no one absolute objective number, does not mean there isn't a range over which many people find music and voice clarity to be subjectively pleasing.  That's what I have been looking for, and so far have found no studies that correlate those three measurements, frequency ranges and subjective clarity.  I think it is possible to give a range ... some guidance ...

The best I have found is somewhere someone said values above 20db are generally good.  The reason I asked about the specific values above is that they are the rough values that I measure for my center channel above 200Hz.  I would say my subjective view is in my room speech is VERY clear and music sounds amazing ... especially with this new CCR center speaker.  I have about 180sqft of 1D and 2D diffusion and very little broadband absorption.  So I'll toss out there that a Clarity measurement like this would probably sound good to most people.


View AttachmentView AttachmentView Attachment

Great example!  From way back when I built the room I had a lot of Sonex on the ceiling.  When it finally rotted I pulled it down and imaging went to hell.  I put absorbers back on the ceiling half way between the couch and speakers and the imaging came back.  I experimented with absorption, reflection and diffusion on the side walls half way between the speakers and couch, and the absorption gave the best imaging.  I have measurements back to 2012 so I checked and the Clarity measurements generally track ... improving incrementally as I added room treatment with imaging and soundstage being the criteria for success.

My intent in all of this is to correlate measurements with what I did and what I heard ... each step of the way.  Without writing a book about it, I can say that overall the trend has been there are measurements that correlate with scientific theory on the room treatments and speaker placement, as well as my listening along the way.  It doesn't always leap out at you, but it definitely does not contradict.

[marcl]

On the theme of correlating audible phenomena, REW measurements ... and solutions ...

Early Pandemia I was making a lot of changes in my system and at one point in the Fall 2020 I was hearing an asymmetry in soundstage ... like a hole on the right side of the right speaker.  One thing about it was that I could not move the imaging farther to the right by increasing levels on the right.  It got louder, but the image didn't move.  This suggested it might be due to some reflection/cancellation scenario where changing level just causes the cancellation to deepen.  So I looked at REW Impulse Response in % mode, Energy Time Curve (ETC) plot.



I saw this reflection at 18ms.  So what could be causing that and could it be related to what I'm hearing?  I have a couple 2x4ft absorbers so I started moving them around and remeasuring.  I found that when I put them on the reflective wall 90 degrees to the right of the MLP, the 18ms reflection went away and the soundstage was immediately fixed!



I had a 3x3ft absorber made to fill that space on the wall and the problem was solved!

Was there a correlation with any other measurements?  I can't say for sure, but the Clarity measurements for the right channel with and without the absorber indicate an improvement with the absorber.   



A year and a half passed and I made a change to the location of the front speakers.  After the process settled down I observed that soundstage issue on the right again.  I did various things and wasn't really thinking back to that reflection.  There was nothing in the measurements that indicated a reflection problem and the 18ms peak had not returned.  Eventually it occurred to me to take down the 3x3 absorber.  The soundstage issue was immediately fixed ... and, the 18ms reflection had not returned.  So the movement of my front speakers toward the front wall had either blocked the reflection or sent it out the doorway to my left.  And for some reason having absorption to my right was interfering with some beneficial reflections.  And BTW, I had always preferred to have diffusion on that right wall but had taken them down when I moved my side surrounds forward a bit (and hung a picture there).  So I took down the picture, put the diffusers back up and everything sounded fine.  It seems diffusion to my right has a pleasing affect of balancing the open doorway to my left.

Symmetry ... correlation ... listen ... verify ...

[fbczar]

Feb 28, 2023 7:55:47 GMT -5 marcl said:

On the theme of correlating audible phenomena, REW measurements ... and solutions ...

Early Pandemia I was making a lot of changes in my system and at one point in the Fall 2020 I was hearing an asymmetry in soundstage ... like a hole on the right side of the right speaker.  One thing about it was that I could not move the imaging farther to the right by increasing levels on the right.  It got louder, but the image didn't move.  This suggested it might be due to some reflection/cancellation scenario where changing level just causes the cancellation to deepen.  So I looked at REW Impulse Response in % mode, Energy Time Curve (ETC) plot.

View Attachment

I saw this reflection at 18ms.  So what could be causing that and could it be related to what I'm hearing?  I have a couple 2x4ft absorbers so I started moving them around and remeasuring.  I found that when I put them on the reflective wall 90 degrees to the right of the MLP, the 18ms reflection went away and the soundstage was immediately fixed!

View Attachment

I had a 3x3ft absorber made to fill that space on the wall and the problem was solved!

Was there a correlation with any other measurements?  I can't say for sure, but the Clarity measurements for the right channel with and without the absorber indicate an improvement with the absorber.   

