Two dips in L/R FR but not in Left and Right Individually

Trdat

Member
Thread Starter
Joined
Nov 6, 2019
Messages
186
I have a more pronounced dip at the crossover point of 950hz in the measurement with Left and Right together than in the individual measurements and there is another dip further along at 3500hz.

These measurements are with REW using a correction filter that looked good enough in Audiolense.

Any ideas? Is the dip at the crossover pont normal? How can I work it to ensure a flatter response?
 

Attachments

  • Leftright.jpg
    Leftright.jpg
    78.9 KB · Views: 50
  • Right.jpg
    Right.jpg
    77.4 KB · Views: 51
  • LEft.jpg
    LEft.jpg
    78 KB · Views: 55

juicehifi

Audiolense
Staff member
Joined
Feb 5, 2018
Messages
697
These are almost guaranteed to be due to phase differences after correction. Mainly you should improve your acoustic environment to get a flatter sum, but what you see here is 3% of what you should expect to see from a system that doesn't do FIR correction ... typically it sounds balanced, but the sound stage is very unfocused compared to using a set of very good FIR filters. So how does the sound stage appear in your place? My guess is that it sounds like the walls have disappeared ... compared to not using the Audiolense filters.
 

Trdat

Member
Thread Starter
Joined
Nov 6, 2019
Messages
186
These are almost guaranteed to be due to phase differences after correction. Mainly you should improve your acoustic environment to get a flatter sum, but what you see here is 3% of what you should expect to see from a system that doesn't do FIR correction ... typically it sounds balanced, but the sound stage is very unfocused compared to using a set of very good FIR filters. So how does the sound stage appear in your place? My guess is that it sounds like the walls have disappeared ... compared to not using the Audiolense filters.

Well, the above measurements are on my XO system so I don't have the option to toggle on or off Audiolense but overall it sounds fantastic I was just curious if that was normal and if we could do anything about it?

My room is heavily treated. Plus, it was weird for the L and R to be very flat and the L/R to be not as flat.

So you are saying its perfectly normal to get phase issues that might effect the frequency repsonse?
 

MediumRare

Member
Joined
Nov 8, 2019
Messages
66
Well, the above measurements are on my XO system so I don't have the option to toggle on or off Audiolense but overall it sounds fantastic I was just curious if that was normal and if we could do anything about it?

My room is heavily treated. Plus, it was weird for the L and R to be very flat and the L/R to be not as flat.

So you are saying its perfectly normal to get phase issues that might effect the frequency repsonse?
Yes. Also, try applying psychoacoustic smoothing and check the results. Almost certainly inaudible. You might end up pulling down some of the HF bumps to get a proper downslope. Otherwise your curves look outstanding.
 

DanDan

Moderator
Staff member
Joined
Mar 10, 2018
Messages
743
Measurements with a single mic do not represent what we receive at either ear nor the combination which the brain makes. Either measure at the ear positions or use the summing features in REW.
 
Last edited:

Trdat

Member
Thread Starter
Joined
Nov 6, 2019
Messages
186
Yes. Also, try applying psychoacoustic smoothing and check the results. Almost certainly inaudible. You might end up pulling down some of the HF bumps to get a proper downslope. Otherwise your curves look outstanding.
Great news, yes I got a decent step response to.

Pshycoacoustic smoothing in Audoilense? Where do I find it?
 

Trdat

Member
Thread Starter
Joined
Nov 6, 2019
Messages
186
Measurements with a single mic do not represent what we receive at either ear nor the combination which the brain makes. Either measure at the ear positions or use the summing features in REW.
Ear positions as in multiple varying listening positions?

And summing feature where do I find it? I am presuming that it combines all the measerements are you saying that is a better indicator of what we hear...?
 

Trdat

Member
Thread Starter
Joined
Nov 6, 2019
Messages
186
Smoothing option in REW graph.
Got it. Yeh.

But your second recommendation is to use the target curve in Audiolense to pull down the HF which might give a flatter response towards the high frequency right?
 

DanDan

Moderator
Staff member
Joined
Mar 10, 2018
Messages
743
Ear positions as in multiple varying listening positions?

Sure but at least two. Nobody nose what goes on in the centre!

And summing feature where do I find it? I am presuming that it combines all the measerements are you saying that is a better indicator of what we hear...?

