Lately I've been using the moving measurement with Audio Tools on the iPhone along with the iMM-6 microphone. Results seem steady and useful. 1 sec or 3 sec average seems about the same, the graph settles down quickly.
MMM how to
First I'd like to make clear that I want to measure the frequency range of my system within my listening room which is a 'normal' living room ie not acoustically treated apart from lots of bookshelves on the side walls and in the corners.
The speakers response can be tweaked through dsp. But before that I'd like to find the best position for them, as well as the best listening sweet spot, within the room. To a degree my ears/brain are not fine enough (I don't have golden ears) to appreciate subtle differences.
So I'd be happy 1) to measure their frequency range, 2) to experiment with corrections and 3) to measure again with the correction.
In searching the internet, after having tried unsuccessfully Mathaudio and Dirac (Acourate and Audiolense are too technical for me), I found an article using REW (used to measure and to make filters for a convolution engine) which included the MMM paper published by jlo which piqued my interest.
My questions regarding MMM:
1) by spiraling the mic around me there will be some spots where the mic won't 'see' the speaker under measurement. Does it matter or should I spiral the mic with myself outside of the spiral?
2) as LF wave lengths are longer than HF ones should the speed of the moving mic vary to cater for both LF and HF? If so can you give me a clue on how to do it?
3) how long (in seconds) should last a measurement?
4) my listening chair is a bulky, comfortable armchair which might absorb and diffuse the sound at some frequencies. Should I make the measurements with or without the listening chair?
My question regarding REW: how to compare the before correction and the after correction signals?
Thanks.
Thanks Juhazi,
First I'd like to make clear that I want to measure the frequency range of my system within my listening room which is a 'normal' living room ie not acoustically treated apart from lots of bookshelves on the side walls and in the corners.
The speakers response can be tweaked through dsp. But before that I'd like to find the best position for them, as well as the best listening sweet spot, within the room. To a degree my ears/brain are not fine enough (I don't have golden ears) to appreciate subtle differences.
So I'd be happy 1) to measure their frequency range, 2) to experiment with corrections and 3) to measure again with the correction.
In searching the internet, after having tried unsuccessfully Mathaudio and Dirac (Acourate and Audiolense are too technical for me), I found an article using REW (used to measure and to make filters for a convolution engine) which included the MMM paper published by jlo which piqued my interest.
My questions regarding MMM:
1) by spiraling the mic around me there will be some spots where the mic won't 'see' the speaker under measurement. Does it matter or should I spiral the mic with myself outside of the spiral?
2) as LF wave lengths are longer than HF ones should the speed of the moving mic vary to cater for both LF and HF? If so can you give me a clue on how to do it?
3) how long (in seconds) should last a measurement?
4) my listening chair is a bulky, comfortable armchair which might absorb and diffuse the sound at some frequencies. Should I make the measurements with or without the listening chair?
My question regarding REW: how to compare the before correction and the after correction signals?
Thanks.
Last edited:
1) I haven't noticed any problems there.
2) Move it slowly like a yoga excercise.
3) Measurement should last about 10 seconds. The graph sets stable usually after 5 sec.
4) With the chair, listener and everything else on normal locations.
I use MMM to see the overall large scale balance of sound. I don't make room-eq corrections by it, only "voicing" which means something like 1 octave band "contouring". This is very useful when a (dsp-capable) speaker is set up in a new room.
You can also take say 5-7 normal sweeps by holding the mic in hand at various positions like in MMM. Then overlay them and you will see how crazy that looks! How to eq???
2) Move it slowly like a yoga excercise.
3) Measurement should last about 10 seconds. The graph sets stable usually after 5 sec.
4) With the chair, listener and everything else on normal locations.
I use MMM to see the overall large scale balance of sound. I don't make room-eq corrections by it, only "voicing" which means something like 1 octave band "contouring". This is very useful when a (dsp-capable) speaker is set up in a new room.
You can also take say 5-7 normal sweeps by holding the mic in hand at various positions like in MMM. Then overlay them and you will see how crazy that looks! How to eq???
I mean that doing multiple "normal sweep" measurements at various locations reveals how location-spesific single measurements are. Still many people do eq based on them, which is not wise, or even wrong
Oh, and measure L and R speakers separately, and both playing. You will notice that MMM smooths out diffraction effects nicely. Change smoothing settings and use both bars and lines to learn how measurements can lie! Farfield/room measurements contain all the interferences added to direct sound and it is not wise to eq sharp peaks.
I don't like room-eq at all, just find the right locations for speakers and the listener! Various decayed signal/energy analysis' helps to do acoustic "treatment" to walls and other boundaries.
