Calibrated Microphone - Would Like Opinions

[whoareyou]

I just received a new UMIK-1 Cross Spectrum calibrated microphone with calibration that goes out to 25khz

Measurements at approximately 20khz and above seem "off"?
For example, take a look at the following portion of measurement (I've taken many measurements and all look similar) along with the mic calibration file
To me the calibration doesn't seem right. My front 3 speakers are rated 20 - 23khz (not sure about the sides), but I'd be surprised if any of them put out anything meaningful at 25khz.
I'm waiting to hear back from XSpectrum, but I'd like to get opinion of more experienced people regarding this.

Does this seem reasonable?
My measurement sweep was configured to end at 24khz, so should this now be configured to end at 25000 to match the microphone calibration file?
Should I modify the calibration file to match the frequency range specs for this umik-1 microphone? I believe it's rated to 20000hz?

Of course, I can't hear any of this

Corresponding mic calibration info

 

[jjazdk]

That is a valid concern.
Considering that the calibration applies almost 25dB of gain at 25kHz compared to 10kHz, no I would not trust measurements at 25kHz.
Have you compared the original microphone calibration to the Cross Spectrum one?

 

[whoareyou]

That is a valid concern.
Considering that the calibration applies almost 25dB of gain at 25kHz compared to 10kHz, no I would not trust measurements at 25kHz.
Have you compared the original microphone calibration to the Cross Spectrum one?

Yes, I compared them, but UMIK-1 only provides calibration data to 20kHZ.

 

[Fon]

My mic calibration and measurements look the same as yours, Umik-1 Cross Spectrum..

 

[whoareyou]

My mic calibration and measurements look the same as yours, Umik-1 Cross Spectrum..

Interesting and good to know, but barring an explanation (and maybe it's a simple explanation) it still seems wonky to me.

 

[jjazdk]

It is wonky, because the microphone was not made for extension to 25kHz.

Interesting and good to know, but barring an explanation (and maybe it's a simple explanation) it still seems wonky to me.

 

[whoareyou]

It is wonky, because the microphone was not made for extension to 25kHz.

That's exactly what I think.

Guess there's a reason MiniDSP only goes out to 20kHZ with their data. Looks like I'll be trimming down the calibration file.

Thanks @jjazdk

 

[BenToronto]

Except for doubts about super-sonic frequencies, the calibration of these mics (and similar non-USB) is so close to flat that hardly worth fussing about using any correction (esp since the room acoustic tests are vastly wilder).

BTW, I''ve compared the mics in a Mac laptop to my individually calibrated mic and not much different, north of the bass region. And, since it is testimony of your ears that ultimately matter and since testing tends to be iterative, the mic reading is just a relative benchmark anyway to examine as you adjust the DSP step by step.

B.

 

[whoareyou]

Except for doubts about super-sonic frequencies, the calibration of these mics (and similar non-USB) is so close to flat that hardly worth fussing about using any correction (esp since the room acoustic tests are vastly wilder).

True, I can't hear these frequencies so from that standpoint I am not concerned, but they influence simulations and correction filter loss, and that I find important when generating filters.

BTW, I''ve compared the mics in a Mac laptop to my individually calibrated mic and not much different, north of the bass region. And, since it is testimony of your ears that ultimately matter and since testing tends to be iterative, the mic reading is just a relative benchmark anyway to examine as you adjust the DSP step by step.

To which portion of the DSP are you referring? For example, do you believe that target curve is more important than an accurate measurement?

 

[BenToronto]

Funny thing, I've been contemplating this all morning.

I'd say the goal is to sit and hear satisfying music. Which I think means a "house curve" that produces that satisfaction (at at least one loudness level, since everybody just hates variable loudness compensation even with 4-gang volume controls or fantasies of achieving it with miniDSP, eh).*

I believe at present, there is no single or multiple physical measurements that adequately capture human hearing, as much as I admire Toole's Spinorama. Certainly not any Spinorama data that comes in the shipping box, so to speak. That's because the sum of the acoustic parts in your music room can not be calculated. But REW testing provides a snapshot of the room from which you can further tinker.