Can't remember, but I would guess in the Graph Controls pane top right. Graph Arithmetic.......
If one ear hears a narrow notch and the other does not, it seems most unlikely that the brain would take notice of the notch.
If one were doing extremely critical balancing in a studio mix context perhaps. There was a school of thought that came up with third octave practices on the basis that this is the smallest band of frequencies we readily notice.... So narrow stuff... , particularly on one side, no pasa nada.
 

jjazdk

Member
Joined
Aug 17, 2018
Messages
95
Ear positions as in multiple varying listening positions?

Sure but at least two. Nobody nose what goes on in the centre!

And summing feature where do I find it? I am presuming that it combines all the measerements are you saying that is a better indicator of what we hear...?

Can't remember, but I would guess in the Graph Controls pane top right. Graph Arithmetic.......
If one ear hears a narrow notch and the other does not, it seems most unlikely that the brain would take notice of the notch.
If one were doing extremely critical balancing in a studio mix context perhaps. There was a school of thought that came up with third octave practices on the basis that this is the smallest band of frequencies we readily notice.... So narrow stuff... , particularly on one side, no pasa nada.
Agreed, narrow notches are not much detrimental to the sound quality. Narrow peaks on the other hand, can be very annoying.
Wether or not a single microphone position is enough is not as clear cut. But if you want to measure at the ears, I would propose doing a long average (80-100 averages) by waving the mic while REW averages a pink noise stimuli.
 

DanDan

Moderator
Staff member
Joined
Mar 10, 2018
Messages
743
A second mic position a head width apart is 100% more information. Plus both can be at actual ear locations, rather than the nose.
Unfortunately Sonarworks use Hand Held Measurement and there are many advocates of the Moving Mic Method on the Internet.
I don't recommend it unless one includes a Boom Pole to distance the mic from the operator's body.
B&K Near the Mic.png
 

wahoospiff

New Member
Joined
Jan 15, 2018
Messages
11
The summing feature of REW that I use is in the "All SPL" sheet of the main REW window, then click on the upper right "Controls" button, then click on "Alignment Tool" in the window that pops up which brings up another window called "Alignment Tool". In this last window, if you select "R" and "L", REW will sum the left and right channels to get L+R, ie., it should match what REW records when both speakers are playing. I'm sure there are a million other ways to sum but I use this.
 

MediumRare

Member
Joined
Nov 8, 2019
Messages
66
Got it. Yeh.

But your second recommendation is to use the target curve in Audiolense to pull down the HF which might give a flatter response towards the high frequency right?
I’m not familiar with Audiolense, but using the guidance from Harman research, the ideal in-room curve is downward-sloped 10 db from about 200hz to 20k hz.

Regarding the "two-ears needs two measurements" question, multiple measurements are better but not because of two ears. It’s because you don’t listen with your head in a vise; you don’t want your only measurement to be unluckily in a room mode or other node and give a misleading indication of what EQ is needed. There is a more sophisticated technical reason but never mind that.

This tells you what you need: https://audiosciencereview.com/forum/index.php?threads/rew-moving-microphone-method-help.12641/
 

DanDan

Moderator
Staff member
Joined
Mar 10, 2018
Messages
743
Regarding the "two-ears needs two measurements" question, multiple measurements are better but not because of two ears. It’s because you don’t listen with your head in a vise; you don’t want your only measurement to be unluckily in a room mode or other node and give a misleading indication of what EQ is needed. There is a more sophisticated technical reason but never mind that.

Multiple measurements are very welcome. Remembering that each mic location requires 3 speaker sweeps, i.e. L, R, L+R
Moving Mic measurement may be an option, but will definitely be flawed unless a boom is used to keep the mic well away from the operator. Please see the Bruel and Kjaer advice in my previous post.
 

MediumRare

Member
Joined
Nov 8, 2019
Messages
66
Regarding the "two-ears needs two measurements" question, multiple measurements are better but not because of two ears. It’s because you don’t listen with your head in a vise; you don’t want your only measurement to be unluckily in a room mode or other node and give a misleading indication of what EQ is needed. There is a more sophisticated technical reason but never mind that.

