Room correction measurments with Mathaudio Room EQ

[AnalogSteph]

By the way do i need for my Superlux ECM999 also a calibration file because Superlux does not provide such file. However i saw that the freq respoinse was already quite flat of the Superlux.

The crux with cheap measurement mics is sample variation. They can stray quite substantially, above a few kHz in particular. Once calibrated, they tend to be perfectly usable, but you pretty much have to send them off to get them measured. (0°/90°.)

 

[SIY]

The crux with cheap measurement mics is sample variation. They can stray quite substantially, above a few kHz in particular. Once calibrated, they tend to be perfectly usable, but you pretty much have to send them off to get them measured. (0°/90°.)

This is indeed often true, especially with electrets, but the newer MEMS mikes tend to be a whole lot more consistent part-to-part.

 

[Snarfie]

The crux with cheap measurement mics is sample variation. They can stray quite substantially, above a few kHz in particular. Once calibrated, they tend to be perfectly usable, but you pretty much have to send them off to get them measured. (0°/90°.)

Not shure but this is how mathaudio copes with changing samplerates regarding measuring

Supports full range of sample rates from 44,056 kHz up to 384 kHz. All sample rates are supported without resampling to avoid any possible loss in quality.

https://mathaudio.com/room-eq.htm

 

[Snarfie]

Bit embarrassing first measurements with the MEM's mic where made not correct. Using a Yamaha MG102c which provides phantom power. I used the jack stereo out (4dbu) which probably resulted in a faulty frequency response (see earlier picture page 1). Now using the stereo line out signal result in a more recognizable frequency response. Still the result is better than the bypass (original) sound. Probably it is Psychoacoustic and/or in the ear of the beholder what you preference is.


The frequency response from the new Superlux ECM999.

Gray line is the original found response white line is Mathaudio correction. Did brought the frequencies around 50 to 80 Hz a bit up (small white arc) did sound better. Because mathaudio does not correct the bump around 160 Hz I compensated that manually (more or less) using the 31 band Graphic equalizer.

 

[Dogan]

Thanks for the comment. Yes i know REW problem is i can't get the REW file not in Foobar2000 or any other mediacenter like VLC in a simple plug&play manner as Mathaudio does in Foobar2000.

No expert here but you should be able create the filter in REW and export a filter impulse response file as wav (file>export). This file can then be linked to the foobar convolver component that should do the same trick rather easily.

I will give Mathaudio plugin a try to see what it does at some point.

Btw, do you know a simple parametric eq component for foobar? I tried a few in the past but the practical ones use bandwidth instead of Q factor that require some conversion and the rest were not user friendly.

Dogan

 

[Snarfie]

No expert here but you should be able create the filter in REW and export a filter impulse response file as wav (file>export). This file can then be linked to the foobar convolver component that should do the same trick rather easily.

I will give Mathaudio plugin a try to see what it does at some point.

Btw, do you know a simple parametric eq component for foobar? I tried a few in the past but the practical ones use bandwidth instead of Q factor that require some conversion and the rest were not user friendly.

Dogan

Ha ha I'm too not an expert in a long shot. A parametric eq component no i don't know such plug-in for foobar2000 (or any other device/software).

Lately i bought the IMF CM2 (Compact Monitor 2) I found it a big surprise that the linearity (see grey line of the CM2 around 3db headroom compared with white line) was way better in comparison with my old IMF Compact II (see grey line of the Compact II around 9db headroom compared with white line). Basically from a listning point of view Mathaudio did not made the sound of the CM2 much better. However the IMF Compact II using room correction or bypass mode is spectacular it became another speaker. So my conclusion for now is that a speaker with a horrible frequency curve does benefit much more from room-correction than a speaker with an already strait frequency curve. I'm now only puzzled what roomcorrection roll is in a better or worse frequency curve of any speaker taking in account your specific roomacoustics. So are there speakers with a near perfect frequency curve despite your room accoustics.

ps. if you gone use Mathaudio Room Eq let us know your experience an/or some pictures of your found Frequency response.