View AttachmentView Attachment

A year and a half passed and I made a change to the location of the front speakers.  After the process settled down I observed that soundstage issue on the right again.  I did various things and wasn't really thinking back to that reflection.  There was nothing in the measurements that indicated a reflection problem and the 18ms peak had not returned.  Eventually it occurred to me to take down the 3x3 absorber.  The soundstage issue was immediately fixed ... and, the 18ms reflection had not returned.  So the movement of my front speakers toward the front wall had either blocked the reflection or sent it out the doorway to my left.  And for some reason having absorption to my right was interfering with some beneficial reflections.  And BTW, I had always preferred to have diffusion on that right wall but had taken them down when I moved my side surrounds forward a bit (and hung a picture there).  So I took down the picture, put the diffusers back up and everything sounded fine.  It seems diffusion to my right has a pleasing affect of balancing the open doorway to my left.

Symmetry ... correlation ... listen ... verify ...

Great post Marc. As a fellow Magneplanar owner I am concerned with reflections and the impulse response of my system. Your description of using REW to eliminate unwanted reflections and poor impulse response is enlightening. I am sure you use Dirac for the same purpose. 

[marcl]

Feb 28, 2023 21:14:45 GMT -5 fbczar said:

Feb 28, 2023 7:55:47 GMT -5 marcl said:

On the theme of correlating audible phenomena, REW measurements ... and solutions ...

Early Pandemia I was making a lot of changes in my system and at one point in the Fall 2020 I was hearing an asymmetry in soundstage ... like a hole on the right side of the right speaker.  One thing about it was that I could not move the imaging farther to the right by increasing levels on the right.  It got louder, but the image didn't move.  This suggested it might be due to some reflection/cancellation scenario where changing level just causes the cancellation to deepen.  So I looked at REW Impulse Response in % mode, Energy Time Curve (ETC) plot.

View Attachment

I saw this reflection at 18ms.  So what could be causing that and could it be related to what I'm hearing?  I have a couple 2x4ft absorbers so I started moving them around and remeasuring.  I found that when I put them on the reflective wall 90 degrees to the right of the MLP, the 18ms reflection went away and the soundstage was immediately fixed!

View Attachment

I had a 3x3ft absorber made to fill that space on the wall and the problem was solved!

Was there a correlation with any other measurements?  I can't say for sure, but the Clarity measurements for the right channel with and without the absorber indicate an improvement with the absorber.   

View AttachmentView Attachment

A year and a half passed and I made a change to the location of the front speakers.  After the process settled down I observed that soundstage issue on the right again.  I did various things and wasn't really thinking back to that reflection.  There was nothing in the measurements that indicated a reflection problem and the 18ms peak had not returned.  Eventually it occurred to me to take down the 3x3 absorber.  The soundstage issue was immediately fixed ... and, the 18ms reflection had not returned.  So the movement of my front speakers toward the front wall had either blocked the reflection or sent it out the doorway to my left.  And for some reason having absorption to my right was interfering with some beneficial reflections.  And BTW, I had always preferred to have diffusion on that right wall but had taken them down when I moved my side surrounds forward a bit (and hung a picture there).  So I took down the picture, put the diffusers back up and everything sounded fine.  It seems diffusion to my right has a pleasing affect of balancing the open doorway to my left.

Symmetry ... correlation ... listen ... verify ...

Great post Marc. As a fellow Magneplanar owner I am concerned with reflections and the impulse response of my system. Your description of using REW to eliminate unwanted reflections and poor impulse response is enlightening. I am sure you use Dirac for the same purpose. 

Well Dirac cleans up the impulse responses for sure and I think the audible result of that is clarity especially in transients ... and it kind of removes some fog.  But of course Dirac can't do anything about reflections so REW is the only way to track those down.  ttocs told me yesterday he taped an old CD on the wall and used a laser measurer to bounce reflections and correlate distance to time on the ETC plot.

[lhracing]

I have a question for you REW experts. When measuring a 2 channel only system should you have the microphone pointing up or straight forward?

[marcl]

Mar 3, 2023 13:24:17 GMT -5 lhracing said:

I have a question for you REW experts. When measuring a 2 channel only system should you have the microphone pointing up or straight forward?

I guess either one is okay, but be sure to use the right calibration file.  I always measure with it pointing up.

[ttocs]

I aim the mic up, 90º, when measuring the room.

If I wanted to measure speakers outside I'd point the mic at the speaker.

[KeithL]

The right answer depends on both the situation and the microphone.

