Topic: Insta360 Mic Pro - New wireless mic with 32-bit stereo internal recording

[Ozpeter]

While doing a test recording, I accidentally dropped an aluminium ladder flat onto a concrete floor in the garage, quite close to the device.  I nearly jumped to the garage ceiling.  Obviously the waveform at that point was seriously clipped-looking, but lowering the level by 15dB in Audition reduced it to a nice rounded transient peak.  It looks like it should handle most real-world levels.

[Ozpeter]

Today's shortish (phew) report on this device -

- I was wrong about the LED lights being on all the time.  There's a switch in the phone app to turn them off (remotely of course).  That actually makes a significant difference to using it low profile, of course.

- I am now using a simple and cheap USB to USB OTG adapter to access the files on the device.  You don't have to turn on the device to read its content.  You can replay the audio with a phone that way using a suitable file management app to access the files via the OTG, and listen on bluetooth phones or earplugs. [Edited - I have now discovered that the short USB-C to USB-C lead supplied with the device is an OTG lead in itself, so that's all that needs to be used.  So I wasted $7AU on that adapter.  Huh. So really, playback is not that hard a task.]

- It's pretty easy to conceal the device - especially with the LEDs off! - underneath a magnet-friendly can or the like, or using the magnet clip to attach it beneath a paper carrying bag or similar.  Or under the peak of a baseball cap, again especially with the LEDs off!

- I'm still not back at home with my normal gear but I do get the impression that the device has a bass-heavy sound, but I used an eq app on my phone and that brought back plenty of the HF content - it's not missing, just a bit rounded off.  But in a day or two I should be able to do some slightly more objective testing of that aspect.

[Ozpeter]

I am almost certain that the mic array is actually the same as that on the Instamic.  The front facing mic aperture is quite wide, allowing space for more than one of those little MEMS devices.  The supplied windshield covers those mics and not the side ones, probably because if the side mics supply the stereo content, those using the mic just for transmitting audio to the receiver do not need to use those side mics.  They only come into place for recording 32 bit float stereo.  But the front mic, or mics, is used for creating various polar patterns, which I believe is what you can do with (say) 4 MEMS capsules.  So it really is a direct competitor with the Instamic.

I bet Insta360 would have called it the Instamic if Zoom/Instamic hadn't already bagged that name...

[Ozpeter]

At last I am back at base and able to be a bit more scientific about this device.  Well, slightly.

First, here is a frequency plot of a eight minute or so recording made with the device attached beneath the peak of my baseball cap.  I simply recorded some location sound when I walked into a local chicken takeaway place, ordered my chicken, stood around for a bit with some people talking in the space and background music playing, and then walked home along a suburban road with traffic passing.  To me, the plot looks quite even, with all those random frequencies I recorded not resulting in major peaks, and a good extension at the top and bottom of the frequency range.  Tomorrow I'll do some indoor tests using maybe white noise from hifi speakers - which won't reproduce the noise particularly flat - but I will compare what this device captures vs maybe the Zoom M2 and perhaps even the Sony A10 which many people seem to like.

[Ozpeter]

I tried recording a jazz big band with vocal track played on my hifi, using the Mic Pro, the Zoom M2, the Zoom H2e, and the Sony A10.  Below is the frequency analysis of the normalised files, with green = Mic Pro, red = M2, purple = H2e, and the A10 in yellow.  But I'm not sure it really shows much, compared with playback listening.  That seems to show that the Mic Pro has a wide stereo image, more than 120 degrees I'd say, and that it does benefit from a bit of added eq in a gradual slope up and back down from about 1kHz to 6kHz, and maybe a slight reduction in low frequencies below say 400Hz.  In other words, it lacks a little brightness and it's a bit bass heavy, but both are correctable.  And the stereo width might need a bit of narrowing depending on its placement in relation to the source.  But each of these 4 devices has its own sound, without any of them being horrible IMHO, just... different.

But once again, given the tiny size and long battery life, together with extreme simplicity of use, the Mic Pro remains in my opinion a remarkable little field / location recording device.  And it's cheap.