Multiple measurements are very welcome. Remembering that each mic location requires 3 speaker sweeps, i.e. L, R, L+R
Moving Mic measurement may be an option, but will definitely be flawed unless a boom is used to keep the mic well away from the operator. Please see the Bruel and Kjaer advice in my previous post.
Funny story: I just did a moving mike test on my own listening room. I did it with and without a person sitting in the primary listening position. Then I left the mike on a tripod and repeated the measurement. There were barely a handful of ripples of differences between the three trials, comparable to run differences with a stationary mike. If one stands to the side while they move the mike I’m confident there will be no audible impact at all.
 

DanDan

Moderator
Staff member
Joined
Mar 10, 2018
Messages
743
Funny story: I just did a moving mike test on my own listening room. I did it with and without a person sitting in the primary listening position. Then I left the mike on a tripod and repeated the measurement. There were barely a handful of ripples of differences between the three trials, comparable to run differences with a stationary mike. If one stands to the side while they move the mike I’m confident there will be no audible impact at all.

I don't really see a humorous aspect. You are presenting an anecdote which undermines Bruel and Kjaer's work and the cautions regarding the use of Moving Mics without distancing poles. B&K warn about "errors of up to 6dB around 400Hz" and you seem to be suggesting they are incorrect according to your 'research'. What now Fake Acoustics?
 

jtalden

Senior Member
Joined
May 22, 2017
Messages
887
Location
Arizona, USA
More  
Preamp, Processor or Receiver
Marantz AV7705 Pre/Pro
Main Amp
VTV 6 chnl NC252MP P-amp x 2
Additional Amp
Behringer DCX2496 x 2
Universal / Blu-ray / CD Player
OPPO BDP-103 Universal Player
Front Speakers
DIY SEAS H1456/H1212 Spkr x 5
Subwoofers
DIY JBL 2235H 15" SW x 2
Video Display Device
JVC DLA-X790R
Screen
Da-Lite Da-Snap 39105V - 92"
I would like to add my support for the use of MMM as a reasonable way to make suitable measurements for the purposes of PEQ. It is quick and easy way to get very repeatable average SPL measurement over a chosen area or volume. It works well for many typical room situations. I am not suggesting that other methods are not viable, nor suggesting it is better than all others. I am also not dismissing B&K findings. I am sure their results are accurate for the measuring conditions they chose.
The MM method has worked well for me and many others. This is a good website for more detailed information. I have repeatedly used many the other suggested measuring methods for LP EQ purposes and I keep returning to MMM as the easiest to implement and still provide good results. Below are some of my own measurements from 2013.

Below 6 MMM measurements were taken centered on the LP and covering an area about 20"H x 36"W. This shows excellent repeatability. The starting point of the mic is not a significant factor. Since this time I have added about 12" of depth to the area as well, but it doesn't seem to make a significant difference in the 2 main rooms I work with.
1 6 LP area MMM Measurements.jpg


Below are 7 sweep measurements in the same area. The first one being with mic at the LP and the other were offset at the extremes of the same area. The average of 7 is the black trace.
2 7 LP Sweeps with their average.jpg


Below is the main LP sweep compared to the average of them all. There is reasonably good agreement in this case. Smoothing the LP measurement would lead to close agreement except in the 400 to 1000 Hz range where the deviation may be enough to be significant for EQ.
3 Main LP Sweep vs 7 sweep avg around LP.jpg


Below is one the MMM measurements compared to the 7 sweep average. Someplace I have data showing that the more random sweep locations that are taken the closer the sweep average approaches the MMM measurement. There is no clear indication the handheld MMM process led to a measurement dip in the midrange.
4 MMM vs 7 sweeps avg around LP.jpg


My conclusion is that MMM is a viable option for those of us who want to use average measurements in the LP area for our PEQ work. We can adjust the area as desired and obtain an accurate and repeatable, average measurement with a minimum of time and effort. Confirming this is the case for a particular room setup is not very difficult. That said, just using a single LP sweep measurement can also lead to good PEQ results so it is difficult to completely dismiss any of the commonly proposed methods for PEQ.
 

DanDan

Moderator
Staff member
Joined
Mar 10, 2018
Messages
743
Hi jtaiden. Could you clarify some details please. Your mic moving area is 12" deep right? So all of your moves are in the B&K no fly zone? When you did the sweep measurements were you sitting behind the mic?

All dual speaker measurements which are not precisely central suffer from HF anomalies. This is due to comb filtering caused by the different distances (time of flight) to the tweeters. This is a false measurement of your HF response because neither ear nor both can experience this cancellation because of the blocking effect of the head. So any Eq derived from this false measure is fruit of the poisoned tree.

