Replaced my UMIK-2 mic with an Earthworks M23R... I'm confused

Urs said:

I have a 2.0 setup in a difficult but acoustically well treated room. I wanted to improve or at least confirm my previous measurements done with a UMIK-2 USB microphone with a better measurement mic, that's why I bought an Earthworks M23R.
My audio interface is rather nice and robust, it's a RME Digiface AES (contains a Mic preamp).
With the UMIK-2 I used these devices in Audiolense: RME Digiface WASAPI for Playback and miniDSP ASIO for microphone input.
With the Earthworks I use RME Digiface ASIO for Playback as well as for microphone input.

I am using since quite some time two or three Target curves, which I like, also a few slightly different correction procedures.

After two days of careful measurements and trying different targets and correction procedures, I am confused and disappointed by the Earthworks M23R, but maybe I am missing a thing or two? All my correction filters based on measurements with the M23R are significantly (!) worse than those based on the UMIK-2. They are lifeless, rather veiled and dull.

I think the reason for this is the "hill" between 200 Hz and 10 kHz in the M23R measurement(s). This "hill" causes a large "valley" in the correction filter, which in the end might not help the sound. But I am only guessing and as I said... rather confused ;-)

(For both microphones I used the appropriate calibration files. Without using the calibration file the situation remains basically the same)

Click to expand...

Hi Urs,

I am not able to make much out of these screendumps. But I would like to take a look at one or ideally two measurements. Ideally two made on the fly, with different sweep duration. You can send me one or two measurement files if you want me to take a look.

Urs said:

Thank you Bernt
I attach three measurement files, two were done with the M23R and one with the UMIK2. All were done with a 10 seconds sweep. If it would be helpful, I can make another measurement with a different sweep duration on the weekend.

Click to expand...

Hi Urs,

I finally got around to look at your files.

The timing looks good on all three, so I do not really see any of the usual suspects here that can cause serious problems with the time domain correction.

But the frequency response is very different between the umik and the M23R. Either you have flipped one of the calibrations around or the umik is very much off. FYI, the mic calibration that you load into Audiolense should be the measuremed response of the microphone, not the inverse. Audiolense will inverse it. So perhaps check that the calibration file you loaded for the earthworks was the actual measurement and not its inverse. With that under control I am confident that you will experience a very good correlation between simulated responses and what you hear.

Here's the Earthworks calibration. As expected it doesn't do much. The Earthworks' tend to be very precise even before calibration.





Here's the UMIK



This calibartion file looks clean, and does not have typical reflection artifacts from the environment that sometimes can be found in lesser calibration.

Nevertheless, something is off with one of the calibrated microphones. My vote goes for the Earthworks.

To quote your sound quality asessment:
After two days of careful measurements and trying different targets and correction procedures, I am confused and disappointed by the Earthworks M23R, but maybe I am missing a thing or two? All my correction filters based on measurements with the M23R are significantly (!) worse than those based on the UMIK-2. They are lifeless, rather veiled and dull.

I think the reason for this is the "hill" between 200 Hz and 10 kHz in the M23R measurement(s). This "hill" causes a large "valley" in the correction filter, which in the end might not help the sound. But I am only guessing and as I said... rather confused ;-)

Your uncorrected speaker has low energy in the lower midrange and by comparison too much energy in the kHz area, centered around 1-2 kHz. When you take down the high energy and fill in with energy in the lower midrange, the sound will typically be experienced as less transparent - at least immidiately. And the sound stage can be perceived as having become smaller. But in the long run, you may notice nuances that has been masked before ... and perhaps that colorations/distortion you thought were on the recordings disappear.

Here's a correction of the stereo sample, which happens to be a pair of speakers I am still using :-)



The off-balance here before correction is somewhat similar to what you measure with your speakers. Before correction there is too little energy below 1 kHz and clearly too much above. But I was used to it, and it sounded great on some recordings. Without correction the speaker had a good and strong bass and a great deal of perceived transparency. But it could be a bit straining on recordings with a lot of hf energy.

It took me some time to get adjusted to the corrected sound. First it was less transparent, more veiled and more "cellar-like" where it had an "outdoor transparency" before. But when I got used to it, the perception flipped around. The pitches and the air came back to me.

Especially male voices and rhythm instruments (who has a lot of energy in the 100 Hz-es took on a more realistic body ... and the sound of "microphones and electronics" which we are used to hear on live venues and most domestic rigs completely disappeared on most of the recordings. Everything just sounded and still sounds more organic. And the bass became much faster, much more rhythmic. Actually the whole speaker became more rhytymic. The leading edge of bass notes need midrange energy to shine...

And eventually I started to hear nuances in the ambience that I hadn't heard before. The speaker behaved much more chameleon-like and discriminated between different recording ambiences like I'd never heard before. This was particularly beneficial on classical recordings and other high quality recordings. At some recordings the ambience became extremely deep an wide and long, at others there was hardly any ambience at all. But all of this was a reflection of the recordings and not the speaker + room.

My point is this: When you change the voicing so much during correction, and especially when you add energy to the lower midrange and remove energy from the presence / treble, the speaker will immediately sound less engaging. But if you have corrected towards a more neutral presentation, the sound will likely grow on you within a few days.

Quick A-B comparisons can be a bit misleading here, since your perception needs time to recalibrate.

Lastly it needs to be said that people prefer different voicing. I prefer a rasor straight target that falls a few dB. But many users - also seasoned users - prefer a gentle lift in the bass of a few dB and, by comparison to everything else, a slightly recessed lower midrange. So you need to figure out what your preferences are.

Perhaps you prefer some variation ofver this theme:

If you don't make too sharp curves along the target, you may be able to avoid colorations that are clearly audible and get an overall voicing that sounds really big and transparent etc.

But that will be easier with a microphone that you can trust and my money is on the Earthworks.

Please keep us posted!

THe comparison says it all, really. There's no reason to trust your M23 measurements after this.
And my compliments for your hearing skills!