[Gutbucket]

Thanks for these reports Peter.  These little devices and others like them are quite interesting and promising.  Especially interesting to me at a fundamental level is how they enable modification of pickup pattern and stereo recording.  I'm interested in finding out how well they are able to do those things such as how accurate the shape of the derived sensitivity patterns across a wide frequency range.

- According to AI search, this device uses MEMS mics and associated techniques to obtain stereo recording from closely spaced omni mics.  I can't find where AI might be getting that info from, but the features of the mic do strongly suggest that it is indeed a MEMS device.

- If the device is set to record 32 bit float stereo internally, then you can't record it with any of the built in 'effects'.  It's just the pure stereo thing, which is fine.  If you set it to mono, and ask it to record that with processing, then you can set the mic array to record internally as if it was omni, cardioid, fig of 8, or narrow.  I've tried cardioid so far, and it certainly rejects the sides and back.  Now whether these settings would allow a pair of these to be used as a pair of cardioids or omnis instead of just one for single point stereo I'm not sure, in terms of how good the result would be.  But it's possible on the face of it.  Maybe one could even create an MS pair with one set to cardioid and the other to fig of 8.

I am almost certain that the mic array is actually the same as that on the Instamic.  The front facing mic aperture is quite wide, allowing space for more than one of those little MEMS devices.  The supplied windshield covers those mics and not the side ones, probably because if the side mics supply the stereo content, those using the mic just for transmitting audio to the receiver do not need to use those side mics.  They only come into place for recording 32 bit float stereo.  But the front mic, or mics, is used for creating various polar patterns, which I believe is what you can do with (say) 4 MEMS capsules.  So it really is a direct competitor with the Instamic.

As far as I'm aware, all individual MEMS elements are omnidirectional monophonic devices, and it's the combination of more than one of them in a single device along with sum/difference/phase-processing provides potential for lower noise and/or control of polar pattern. That implies that our discussion about using a pair of miniature omnis to produce a fig-8 sensitivity pattern (for use as a Side channel) or other patterns as speculated about a year ago in the Omnis to figure of eight thread may have somewhat more merit than originally received.  I've put off the experiments I intended to do with two closely spaced / somewhat baffled omnis in combination with phase-rotation processing afterward, but am not thinking of returning it it.

[Massive Dynamic]

But once again, given the tiny size and long battery life, together with extreme simplicity of use, the Mic Pro remains in my opinion a remarkable little field / location recording device.  And it's cheap.

Your results make me wonder if this mic has surveillance applications in mind.

[Ozpeter]

Here, at last, is my YouTube review, which includes a range of audio samples.  My YouTube channel is not monetized so including this link doesn't make me any money - sadly!

https://youtu.be/QpTtksHt0gg

As for audio surveillance - it is actually large compared with most wireless mics.  But for discreet recording of music events, it's indeed the smallest 32 bit float stereo multi-mic recording device.  In the video I did include some ambient recordings in a takeaway cafe and in a supermarket, but nothing the mic heard would not be heard by anyone else there.

Obviously there are many factors involved but in the video I did comment that compared to the Instamic, the Mic Pro does record 32 bit float stereo, has twice the memory, and twice the battery - and costs half the price.  But I don't think it's waterproof!

[Ozpeter]

.... Especially interesting to me at a fundamental level is how they enable modification of pickup pattern and stereo recording.  I'm interested in finding out how well they are able to do those things such as how accurate the shape of the derived sensitivity patterns across a wide frequency range.

The Mic Pro can do different patterns for mono internal recording.  In the app, if you set it to mono you can also choose "Rich" or "Bright", use auto gain setting ("Prevent audio clipping" or "Auto Gain Control") and choose "Sound direction" options including Omni, Voice Focus, Cardioid, Figure of 8.  And there's a Low Cut setting.  As these don't apply to stereo recordings, I haven't tried them, but I suppose I should.  However, most of the YouTube reviews of the device (video-centric ones) do show how those settings sound, eg by someone walking around the device while talking.  Polar patterns were well defined but they may be geared towards voice recording rather than music.