If I took a photo of you, would I have a much of a picture of you? Likewise, truly awful meaningless sound if you listen to a recording of music from your measurement mic in its testing location.

I am thoroughly committed to measurement, physical if possible but also human perception, as key to sorting out HiFi systems.

B.
* Hint: half the readers of this post have hearing deficiencies and most don't know it. Not sure how that works with establishing a house curve, hearing aids or not?

 

[jjazdk]

Except for doubts about super-sonic frequencies, the calibration of these mics (and similar non-USB) is so close to flat that hardly worth fussing about using any correction (esp since the room acoustic tests are vastly wilder).

BTW, I''ve compared the mics in a Mac laptop to my individually calibrated mic and not much different, north of the bass region. And, since it is testimony of your ears that ultimately matter and since testing tends to be iterative, the mic reading is just a relative benchmark anyway to examine as you adjust the DSP step by step.

B.

I disagree. For most microphones, the variations in their frequency response is much bigger than what an objective measurement can tolerate, therefore there is a need for the calibration file.

And no, the microphone in your Mac laptop (or any other laptop for that matter) is not even close to the measurements you can make with a proper microphone.

 

[BenToronto]

I disagree. For most microphones, the variations in their frequency response is much bigger than what an objective measurement can tolerate, therefore there is a need for the calibration file.

And no, the microphone in your Mac laptop (or any other laptop for that matter) is not even close to the measurements you can make with a proper microphone.

As someone who has actually done measurements in the famous Bell Labs anechoic chamber and in his little music room too, I can claim a bit of insight into measurements. Whatever "objective measurement can tolerate" could possibly mean in the context of home acoustics, your meaning must be astonishingly persnickety, at least compared to most of the rest of us.

Other than the most extreme ends of my calibrated mic, the calibration correction is barely a dB or so from flat. And my room can be way over 10 dB, even shifting the mic a few inches. So a waste of time to cry about what is "proper".

B.

 

[Ofer]

As someone who has actually done measurements in the famous Bell Labs anechoic chamber and in his little music room too, I can claim a bit of insight into measurements. Whatever "objective measurement can tolerate" could possibly mean in the context of home acoustics, your meaning must be astonishingly persnickety, at least compared to most of the rest of us.

Other than the most extreme ends of my calibrated mic, the calibration correction is barely a dB or so from flat. And my room can be way over 10 dB, even shifting the mic a few inches. So a waste of time to cry about what is "proper".

B.

This is more experience then most (if not all) of the users here. Just that today finding a measurement mic with a calibration file under 100$ isn't so difficult for both USB and XLR options. So there is really no reason to pass on the option.

 

[juicehifi]

I strongly recommend individually calibrated microphones. Significant frequency deviations that are audible occur regularly and it is more or less impossible to account for the deviation by ear. The second frequency chart posted by OP here has deviations between 20 Hz and 10 kHz that are likely to be audible. And based on my own experience I'd say that it is almost impossible to correct by ear the errors introduced by a microphone that isn't properly calibrated.

The capsules used for measurement microphones vary substantially from batch to batch and also significantly within batches. Therefore, an individual calibration is needed.

 

[Anatoliy Gavrilov]

Corresponding mic calibration info

The attachment contains the wrong calibration file - there the SPL is raised by 25 kHz by + 47 dB. This cannot be.
I suggest setting the microphone calibration to "flat" in the program settings. Another option is to take another measurement microphone for comparison.

My front 3 speakers are rated 20 - 23khz

Then the upper measurement frequency must be turned on at 48 kHz. The sound card must be with a sampling rate of 192 kHz.

 

[BenToronto]

Yes, that is some good advice.

My Mac laptops are excellent north of the bass region (and I suppose, south of super-sonic). So for comparison, put your mic real close to the laptop mic and try it.

