miniDSP UMIK-2 USB measurement microphone (also ambient noise floor, measurements of my system using Dirac Live with Paradigm + SVS subs)

Well, as mentioned previously, it was a great holiday season and here's another gift from Audio-Santa 🙂 - the miniDSP UMIK-2 measurement microphone. I bought this through Solen.ca here in Canada and they did a great job with the order and quick shipment. Amazon Canada often has limited stock of miniDSP products.

I'm sure the miniDSP company is well known to many audiophiles at this point. I bet many of you, like myself, already have one of their UMIK-1 microphones which I've used in a number of the acoustic measurements on this blog since 2019.

As you can imagine, the UMIK-2, initially released in late 2020, is an evolution of the UMIK-1 with upgrades which we'll talk about. First, let's have a peek at the physical device itself.

Here's what's inside the box:

Compactly packaged inside, we have the microphone itself - a matte black metal unit about 7.25" long and shy of 1" widest diameter. It's connected through USB-C, a small tripod with microphone holder/clamp, foam windscreen, 2m shielded male USB-A to USB-C cable, and info sheet to access online details including the manual and driver software are also in there.

Here's a look at my UMIK-1 beside the UMIK-2:

In terms of construction, the metallic gray UMIK-1 looks almost the same as the UMIK-2. Without putting them side-by-side, I might not be able to tell the difference. The UMIK-2 is slightly longer and wider diameter, whereas the 2019 UMIK-1 has a mini-USB port rather than the USB-C. Notice that both have a bluish LED to indicate power, and there's a serial number sticker on the body of the microphones for easy access. You'll be using that serial number on the UMIK-2 webpage to download the calibration files. There are two files; one for 0° on-axis "free field" measurements which might be more useful for stereo systems with mic pointing towards the speaker and the other with mic at 90° "diffuse field" pointing up to the ceiling useful for surround/multichannel setups.

Beyond the exterior, what has improved under the surface?

Quite a lot! The omnidirectional microphone capsule is now larger, increasing from 1/4" in the UMIK-1 to 1/2" in the UMIK-2. The updated microphone is said to achieve significantly better low-noise and lower distortion performance. miniDSP has also beefed up the electronics with an updated 32-bit ADC, and the XMOS XUF208-128-QF48-based USB interface is an asynchronous USB Audio Class 2 device. It should be plug-n-play with modern Windows, Mac, and Linux computers. Unlike the 16/48 UMIK-1, this microphone can record at 24/32-bits resolution and sample rates ranging from 44.1kHz up to 192kHz. An ASIO driver is also available in Windows for low-latency recordings.

There are a couple of areas in the specifications to be mindful about where the UMIK-1 might have an edge. One is with the maximum SPL of the UMIK-1 at 133dB compared to the lower UMIK-2 at 125dB. For typical room measurements this is not a problem. It's highly unlikely that I'll ever need to measure extreme volumes like >100dB (anyone recording at a rock concert?)! The other thing to be mindful about is that the larger 1/2" capsule could become more directional at higher frequencies (high frequencies = shorter wavelengths, approaching the size of the diaphragm could result in diffraction).

Let's have a look at the opening lines of the calibration file for the UMIK-2 (serial number blurred to protect the innocent):

"Sens Factor" refers to the calibration factor for loudness (referenced to 1kHz, 94dB SPL). "AGain" refers to the amount of internal analogue gain being applied to the microphone signal. Unlike the UMIK-1 where you could change this internal gain with dip switches:

UMIK-1 gain dip switch for reference.
Original images from here.

The UMIK-2 has a controllable gain (range 0 to +42dB) which can be adjusted with their "miniDSP-MIK" software. You'll have to contact miniDSP if you want a copy of this. They quickly responded within a day. The program is free, but you have to agree not to disseminate it.

In the unlikely event that you must measure insanely loud levels approaching 120dB SPL, you can use the control software to turn it down to 0dB gain. Needless to say, if you fool around with the gain level, make sure to edit the calibration file so that measurement software like REW can accurately show the correct amplitude. For example, if I go from +18dB default to +8dB, I'll need to change the "Sens Factor" down to -23.61dB to reflect the -10dB in order for REW to report the accurate level.

A few measurements...

Obviously, measurement microphones aren't the most exciting thing to talk about for audiophiles, but they're clearly important for those of us running tests and doing acoustic room calibrations!

To be clear, you don't need to spend more than twice the price of the UMIK-1 for the UMIK-2 for most applications. The UMIK-1 is already great for room DSP calibrations and clearly better than the inexpensive measurement microphones that typically come with AV receivers for Audyssey, YPAO, Dirac, etc. auto-calibrations. At normal listening levels like say average 65-85dB SPL, the UMIK-1 will be accurate.

Let's run a comparison sweep between the UMIK-1 and UMIK-2 with miniDSP's calibrations applied in my room:

Good correlation. The two UMIK microphones are pretty much spot-on through the mid-range with a little more deviance into the lower bass and upper treble where human hearing also isn't as sensitive. The difference is arguably insignificant for our hobbyist pursuit; for example at the 10kHz cursor position, the difference was around 2dB. As you may know, there are 3rd party companies like Cross-Spectrum that sell UMIK-1 and UMIK-2's with more precise calibration files for those who need this.

Where we can see the superiority of the UMIK-2 is in significantly lower self-noise. With the microphones connected to my fanless miniPC in the quiet sound room at night, I took some thick blankets and wrapped them around the microphones to see what "silence" looks like! This will give us an idea of the "self-noise" in a very quiet environment:

As you can see, the UMIK-2 performs significantly better than the UMIK-1 as expected. Notice the 120Hz cursor peak which is likely a result of some 60Hz mains-related 2nd harmonic noise probably through the USB interface itself. The UMIK-2 picks up a low ~18.5dB SPL level which is about -10dB better than the UMIK-1. At 1kHz, the difference is even more dramatic with >20dB difference.