1 vs 2 Spkr Comparison.gif


Also...... the windscreens we use professionally are substantial and tested. Unless one moves very slowly there is bound to be VLF disturbance of the diaphragm. And indeed structure borne muscle and bone creaking if you are of my age!

I often speed up finding no fly zones by using Pink Noise and RTA Spectrum. But I Zoom to LF only, say 30-200Hz.
 

jtalden

Senior Member
Joined
May 22, 2017
Messages
887
Location
Arizona, USA
More  
Preamp, Processor or Receiver
Marantz AV7705 Pre/Pro
Main Amp
VTV 6 chnl NC252MP P-amp x 2
Additional Amp
Behringer DCX2496 x 2
Universal / Blu-ray / CD Player
OPPO BDP-103 Universal Player
Front Speakers
DIY SEAS H1456/H1212 Spkr x 5
Subwoofers
DIY JBL 2235H 15" SW x 2
Video Display Device
JVC DLA-X790R
Screen
Da-Lite Da-Snap 39105V - 92"
Hi jtaiden. Could you clarify some details please. Your mic moving area is 12" deep right?
This data from 2013 did not include any front/rear mic motion - just up/down and right/left. I just was commenting that currently I do include some front/rear motion as well and it hasn't made a significant difference in my room.

The LP for the posted set of measurements was almost 4 m in from of the speaker being measured. The LP is at the center of the 20"H X 36"W window that contained the 7 sweeps and the 6 MMM measurements. This is a little different from my current MMM process that also includes some mic motion front/rear along with the up/down and right/left that was use here.
So all of your moves are in the B&K no fly zone? When you did the sweep measurements were you sitting behind the mic?
Yes, In 95% of the cases I have start with the mic on a stand at the LP. I start the RTA forever measurement at the PC a few feet away. I move to the mic and remove it from the stand and sweep through the window range. When finished, I replace the mic on the stand and move to the PC to stop the RTA. The process is very forgiving to using other procedures for the mechanics of the process.
All dual speaker measurements which are not precisely central suffer from HF anomalies. This is due to comb filtering caused by the different distances (time of flight) to the tweeters.
Yes, this is correct and is particularly bad for a single measurement of a stereo pair as you posted.
This is a false measurement of your HF response...
We are measuring one speaker at a time and getting an accurate representation of the taking many simple sweeps in that same area and then SPL averaging them. To discount using this method is discounting using sweep averages.

It is true that, if we instead vector averaged the same sweep measurements, we would get a very distorted SPL due to phase cancelations. That is why we use the simple SPL average.
Also...... the windscreens we use professionally are substantial and tested. Unless one moves very slowly there is bound to be VLF disturbance of the diaphragm. And indeed structure borne muscle and bone creaking if you are of my age!
I have tried moving the mic very slowly and relatively fast with no difference. I therefore just move it a comfortable moderate speed. The impact of any resonable spurious noise as in disconnecting or replacing the mic from the stand would normally be washed out in the large number of averages obtained. The body to mic positioning is always changing as both are moving during the process so again, I think this washes out.

If you tried this at a typical LP room LP measurement position what were your results - any data? It may not connect well with your normal sound activities. Your concerns are all warranted, I am only suggesting that unless you have more direct evidence that there is a significant problem using this method you may be dismissing it too strongly.
 

DanDan

Moderator
Staff member
Joined
Mar 10, 2018
Messages
743
We are measuring one speaker at a time and getting an accurate representation of the taking many simple sweeps in that same area and then SPL averaging them. To discount using this method is discounting using sweep averages.

So you get an RTA Average using one Speaker. Then another average using the other Speaker? Given your scepticism about vector arithmetic, how do you add or average them to get the L+R response? dB's don't play nice with sums.....

If you tried this at a typical LP room LP measurement position what were your results - any data? It may not connect well with your normal sound activities. Your concerns are all warranted, I am only suggesting that unless you have more direct evidence that there is a significant problem using this method you may be dismissing it too strongly.

Bruel and Kjaer are the gold standard of measurement. They work to legal standards.
Their proximity warning was in fact less dramatic than the simple test I just presented. 10dB different around 600Hz and not a particularly narrow dip.