[Ozpeter]

I hadn't planned this but I have now done a quick YouTube presentation testing the mono polar patterns of the Mic Pro, which might interest "Gutbucket" if not anyone else.

https://youtu.be/PmB5LEH2Cc4

I am frankly a bit suspicious of how these polar patterns are created, in respect of this device anyway.  Maybe the Instamic goes about it differently, but it seems to me with the Mic Pro that it is chiefly concerned with voice recording in mono.  It is very good at behaving like "real" mics for voice, but the polar patterns have much less effect on music.  For instance, the most tightly focused setting (narrow pattern, shotgun imitation) discriminates against voice sounds other than right in front of the mic, but discriminates against music much less - and when the voice is heard from the front, music being played to one side starts sounding on axis, then further away if the voice stops.  Hmm.  This stuff might be very useful in certain voice recording situations but it may preclude use of a pair of these devices other than perhaps as spaced omnis, where it seems no actual processing of the signal is happening.

[Gutbucket]

Thanks Peter.  I'll watch/listen when I can with headphones. Was just typing this up when you posted the message above..

Before searching out the website I assumed the Mic Pro incorporated four MEMS arrayed symmetrically around the puck-shaped device facing the four cardinal directions, thus providing the capability for simple beam forming in any horizontal direction.  Turns out it uses an array of 3 MEMS oriented Center / Left / Right without a fourth in back, opposite the center position.  Hmmm. I can see how they might derive forward-facing cardioid and "voice focus" patterns (the latter presumably supercardioid like), by using the side MEMS as virtual "rear vents". However the lack of complete front/back symmetry of the array makes me wonder how they derive a symmetrical forward facing fig-8.

Presumably stereo mode is Mid/Side and derives a decently symmetrical fig-8 Side channel pattern via the two MEMS placed symmetrically on opposite sides.  But how might it derive a forward-facing fig-8 without a MEMS in back opposite the center microphone?  Perhaps in the same way they derive the cardioid and "vocal focus" patterns, only using a still higher ratio from the two Side MEMS, but I would not expect a fig-8 pattern derived in that way to be particularly symmetrical.  Maybe it isn't.

I looked at the website and manual but was unable to find any info about the orientation of the mono directional patterns..  It's not made clear whether on-axis is indeed inline with the center microphone position or one of the side microphones.  Because of that I presume on-axis is in-line with the center microphone, otherwise I'd expect a mention somewhere that in order to use the directional patterns the device must be oriented sideways and include indication of which side is the "front". 

[Gutbucket]

Turns out it uses an array of 3 MEMS oriented Center / Left / Right without a fourth in back, opposite the center position.

Noting that this is essentially the same MEMS array geometry as used in the Instamic, but I presume the spacing between elements in the Mic Pro is greater since the device is physically larger.

[Ozpeter]

The centre mic aperture of the Mic Pro is much wider than the apertures of the side mics.  And the wind muff that normally comes with the Mic Pro does not cover the side mics, which implies they are not used when the device is fulfilling its main role of a mono wireless mic.  The muff that does cover the sides is an optional extra (fortunately very cheap).  So, not that I know much about MEMS, but I strongly suspect that there is more than one MEMS capsule in the front facing mic aperture, which is a kind of grille rather than a series of holes.  If there was just one in there, and assuming I'm right that the side mics don't usually get used, then the variable polar patterns wouldn't be possible.  As I understand it, the Instamic has four capsules on top which are used in mono mode, and the capsules at the sides support stereo.  Perhaps the Mic Pro has kind of ripped off their design!  In terms of size, the Mic Pro seems to be a round version of the Instamic size, if that makes sense...

[Gutbucket]

Yes, the center position in the Instamic also uses multiple MEMS.  I think that's done to improve noise performance over what would be available from a single sensor that is physically small and thus inherently noisy.  Also means improved noise performance from sensors run in parallel applies only to mono mode, not stereo mode.  Since the unit does allow for 32bit float recording in stereo mode, post analysis should be able to show the actual dynamic range of the mono modes verses the stereo mode.

Do you know if it uses dual channel switching ADCs?  If it is, it might be operating like the Deity PR2 does, using the two ADCs in dual switching mode when recording in mono, while reverting to using them in single non-switching mode for stereo.  Can store output as 32bit float either way of course.