Given the improved noise level, the UMIK-2 can be used to better capture the ambient noise level in a quiet sound room. Here's what I get in my sound room in the evenings when I would be typically listening to music; microphone placed at the "main listening position" (MLP, aka "sweet spot"):

The 120Hz electrical hum is there but nice to see everything is below 30dB SPL. The UMIK-1 would be struggling right at its limits to capture this accurately due to higher self-noise.

A-weighting cuts off the highs and lows while accentuating the mid-range (2-5kHz especially) which is more sensitive with human hearing and typically the value quoted for room ambient noise. I would consider <30dB(A) a nice target and I'm certainly pleased with the 25dB(A) result for my domestic hi-fi audio basement listening space (also discussed previously way back in 2014).

[For further discussions exploring other standards like the "Noise Criterion" (NC) curves, check out this thread on Steve Hoffman's Forum. RIP Ingenieur / Arthur who started the discussion, passed recently.]

The ambient noise level is one of those essential parameters to consider when we talk about the quality of the sound room. Be careful about claiming to hear nuances arising from a "black background" unless you're sure you have a very quiet room! I have certainly never been in the presence of such a quiet room at audio shows or at the local audio dealer. 

[Likewise, make sure as audiophiles we have an idea about our auditory acuity by doing a hearing test before claiming to possess Golden Ears; ready to tell the difference between Nordost cables! 🤣]

BTW, notice that the ambient noise graph above was captured up to 48kHz. This implies that I was running the UMIK-2 at the "hi-res" 96kHz samplerate.

Speaking of high samplerates, let's look at another set of sweeps which is my current room set-up with dual subwoofers:

As you can see, normally with DIRAC Live Bass Control DSP turned on (green), the frequency response is significantly flatter than with it turned off (blue). The treble roll-off is dictated by my target curve (in this case, using Dolby Atmos Music Target). 

Since I can capture using the UMIK-2 further into the ultrasonic spectrum, I was curious whether at my listening position, with the Paradigm Signature Reference S8 v.3 speakers I could detect any notable tweeter break-up with the beryllium drivers. 

Notice that if I turn off all DSP (red), there's no steep 24kHz roll-off (internal DSP running at 48kHz) and what we're now seeing is the natural ultrasonic roll-off from the speakers' tweeters. It measures smoothly at the main listening position without any harsh anomaly or tweeter break-up modes. 

Since I'm showing room sweep results with the dual-subwoofer set-up (Paradigm SUB1 + SVS SB-2000 Pro discussed last week), let's have a look at the spectrogram:

I turned the volume up and used a Dolby Atmos Music Target with +3dB bass boost which sounds great in action movies. Note the peak bass level at 93dB SPL as per the scale on the right. Also note the cursor at 500ms and 50Hz. The spectrogram is another way to view the time domain performance like the waterfall plot but using colors to represent intensity of the sound.

With infrasonic extension down to 15Hz, the spectrogram looks well-controlled without excess energy exceeding 500ms except below 50Hz. Notice the null around 55Hz which is accentuated by my ottoman/table/footrest in front of the sofa - necessary sonic sacrifice for convenience. 😉

Finally, using that data above, let's calculate the reverberation time (RT60 decay) curves in this room:

I'm not a fan of excessively "dead" (anechoic) listening rooms. So long as the Topt stays below 500ms across the spectrum, I probably would not complain.

The UMIK-2 microphone capsule can be removed, allowing you to change the transducer portion of the microphone for compatible 60 UNS-threaded units. It's compatible with other pre-polarized electret condenser capsules (like these?); obviously if you replace, it'll need to be accurately recalibrated.

Summary:

Well, if you're looking for a good microphone these days for acoustic measurements and want something with lower noise performance, the miniDSP UMIK-2 looks like a great choice and will compete with more expensive measurement mics. Shop around as I see the US Amazon price is currently listed as around US$300 while it's US$195 on the miniDSP website, I presume this is in part a reflection of the tariff situation as it's shipped from miniDSP through Hong Kong.

In use, the UMIK-2 is certainly a good no-nonsense option that's convenient with the embedded high quality, high speed, USB 2.0 interface. I have used this device with REW and have had no issues with DIRAC Live on the Windows platform.

On my system, the calibration file correlates well with the UMIK-1 (90° calibrations compared) and clearly the UMIK-2 noise level is significantly lower across the audio spectrum by around 10dB or more at the default +18dB internal gain.

With that improvement in noise level, this allowed me to take a reading of the ambient noise floor in my sound room as a reminder that IMO, a truly "hi-fi" audiophile sound room needs to be below 30dB(A) SPL. This will provide at least 70dB of dynamic range for peak levels to 100dB SPL. I keep my average listening levels modest around 75-80dB SPL typically other than the occasional need to "pump up the jam". 🙂

As usual, I encourage you, dear audiophiles, to be involved in exploring your sound system in a rational, objective manner to look for anomalies and qualities that might not be easily audible. A calibrated measurement mic will quickly and accurately tell us things like frequency response anomalies, time-domain decay issues with a waterfall graph or the spectrogram, as well as detection of distortions.

I would hope that those who offer themselves as experts in hardware sonic quality (like these guys) have the capacity to test out their own systems and be able to answer questions with fundamental technical knowledge commensurate with being involved in the "high fidelity" pursuit. Otherwise, high price tags are just reflections of merely luxury "high end".

--------------------

Let's end with a great live recording of Hugh Masekela - "Stimela (The Coal Train)" off Hope (1994, DR15, available also as hybrid SACD):

I hope you're all enjoying the sound and music!