Virtually all calibrated microphones have a very different frequency response at 0 degrees than they do at 90 degrees.
In an anechoic chamber, or an open field, the sound is reaching the microphone in a straight line from the speaker.
Therefore, in those situations, you WILL NOT get accurate results unless you use the appropriate calibration file.
Some microphones include both; some microphones only include one calibration file and specify the direction in which they should be used.
For example, the microphone we include with the RMC-1 and XMC-2 is calibrated to be accurate at 90 degrees and comes with a 90 degree calibration file.
So you're SUPPOSED to use that one pointed straight up.
And, for the microphone that we include with our MC1 and MR1, we specify that you should point the microphone towards each speaker (0 degrees).
So, with that microphone, and EmoQ, THAT is the way you'll get the best results.

However the details become somewhat more complicated in a room.
To put it as simply as possible, in a room, you must account for both direct sound arriving directly from the speaker, and sound that arrives after reflecting off of various surfaces.
This is handled in different ways by different software - which is why you should always follow the directions for your particular measurement system.

Here are a few of the "considerations" involved:

1.
If you use something like pink noise, and a "real time analyzer", or an SPL meter, you are measuring a long term average.
Basically you are measuring both direct and reflected sound, reaching the microphone from all directions, all thoroughly mixed together.
(And so also affected by the differences in the response of the microphone to sound from various directions.)

2.
When you use a program like REW or Dirac Live, the program is able to separate direct and reflected sound by using various types of "time windowing".
Sound arriving directly from the speaker will arrive before reflections from surfaces (because it has a shorter trip).
Some programs "fire off a tone, listen for how loud it is when it first arrives, then stop listening before the reflections have time to get to the microphone"...
Others do something similar with those familiar "chirp tones" (they only listen to each particular frequency when it first arrives then ignore it later)...
Others do a much more sophisticated analysis of all of the sound coming from the chirp tone and use that information to "model the response of the room".
(This is basically what Dirac Live does.)

3.
With those more sophisticated measurements and analyses....
- you still have a microphone whose response will be different in different directions
- and you have a room whose dimensions are basically unknown
- and, furthermore, you are starting out with no knowledge of the acoustic properties of each of the surfaces in your room
- as a result of the large number of unknowns, and the limited number of measurements it is practical to take, room correction software makes a lot of assumptions
   (which is why you sometimes get odd results)

For example, based on analysis, the software can tell that "later reflections contain less high frequency content" - and exactly how much.
But it's less able to distinguish whether that's because the later reflections are from the ceiling - which happens to be covered in shag carpet....
Or whether the ceiling is pretty bright, but the reflections from the sides are different because of differences in the microphone's 90 degree and on-axis response calibrations...
(And that's were it's going to be making some assumptions based on what it knows about "typical microphones" and "typical rooms"... )

When we use a calibrated microphone to measure speakers for design purposes we "clamp down the window so we're ONLY measuring the sound coming directly form the speaker"....
(Or we measure them out in the open - or in something close to an anechoic chamber - or we compensate for "room response".)
However, with room correction, you don't want to do that; you WANT to include both the speaker and the room...
But, at the same time, you want to be able to keep track of which is which, and handle them a bit differently...

Mar 3, 2023 13:52:25 GMT -5 ttocs said:

I aim the mic up, 90º, when measuring the room.

If I wanted to measure speakers outside I'd point the mic at the speaker.

[marcl]

This is brilliant ... it's what we've always wanted!  Here is how to use REW to measure Atmos channels!

First, here's a video that tells you exactly how to do it.  There's a lot that does not apply to us as we use Emotiva gear and Dirac, but the Atmos part starts at 00:22:46




In order to do the measurements you need the .WAV and .mp4 files located here:

drive.google.com/file/d/1APNO_yTseoPoBoyfwyjGLBTFat8g20_4/view

Watch very carefully.  There are specific steps to disable ASIO and measure From File.  You load the .WAV file after selecting From File in the Measurement screen.  Then you separately play the .mp4 file using Windows Media Player or whatever you have (i.e. you don't play the .mp4 files from within REW).  REW will wait for the chirp and record the sweep.  You get all the usual data.  Brilliant!

Here's the response of my Atmos Tops - Magnepan MMGW Front, Polk OWM5 Rear - with crossovers at 120Hz ... along with all the other channels measured with the usual method.  Psychoacoustic Smoothing.

[ttocs]

Mar 12, 2023 10:47:57 GMT -5 marcl said:

This is brilliant ... it's what we've always wanted!  Here is how to use REW to measure Atmos channels!

Watch very carefully.  There are specific steps to disable ASIO and measure From File.  You load the .WAV file after selecting From File in the Measurement screen.  Then you separately play the .mp4 file using Windows Media Player or whatever you have (i.e. you don't play the .mp4 files from within REW).  REW will wait for the chirp and record the sweep.  You get all the usual data.  Brilliant!

Very cool!

So basically we need to disallow REW from taking control of the sweep and play one manually. So when we click PLAY in REW it's just listening. Neato! as Herman Munster used to say.

Now I gotta figure out how to do this with a Mac.