IMF CM2

IMF Compact II

 

[Dogan]

Ha ha I'm too not an expert in a long shot. A parametric eq component no i don't know such plug-in for foobar2000 (or any other device/software).

Lately i bought the IMF CM2 (Compact Monitor 2) I found it a big surprise that the linearity (see grey line of the CM2 around 3db headroom compared with white line) was way better in comparison with my old IMF Compact II (see grey line of the Compact II around 9db headroom compared with white line). Basically from a listning point of view Mathaudio did not made the sound of the CM2 much better. However the IMF Compact II using room correction or bypass mode is spectacular it became another speaker. So my conclusion for now is that a speaker with a horrible frequency curve does benefit much more from room-correction than a speaker with an already strait frequency curve. I'm now only puzzled what roomcorrection roll is in a better or worse frequency curve of any speaker taking in account your specific roomacoustics. So are there speakers with a near perfect frequency curve despite your room accoustics.

ps. if you gone use Mathaudio Room Eq let us know your experience an/or some pictures of your found Frequency response.

IMF CM2

IMF Compact II

I installed the mathaudio foobar plugin and took more than 10 measurements to set it up. I have a 2.1 setup in a very messy living room. The subwoofer is a B&W PV1D that was purchased because of the looks and it is placed near the right speaker. That might possibly explain the dip between 40-80 hz at the right channel and the absence of it at the left side.

Foobar is acting as a upnp server and the audio is processed while streaming.

I tried to compare a few different dsp outputs: the response as is, convolution filter I previously created with REW and 2 mathaudio plugin outputs with different cutoff levels (-5db and -12db). I was planning to post a few more screenshots but to my surprise they turned out to be pretty much identical in response so I only attached one shot showing all the pressure levels.

It was quite tricky for me to take the measurements since I was playing a REW generated sweep file over the foobar upnp server located in another room and getting the speaker responses from a laptop in the living room so I may have screwed up at some point. The results are consistent with each other though.

In the end, I reached to the conclusion that I can't really tell the difference between the convolution filter and the mathaudio outputs They both sound pretty good when you listen Miles&Coltrane playing together.

Dogan

 

[Snarfie]

I installed the mathaudio foobar plugin and took more than 10 measurements to set it up. I have a 2.1 setup in a very messy living room. The subwoofer is a B&W PV1D that was purchased because of the looks and it is placed near the right speaker. That might possibly explain the dip between 40-80 hz at the right channel and the absence of it at the left side.

Foobar is acting as a upnp server and the audio is processed while streaming.

I tried to compare a few different dsp outputs: the response as is, convolution filter I previously created with REW and 2 mathaudio plugin outputs with different cutoff levels (-5db and -12db). I was planning to post a few more screenshots but to my surprise they turned out to be pretty much identical in response so I only attached one shot showing all the pressure levels.

It was quite tricky for me to take the measurements since I was playing a REW generated sweep file over the foobar upnp server located in another room and getting the speaker responses from a laptop in the living room so I may have screwed up at some point. The results are consistent with each other though.

In the end, I reached to the conclusion that I can't really tell the difference between the convolution filter and the mathaudio outputs They both sound pretty good when you listen Miles&Coltrane playing together.

Dogan

Thanx for your measurments. I'm planning to do one more with Sonarworks (something like 22 measurments requierd). I guess there will not be a big difference. By the way i did returnd the CM2's dispited they measurd much better than the Compact II but to my suprise they where lacking low in comparsion with the Compact II. Jazz was sounding great but some pop and soul/funk tracks the CM2's are much more unforgiving sounding sometimes almost harsh (ha ha true monitors i guess). So even the CM2's where measuring way better/flatter it is still in the ear of the beholder to cope with that.

 

[sophie smith]

Just to clarify....
Is there still no way to use correction file created with REW into MathEQ integrated with Foobar2000?
I do have a correction file I created with REW. But I use FB2000 for playback.

Do you think REW and Math EQ are equally useful, or one is better?

Thanks

 

[QMuse]

Just to clarify....
Is there still no way to use correction file created with REW into MathEQ integrated with Foobar2000?
I do have a correction file I created with REW. But I use FB2000 for playback.

Do you think REW and Math EQ are equally useful, or one is better?

Thanks

Yu don't need MathEQ. IIRC there is a DSP convolution plugin for FB2000 which can use filters in WAV format so you have to create a filter from REW like this:

 

[sophie smith]

Yu don't need MathEQ. IIRC there is a DSP convolution plugin for FB2000 which can use filters in WAV format so you have to create a filter from REW like this:

Got it. Thanks!

View attachment 57334

 

[Snarfie]

Yu don't need MathEQ. IIRC there is a DSP convolution plugin for FB2000 which can use filters in WAV format so you have to create a filter from REW like this:

View attachment 57334

Can u you use convultion systemwide? Also the difference/ working between mathEQ an REW is not known exactly. If you Read the Working of MathEQ atleast you see what it does like avoiding preringing phase an amplitude issiues https://mathaudio.com/room-eq.htm

If u use the paid version mathEQ Will work systemwide. I am i favour to have a sort of (subjectieve) test between the several applications[/URL].

 

[QMuse]

Can u you use convultion systemwide? Also the difference/ working between mathEQ an REW is not known exactly. If you Read the Working of MathEQ atleast you see what it does like avoiding preringing phase an amplitude issiues https://mathaudio.com/room-eq.htm

LOL Avoiding pre-ringing is only a marketing gymmick as there is no pre-ringing when you correct only amplitude and not phase. So sure, you can as well say that with REW filters you are also avoiding pre-ringing.

You still didn't manage to make measurements in REW, right? Once you do you should check impulse and step response and learn what pre-ringing is.

 

[Snarfie]

LOL Avoiding pre-ringing is only a marketing gymmick as there is no pre-ringing when you correct only amplitude and not phase. So sure, you can as well say that with REW filters you are also avoiding pre-ringing.

You still didn't manage to make measurements in REW, right? Once you do you should check impulse and step response and learn what pre-ringing is

Did bought a simple mic amlifier without tone control before i used a Yamaha mc102 so i will do another atempt to work with REW. For me a one stop shop as MathEq is the solution an lots of people that can not manage REW and/or can't convert it result to applications like Convultion basicly (not for you an probaly 80% of ASR members) to work with REW imo is to complex for the avarage people. With MathEQ you are up an running in minutes without calibrating you mic etc etc. And the results looks similair with other aplications like REW what i can find online.

Regarding pre-rininging no i am not an expert. I took out one of the points that MathEQ describe on their website. MathEQ describe quite openly what it does atleast compard with other aplications like Dirac, Lyngdorf etc etc. On ASR their is a thread that handels pre-ringing https://www.audiosciencereview.com/...ging-with-linear-phase-room-eq-filters.11121/ and some see some issiues with it.
Quote:
"A correction filter for excess phase (= time correction) will introduce a pre-ringing.
A pre-ringing is allowed as long as the corrected step response does not show up a pre-ringing or if the pre-ringing is small enough. So pre-ringing is simply allowed if you don't recognize it.
If there is too much pre-ringing you will recognize it. So e.g. a simple drum hit = tok may sound like whoop-tok."

Reason why mathEQ avoids pre-ringing:

Quote:
"Avoids the pre-echo (pre-ringing) problem of conventional convolver-based room correction systems. The absence of pre-echo ensures the neutrality of the sound.

So basicly no pre-ringing is favourble i guess.

 

[Dididi]

If you want to have the correction system-wide, then you can start having a look here: AutoEq
For Windows, you can use EqualizerAPO plus Peace as a frontend.

Or very basic but fast: Use The built-in room correction you find under Audio properties/Enhancements/Room Correction. Definitely better than nothing.

 

[ElNino]

LOL Avoiding pre-ringing is only a marketing gymmick as there is no pre-ringing when you correct only amplitude and not phase.

This is untrue. Pre-ringing is an inevitable result of linear or mixed phase EQ. Mathaudio is a minimum phase EQ system.