BTW, the absolute accuracy of mics is not important, just their repeatability when working iteratively with the basic standard which is the meat mic (AKA your ears). Whatever FR your ears like best, is the correct setting no matter what your mic says. Use your mic just as a short-cut to getting there.

B.

 

[whoareyou]

The attachment contains the wrong calibration file - there the SPL is raised by 25 kHz by + 47 dB. This cannot be.
I suggest setting the microphone calibration to "flat" in the program settings. Another option is to take another measurement microphone for comparison.

The attachment has the "correct" file. It is the reason for my post as it seems very much out of whack in those upper frequencies. I am waiting for explanation from x-spectrum as to what is going on here. Unfortunately, x-spectrum is in midst of office move, and they are very slow to respond these days.

 

[Anatoliy Gavrilov]

The attachment has the "correct" file. It is the reason for my post as it seems very much out of whack in those upper frequencies.

Yes, the second file is good, it contains SPL correction from 9 kHz to +7.9 dB, and it is for measuring at an angle of 90 degrees.
And now make a copy of the file and put 0 dB correction values in it above 9068 Hz. Then substitute this file in REW.

UMIK-1 only provides calibration data to 20kHZ.

Most microphones are advertised by manufacturers as 20-20000, although the microphone itself can measure much higher.

 

[whoareyou]

Yes, the second file is good, it contains SPL correction from 9 kHz to +7.9 dB, and it is for measuring at an angle of 90 degrees.
And now make a copy of the file and put 0 dB correction values in it above 9068 Hz. Then substitute this file in REW.


Most microphones are advertised by manufacturers as 20-20000, although the microphone itself can measure much higher.

Thanks for the suggestion. And yes, I understand that the microphone may be able to measure much higher, but I suspect this one does not.
As I mentioned, it is a crossSpectrum calibrated microphone and they are known for providing quality calibrations. I'll provide update when I hear back from CrossSpectrum.

 

[jjazdk]

As someone who has actually done measurements in the famous Bell Labs anechoic chamber and in his little music room too, I can claim a bit of insight into measurements. Whatever "objective measurement can tolerate" could possibly mean in the context of home acoustics, your meaning must be astonishingly persnickety, at least compared to most of the rest of us.

Other than the most extreme ends of my calibrated mic, the calibration correction is barely a dB or so from flat. And my room can be way over 10 dB, even shifting the mic a few inches. So a waste of time to cry about what is "proper".

B.

Seriously, you think your opinion carries more weight because you have been in an anechoic room?

I have access to anechoic rooms (yes, in plural) at my work everyday, but that does not mean my opinion is worth more than others. I am also a Senior Acoustics engineer, and have been so for many years.

Anyway, this shows the calibration for three different measurement microphones, and this is not an uncalibrated variance I want to have in my measurements.

 

[juicehifi]

Thanks @jjazdk,

Those charts give a clear illustration to the topic. Deviations such as these aren't good enough if you're looking for the best possible result.

And be aware that calibration files can start to malfunction after a while. The tips that hold the capsule are very sensitive to damage.

If you struggle getting good result with a measurement mic that is a few years old and have been just lying around, it may be something wrong with the microphone.

 

[whoareyou]

This may be of interest to some:

Hifi-Selbstbau - 1000 Mikrofonkalibrierungen - eine Übersicht

HiFi-Selbstbau - Das Online-Magazin

www.hifi-selbstbau.de

FYI - This company has calibrated 1000 microphones and presented results for all to see.

 

[whoareyou]

Also, the response from crossspectrum:

"What you’re seeing is normal. The issue is that the UMIK-1 has a fixed sampling rate of 48 kHz, which means that it has an upper frequency response limit of 24 kHz. As a result, the frequency response of the mic starts to roll off as it approaches the 24 kHz cutoff frequency. When the calibration file is applied to your measurement, you’ll likely see a huge spike above 20 kHz because the inverse of the roll-off is being applied as a correction. For that region, you’ll have to teak by hand."