In my Professional work I have tried most options. My recommendation is to use a mic stand. For quickly locating LF No Fly zones I recommend L+R speaker drive with Pink Noise and viewing on a 20-200Hz RTA. The Mic or speakers can be moved quickly and the result seen very soon after one steps well away from the mic. Mark the danger zones on the floor with masking tape. Given the intrinsic flaws I see nothing to commend the so called MMC.
 
Last edited:

jtalden

Senior Member
Joined
May 22, 2017
Messages
887
Location
Arizona, USA
More  
Preamp, Processor or Receiver
Marantz AV7705 Pre/Pro
Main Amp
VTV 6 chnl NC252MP P-amp x 2
Additional Amp
Behringer DCX2496 x 2
Universal / Blu-ray / CD Player
OPPO BDP-103 Universal Player
Front Speakers
DIY SEAS H1456/H1212 Spkr x 5
Subwoofers
DIY JBL 2235H 15" SW x 2
Video Display Device
JVC DLA-X790R
Screen
Da-Lite Da-Snap 39105V - 92"
So you get an RTA Average using one Speaker. Then another average using the other Speaker? Given your scepticism about vector arithmetic, how do you add or average them to get the L+R response? dB's don't play nice with sums.....
I take a sweep measurement of L and R separately. The 2 impulses are then time aligned in REW (if necessary) and vector averaged. That shows the interaction between the 2 accurately. This is helpful for speaker placement and acoustical treatments issues, but I don't EQ using this information.
Bruel and Kjaer are the gold standard of measurement. They work to legal standards.
Their proximity warning was in fact less dramatic than the simple test I just presented. 10dB different around 600Hz and not a particularly narrow dip.
Agreed, but I haven't seen them comment on EQ via MMM (or other method) at the listening position in a typical room setup. I like their old, recommended house curve even though it was developed when 1/3 octave GEQ was the standard. I am unclear if they have a recommend a best practice for listening room PEQ applications.
In my Professional work I have tried most options. My recommendation is to use a mic stand. For quickly locating LF No Fly zones I recommend L+R speaker drive with Pink Noise and viewing on a 20-200Hz RTA. The Mic or speakers can be moved quickly and the result seen very soon after one steps well away from the mic. Mark the danger zones on the floor with masking tape. Given the intrinsic flaws I see nothing to commend the so called MMC.
MMM was never intended for speaker location decisions, I agree it is not effective for that purpose. I didn't understand that was the context you were addressing. It's intended use is for LP PEQ work one speaker at a time. It is not perfect nor necessarily the best possible way to do that, but it compares very well with other processes.
 

DanDan

Moderator
Staff member
Joined
Mar 10, 2018
Messages
743
I take a sweep measurement of L and R separately. The 2 impulses are then time aligned in REW (if necessary) and vector averaged. That shows the interaction between the 2 accurately.

Sorry but NO. Two speakers simultaneously driving the room at two locations will combine in a unique way which cannot be assessed from L and R data. That is why we always recommend L, R, AND L+R

Bruel and Kjaer are measurement experts. Their best practice and equipment standards are probably the world's best, for any application.

MMM was never intended for speaker location decisions, I agree it is not effective for that purpose. I didn't understand that was the context you were addressing. It's intended use is for LP PEQ work one speaker at a time. It is not perfect nor necessarily the best possible way to do that, but it compares very well with other processes.

Again NO. Moving Mic Techniques come from way back. Lab measurements using motorised rotating booms. As I said I mount the mic ON A STAND and move it quickly from place to place observing real time results when optimising Speaker and Listener locations. Such Optimisation at LF has to have both speakers running because of the aforementioned physical interaction between them both and the room.
 

jtalden

Senior Member
Joined
May 22, 2017
Messages
887
Location
Arizona, USA
More  
Preamp, Processor or Receiver
Marantz AV7705 Pre/Pro
Main Amp
VTV 6 chnl NC252MP P-amp x 2
Additional Amp
Behringer DCX2496 x 2
Universal / Blu-ray / CD Player
OPPO BDP-103 Universal Player
Front Speakers
DIY SEAS H1456/H1212 Spkr x 5
Subwoofers
DIY JBL 2235H 15" SW x 2
Video Display Device
JVC DLA-X790R
Screen
Da-Lite Da-Snap 39105V - 92"
Well, my original misunderstanding that your comment was about MMM not being feasible for LP PEQ seems to be veering off into unrelated additional subjects. I think this is a good place for me to again apologize for my misunderstanding and bow out.
 
Top Bottom