I'm not saying one is superior to another (although I have my views), but this is a technical-oriented forum, so we might as well be accurate when talking about DSP.

 

[QMuse]

This is untrue. Pre-ringing is an inevitable result of linear or mixed phase EQ. Mathaudio is a minimum phase EQ system.

I'm not saying one is superior to another (although I have my views), but this is a technical-oriented forum, so we might as well be accurate when talking about DSP.

Read again what i said: "there is no pre-ringing when you correct only amplitude and not phase". What you said is true, but when you don't do phase EQ there will be no pre-ringing, so you are safe when doing only amplitude correction, like with Mathaudio and REW. Pointing out that there will be no pre-ringing when not doing phase correction but only correcting amplitude with simple PEQs is nothing but a poor marketing gymmick.

 

[ElNino]

Read again what i said: "there is no pre-ringing when you correct only amplitude and not phase". What you said is true, but when you don't do phase EQ there will be no pre-ringing, so you are safe when doing only amplitude correction, like with Mathaudio and REW. Pointing out that there will be no pre-ringing when not doing phase correction is nothing but a poor marketing gymmick.

This is not correct. If you correct amplitude but maintain linear phase, it's impossible not to have pre-ringing. This is signal processing 101.

 

[QMuse]

This is not correct. If you correct amplitude but maintain linear phase, it's impossible not to have pre-ringing. This is signal processing 101.

Mathaudio uses simple minimum-phase PEQs, just like REW.

With minimum-phase filters, such are simple PEQs, the phase(dotted line) WILL change but you will NOT get pre-ringing.

This is step response of that minimum phase filter:

This is the same filter, but this time created as linear-phase. This one will NOT change the phase but WILL cause pre-ringing:

Remember however that I was talking about simple minimum-phase PEQs used in Mathaudio and REW, so your linear phase remark was out of that context. So again, emphasizing that Mathaudio is avoiding pre-ringing when using minimum-phase PEQs is indeed nothing but poor marketing gymmick.

Ok, class dismissed due to lunch break.

 

[Alec Kinnear]

Thanks for your efforts to explain in more detail pre-ringing QMuse. I've just jumped in with Mathaudio Room EQ. I've found it works best with absolutely minimal adjustment. My B&W 603 S3 have a very nasty double hump at 100 Hz and 130 Hz which can lead to overwhelming mid-bass. I'd been compensating by boosting the subwoofers (stereo) and then having a very steep slope down through mid-range and treble. That sometimes sounded good (there was adequate low-end, not just mid-bass) but often sounded excessively bass-heavy (it was). I've dialled down the subwoofers and just use MathAudio Room EQ to tame the mid-bass. If I push harder to flatten the response, music becomes very tinny.

Happily the REW end result looks much better than what MathAudio Room EQ shows (MAREQ calibration superimposed on the after REW measurement).

What is curious is that although I'm not really directly touching the regions where vocals are located, I find some vocals are now much, much clearer. Isn't that the consequence of the pre-echo reduction which MathAudio touts:

In the case of my vocals, while asking the question I've found a possible answer. By cleaning out the huge mid-bass hump, the B&W main kevlar drivers are not being driven as hard leading to much cleaner mid-range (vocals). Makes the spend on MathAudio Room EQ more palatable (for some of my music Room EQ makes the sound less embracing and stereo imaging less 3D, I'm guessing that any EQ system not just MathAudio would but Dirac and ARC would have the same consequences for stereo imaging, nature of the beast). Less muddy vocals is a good thing and for which it is perhaps worth sacrificing some depth of soundstage. I haven't decided yet. Speakers with no mid-bass hump would be better. Stereophile measured a new pair of B&W 603 S3 with almost flat response in 2005 so perhaps any speakers would suffer the same issue in my office space.

Anyway my question to you is about the pre-ringing and EQ correction. Is minimum-phase PEQ enough for room/speaker correction? What are the alternatives and does changing phase end up improving the sound or just making DSP EQ more complicated?