Oh, and measure L and R speakers separately, and both playing. You will notice that MMM smooths out diffraction effects nicely. Change smoothing settings and use both bars and lines to learn how measurements can lie! Farfield/room measurements contain all the interferences added to direct sound and it is not wise to eq sharp peaks.
I don't like room-eq at all, just find the right locations for speakers and the listener! Various decayed signal/energy analysis' helps to do acoustic "treatment" to walls and other boundaries.
Juhazi has summed it up nicely. What I do is very similar and results are consistent. I tend to do a slow figure 8 sweep with the mic trying to cover the listening area. Usually I count to 12, but if the area is larger, a longer time is needed. You don't want to move the mic too quickly.
As for checking post EQ, I have not found that to be easy if using computer EQ. You have to route the output of test singal (REW) thru your EQ. Not always easy, especially with ASIO drivers. If your EQ is outboard, then no problem.
As for checking post EQ, I have not found that to be easy if using computer EQ. You have to route the output of test singal (REW) thru your EQ. Not always easy, especially with ASIO drivers. If your EQ is outboard, then no problem.
While trying mmm I get OK results. But rather too loosey-goosey for my techie tastes.
About EQ, you need to start with a grasp of the shortcomings of FR you want to patch. From that point, you should work with a single unmoving mic location, preferably using a mic stand. You adjust the EQ (really a cinch with a Behringer DSP box): you iteratively adjust until you have tweaked the curve for that location according to the plan you had earlier chosen. Takes a bit of trial and error to introduce the tweak you want, but at least you are working with an unchanging benchmark.
Let's say you plan is to add 5 dB centred on 800 Hz, with a low Q. That's what you do with your single location.
Then you go back to the method you like for the big picture to see if that big picture now looks OK.
B.
About EQ, you need to start with a grasp of the shortcomings of FR you want to patch. From that point, you should work with a single unmoving mic location, preferably using a mic stand. You adjust the EQ (really a cinch with a Behringer DSP box): you iteratively adjust until you have tweaked the curve for that location according to the plan you had earlier chosen. Takes a bit of trial and error to introduce the tweak you want, but at least you are working with an unchanging benchmark.
Let's say you plan is to add 5 dB centred on 800 Hz, with a low Q. That's what you do with your single location.
Then you go back to the method you like for the big picture to see if that big picture now looks OK.
B.
In a few tries long ago, seemed to produce curves resembling the variable jumble you get from different locations of a mic. Surprised me that it worked at all. Today, I use REW mic averaging, combining three locations near my head location as the general FR. But that tends to be a minor interest for me, at least as compared to say sorting-out the many variables with DSP crossovers (which can be done with a single mic location pretty well).
Does "loose-goosey" need clarification?
B.
I see. What you call "loosey-goosey" I might call the point of the measurement. 🙂 The MMM is an average over an area. It's a tool just like any other, and similar to doing different locations and graphing them into REW. It's just that the MMM averages more points which can have advantages and disadvantages.
Way back when I was doing Dolby and THX alignments in cinemas, it was generally a 5 point or more average. With the the points spread that far across the seating area, you could at least identify what parts of the audience might have problems, and what the problems were. A moving mic measurement across the entire seating area would be a nice average, and would probably give a good idea of bass levels, but wouldn't tell you any specific trouble spots.
For a domestic system, I find that the MMM gives me a good view of the overall tonality of the listening spot. The graph looks like what I hear. No single point measurement ever does. That's its advantage for me. Single point, multi point and MMM are all tools in understanding what's going on.
Way back when I was doing Dolby and THX alignments in cinemas, it was generally a 5 point or more average. With the the points spread that far across the seating area, you could at least identify what parts of the audience might have problems, and what the problems were. A moving mic measurement across the entire seating area would be a nice average, and would probably give a good idea of bass levels, but wouldn't tell you any specific trouble spots.
For a domestic system, I find that the MMM gives me a good view of the overall tonality of the listening spot. The graph looks like what I hear. No single point measurement ever does. That's its advantage for me. Single point, multi point and MMM are all tools in understanding what's going on.
-What distanse is preferable for MMM? 1m or listening position?
- In what direction mic should be pointed? Ceiling or speaker (0 or 90 correction file you need to use).
- In what direction mic should be pointed? Ceiling or speaker (0 or 90 correction file you need to use).
MMM measurement should be done at listening position, scanning a volume of about 1X1X0.5m.
Better to use mic at 90° (facing to ceiling) with the corresponding correction file.
Thanks! I'll try out new method for me.
Can i use any type of movement? Just a shere in sweetspot?
Can i use any type of movement? Just a shere in sweetspot?
Last edited:
Last edited: