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

[Ozpeter]

Tested the mic very unconventionally. [.....]

Probably needs a lot more tinkering but wow! 9/10. Only downside has been accessories.

That all sounds really interesting.  You observation about the low end being centred would explain why in the spectral frequency display in Audition the centre of the display is very bright.  

Two notes related to your post - sadly the mic input is mono only, even though the connector is a stereo type. And, polar pattern selection only applies to internal mono recordings.  In my latest YT video I explored the mono recordings a bit more from the perspective of using the Mic Pro for voiceover work at home, testing the various polar patterns, and I noted that internal mono recordings are not actually real mono - the files remain stereo and when there are gaps in speech, some stereo content is shown in Audition.

The key thing with this device is that it's using MEMS mics, almost certainly basically omnis, and the combined output either as stereo or multi-pattern mono is artificially constructed - 3 omnis very close together can't naturally do this in traditional unprocessed ways.

[Ozpeter]

I'm hesitant to post something so long, but I think many participants in this thread will find it helpful in understanding the workings of this fascinating device.

This is a really interesting chat I had with Gemini about how the Mic Pro works internally - starting from how the Sony M10 works, which struck me as being quite relevant.

Q - "The Sony M10 audio recorder had quite closely spaced omnidirectional mics. They produced a rather poor stereo image with low frequencies particularly in the centre, and some high frequencies at the sides. Is this a natural consequence of using closely spaced omnidirectional microphones in this way?"

A - "Yes, this is a direct, textbook consequence of the physics of closely spaced pressure transducers (omnis). What you are hearing is a physical limitation of the extremely narrow capsule spacing (approximately 58 mm on the PCM-M10) combined with the high-frequency directivity of omnidirectional capsules.

The phenomenon breaks down into two distinct acoustic mechanisms:

1. Low Frequencies: The Time-Arrival (ITD) Deficit
Spaced omnidirectional arrays (A-B stereo) rely entirely on Interaural Time Differences (ITD)—phase and arrival-time variations—to build a stereo image. Unlike directional mics (which capture level differences, or ILD), omnis have no inherent directional patterns to exploit at low and mid frequencies.
At a capsule spacing of just 5.8 cm, the distance is mathematically too small to create meaningful time-arrival differences for long wavelengths:

• For a 100 Hz wave, the wavelength is roughly $3.4\text{ meters}$. A 5.8 cm spacing represents a mere 1.7% of a single wavelength.
• Because the sound wave hits both diaphragms at virtually identical phases and identical amplitudes, the signals in both channels are nearly identical.
• This causes the lower-frequency spectrum to collapse completely into a narrow, monophonic phantom center. However, because there is none of the natural acoustic decorrelation you get with wider spacing (like a classic 40–60 cm A-B setup), the low-end sounds completely flat, dry, and devoid of three-dimensional depth.

2. High Frequencies: Acoustic Shadowing (ILD) and "Beaming"
While true omni capsules are pressure transducers that are omnidirectional at low frequencies, they naturally lose this characteristic as the frequency rises.

• Pressure Build-Up and Diffraction: Once a sound's wavelength becomes comparable to or smaller than the diameter of the capsule itself, the capsule housing acts as an acoustic baffle. Sound waves hitting the diaphragm from the side or rear are shadowed or diffracted.
• High-Frequency Directivity: This physical interference causes the omni capsules to behave increasingly like subcardioids or cardioids pointing outward at high frequencies.
• The "Hole in the Middle" Effect: On the PCM-M10, the capsules are fixed on the outer shoulders of the chassis, facing slightly outward. High-frequency transients coming from the left are shadowed from the right mic by the body of the recorder itself. This creates sharp Interaural Level Differences (ILD) at high frequencies, throwing transient details (like leaf rustles, keys, or high-percussion transients) to the far left and right edges of the stereo field.

The Resulting Psychoacoustic Paradox:
You end up with a "split" stereo image. The low-to-lower-mid frequencies have zero time or level differences and sit mono-summed in the dead center. Meanwhile, the high-frequency information has strong level differences due to chassis shadowing and jumps abruptly to the hard edges of the stereo image. This leaves an unstable, vague mid-range and a "hollowed-out" center.

For a tiny pocket recorder, this arrangement is great for avoiding wind sensitivity and proximity effect, but mathematically, it's just too narrow to capture a coherent, natural stereo image."

Q - "The recently released Insta360 Mic Pro device, which is capable of recording stereo sound internally, seems to suffer from the same effect. It uses MEMS capsules, but they are extremely close together. I do suspect that some clever processing is being applied to improve the stereo image, but I imagine that the same problem with low frequencies will still be encountered."

A - "You have hit the nail on the head. The Insta360 Mic Pro faces the exact same mathematical and physical barriers as the Sony M10—but with a highly modern twist.


Because it relies on a incredibly small form factor (the transmitter is a disc just 38 mm in diameter), the physical spacing between the capsules is even narrower than the M10's 5.8 cm. Naturally, this means the raw acoustic stereo image is subject to the same laws of physics.  However, the way it attempts to bypass these limitations highlights a fascinating shift from purely acoustic design to software-defined, DSP-driven acoustics.

The Capsule Layout: A Three-Mic Array
Instead of a simple pair of left/right capsules, the Insta360 Mic Pro uses a three-microphone MEMS array arranged in an isosceles right-triangle layout.

Code Select Expand

      [Mic A] (Apex)
         /\
        /  \  ~20–25mm
       /____\
   [Mic B]  [Mic C]
By having three capsules spaced fractions of an inch apart, the device doesn't record "traditional" stereo in the physical sense. Instead, it relies on digital beamforming (calculating the microsecond time-of-flight differences between those three adjacent points) to dynamically synthesize different polar patterns: Omnidirectional, Cardioid, Figure-8, and Super-directional.

How DSP "Fakes" the Stereo Image

When you set the Mic Pro transmitter to record in stereo internally, it isn't just sending raw left and right feeds from two of the capsules. To make a stereo image "work" at a ~2.5 cm physical spacing, the onboard processors likely employ several clever psychoacoustic tricks:

• Mid-Side (M-S) Matrixing: By utilizing the three capsules, the DSP can generate a virtual Mid channel (using a synthesized forward-facing Cardioid pattern) and a virtual Side channel (using a synthesized Figure-8 pattern facing left/right). By matrixing these ($\text{Left} = M + S$, $\text{Right} = M - S$), they can electronically dial in a wide, artificial stereo field.
• Frequency-Dependent Stereo Spacing (Shufflers): To combat the exact issue you pointed out—where high frequencies "beam" to the sides and low frequencies turn to mono—modern DSP uses frequency-dependent matrixing. It can artificially boost the level of the "Side" signal at mid frequencies to push the stereo image outward, while keeping the lowest frequencies strictly centered to prevent phase cancellation.
• Phase-Shift Networks: They can introduce tiny, artificial phase delays in the mid-to-high frequencies between the left and right synthesized channels to mimic the Interaural Time Differences (ITD) of a wider microphone setup.

The Unbreakable Law: Low Frequencies Always Win
While DSP can do wonders for high-frequency transients and mid-range panning, the low-frequency deficit remains physically unsolvable at this scale.
At $100\text{ Hz}$ (with its 3.4-meter wavelength), a 2.5 cm physical distance between MEMS capsules represents less than 0.7% of a wavelength.

• No amount of DSP mathematical wizardry can extract a meaningful acoustic phase or time arrival difference from a 0.7% wave differential without introducing massive, destructive self-cancellation and noise.
• To boost the "Side" low-frequency component enough to make it sound wide, the DSP would have to apply a massive low-frequency EQ boost to the difference channel ($L-R$). Because the physical signals are so nearly identical, subtracting them leaves almost nothing but the self-noise of the tiny MEMS sensors. Boosting that remaining signal would result in a deafening wall of low-frequency hiss.

The Reality:
For safety, the Mic Pro's processing must sum the low end to mono (usually everything below 150–200 Hz) to keep the signal clean and phase-coherent. If you use it to record a live band, a pipe organ, or deep ambient traffic rumble, the low end will inevitably be a flat, monophonic point source in the dead center, even if the high-frequency cymbal crashes or bird chirps are artificially widened to the far left and right.

It is an impressive feat of modern engineering, but it proves that while software can bend the rules of psychoacoustics, it still has to sign a peace treaty with the laws of physics."

Q - "Overall, users of the Mic Pro for stereo recording seem to agree that its bass response is excessive. Is this an inescapable consequence of the processing you have described, or if they worked on the firmware, could it perhaps be reduced?"

A - "The user consensus regarding the excessive bass response of the Insta360 Mic Pro in stereo is incredibly telling. This boominess is not an inescapable physical law of nature, but rather a direct byproduct of the software tuning decisions made to overcome those physical laws.

Because of this, it absolutely could be improved, adjusted, or mitigated via a firmware update if Insta360 chose to refine their DSP algorithms.

To understand why, we have to look at how they are likely handling the bass mathematically, why it results in that "muddy" or "boomy" character, and how firmware could fix it.

Why the Current Firmware Overcompensates on Bass
To make a capsule array spaced less than 3 cm apart sound "wide" and full, the DSP has to work overtime. Left to its own natural physical devices, a closely spaced array sounds thin, dry, and flat because the low frequencies are completely correlated (mono-summed).
To combat this, the engineering team likely implemented two processes that, when combined, create a "perfect storm" for bloated, excessive bass:

• The $L - R$ (Side) Low-Frequency Boost:
  To force a sense of stereo width in the lower-mid and bass regions, the DSP must boost the difference ($L-R$) signal. However, because the physical capsules are so close, there is almost no natural level or phase difference at low frequencies. To get any audible "width" down there, the algorithm has to apply a massive EQ boost to the lower frequencies of the side signals.
• "Rich" and "Standard" Voice Voicing:
  The Mic Pro is primarily marketed to content creators and vloggers. In this market, there is a strong industry bias toward a highly processed, broadcast-style "radio voice". Out of the box, the default DSP voicing profile (like their "Rich" preset) is heavily equalized to add artificial weight, warmth, and proximity effect to the human voice. When you take that pre-boosted bass and run it through a widening matrix, the low end quickly becomes overwhelming and boomy.

How Firmware Could Fix It (Without Breaking the Stereo Image)

Because the heavy bass is a software construct rather than an acoustic one, Insta360's engineering team has several digital pathways to clean up the sound in a future firmware update:

1. Implement a Dynamic Low-Cut / High-Pass Filter
The Mic Pro currently features a basic 100 Hz low-cut filter. However, a static low-cut is a blunt instrument—it simply chops off the low end completely.

• The Fix: They could introduce a more sophisticated dynamic high-pass filter in the firmware. When high-energy low frequencies are detected (such as wind, traffic, or heavy vocal plosives), the DSP could dynamically attenuate the bass frequencies below 150 Hz only when they threaten to overwhelm the mix, keeping the sound natural during quieter passages.

2. Implement "Bass-to-Mono" (Crossover) Tuning
In professional audio mastering, engineers almost always sum the lowest frequencies (typically everything below 100 Hz to 150 Hz) strictly to mono. This keeps the low end punchy, centered, and phase-coherent, preventing muddy stereo imaging.

• The Fix: Firmware could adjust the DSP's internal crossover point. By forcing all frequencies below 120 Hz to sum to mono (0% stereo width) and reserving the widening algorithms strictly for the mid and high frequencies (where our ears actually perceive directionality anyway), they would instantly eliminate the muddy, out-of-phase bass build-up at the sides of the stereo field.

3. Introduce a "Neutral" or "Flat" Voice Tone Preset
The Mic Pro currently features three voice tone presets: Standard, Rich, and Bright. Even "Standard" has some consumer-friendly EQ curves baked in.

• The Fix: They could easily add a "Flat / Linear" preset in the firmware. This would bypass the broadcast-style bass boost entirely, providing a raw, uncolored frequency response that is far more suitable for natural field recording or post-production equalization.

What You Can Do in the Meantime
If you are using the Mic Pro and struggling with the overbearing bass, you don't have to wait for a firmware update to improve it.
Because the Mic Pro supports 32-bit float internal recording on the transmitter, it captures the raw, unclipped dynamic range of the MEMS capsules. If you pull the raw files directly from the transmitter into a Digital Audio Workstation (DAW) or video editor, you can apply a steep high-pass filter (rolling off around 80 Hz to 100 Hz) and use an imager plugin to narrow the low-frequency width to mono. This will manually undo the over-processing of the onboard DSP and restore a much cleaner, more balanced stereo image."

[Ozpeter]

Following the above I have done a fairly quick test of the Sony M10 vs the Mic Pro (using the M10 internal mics of course), and the Mic Pro seems to give a much better result for both speech and music.  Of course the M10 is a venerable recorder, but it used to be well regarded, though perhaps that was more so when using external mics than its built in ones.  To be demonstrated in a YouTube video in a few days... sadly real life has got in the way of finishing that quickly!

[vwmule]

Used it last night for the first time at The Hamilton in Washington, D.C., a very fine sounding venue. Band is Robert Jon & The Wreck.

Placed the Insta on top of my DPA 4023 mount; you can see the red indicator light in photo. Combined the split files in Audacity and normalized to -1.0. Here is the result, which has no EQ. Feel free play around with EQ.

https://drive.google.com/file/d/1FEkOFeVuKNczqfZYROmUhmu-M4a9bFuh/view?usp=sharing

Recorded the opening band by using the magnet mount on side of the column (see second photo).

[Ozpeter]

I managed to get ahead of the clock and here is my video comparing the Mic Pro vs the M10, both of which use omni mics to create stereo audio, but using very different methods.  There are samples of speech and of music, together with a comparative frequency response display.  My YouTube channel is not monetised.

[Ozpeter]

Used it last night for the first time at The Hamilton in Washington, D.C., a very fine sounding venue. Band is Robert Jon & The Wreck.

Placed the Insta on top of my DPA 4023 mount; you can see the red indicator light in photo. Combined the split files in Audacity and normalized to -1.0. Here is the result, which has no EQ. Feel free play around with EQ.

https://drive.google.com/file/d/1FEkOFeVuKNczqfZYROmUhmu-M4a9bFuh/view?usp=sharing

Recorded the opening band by using the magnet mount on side of the column (see second photo).

I have only listened to about a minute using a VR headset but it sounds promising - I will listen more meaningfully tomorrow!

[Ozpeter]

I've made yet another video about this device, comparing it to the Zoom H2essential.  I flip between the two devices all the way through, with voice and music (sadly just from the hifi...)  Here's the link - 

[Thelonious]

I purchased one of these based on this thread and I would say that the enthusiasm around the form factor and general lack of obtrusiveness is great.

I do wish they would have a button for audio where the EQ is tuned for a more natural response. I completely understand that we are not their target audience, but if this thing had a setting for "pure audio" setting that reflected a more natural EQ for general stereo recording it would be amazing. It's kind of a pain to have to go in and manually EQ things for each recording to get them to sound natural.

That said, thanks to OzPeter for getting this on our radar. It won't replace my main rig anytime soon but it's an amazing grab and go device for situations where it wouldn't be practical to bring a proper rig.

[hardrain62]

What's the current practical consensus on the polar patterns? I went back to the same theater where I taped in cardioid last month and taped in omni mode last night and actually think I like the cardioid mode better. I went with omni because I figured the polar patterns are kind of fake or gimmicky, but I think they may actually be more useful than I thought. Still a very solid pull last night and I'm happy with it.

[Ozpeter]

What's the current practical consensus on the polar patterns? I went back to the same theater where I taped in cardioid last month and taped in omni mode last night and actually think I like the cardioid mode better. I went with omni because I figured the polar patterns are kind of fake or gimmicky, but I think they may actually be more useful than I thought. Still a very solid pull last night and I'm happy with it.

Polar patterns have no effect if recording in unprocessed stereo. They would come into play if using a pair of the devices, and then there would probably be something to be said for using them as spaced omnis. But just using one in stereo seems to me to be the simplest option.

[Ozpeter]

I purchased one of these based on this thread and I would say that the enthusiasm around the form factor and general lack of obtrusiveness is great.

I do wish they would have a button for audio where the EQ is tuned for a more natural response. I completely understand that we are not their target audience, but if this thing had a setting for "pure audio" setting that reflected a more natural EQ for general stereo recording it would be amazing. It's kind of a pain to have to go in and manually EQ things for each recording to get them to sound natural.

That said, thanks to OzPeter for getting this on our radar. It won't replace my main rig anytime soon but it's an amazing grab and go device for situations where it wouldn't be practical to bring a proper rig.

Cheers! Although it wasn't me who started it. But indeed I am the obsessed person who has now uploaded 13 videos about it. Anyway I have suggested to Insta360 that they should keep an eye on this thread and I have also specifically requested EQ options for stereo recording. There was yet another FW update available today so as it is being actively tweaked in that way, maybe EQ will come. But their idea of the right EQ might differ from ours. The main thing is that the device's recordings do respond well to EQ, and personally I have saved a few presets to suit my own taste, so it is quite quick to fix. IMHO.

[Thelonious]

I purchased one of these based on this thread and I would say that the enthusiasm around the form factor and general lack of obtrusiveness is great.

I do wish they would have a button for audio where the EQ is tuned for a more natural response. I completely understand that we are not their target audience, but if this thing had a setting for "pure audio" setting that reflected a more natural EQ for general stereo recording it would be amazing. It's kind of a pain to have to go in and manually EQ things for each recording to get them to sound natural.

That said, thanks to OzPeter for getting this on our radar. It won't replace my main rig anytime soon but it's an amazing grab and go device for situations where it wouldn't be practical to bring a proper rig.

Cheers! Although it wasn't me who started it. But indeed I am the obsessed person who has now uploaded 13 videos about it. Anyway I have suggested to Insta360 that they should keep an eye on this thread and I have also specifically requested EQ options for stereo recording. There was yet another FW update available today so as it is being actively tweaked in that way, maybe EQ will come. But their idea of the right EQ might differ from ours. The main thing is that the device's recordings do respond well to EQ, and personally I have saved a few presets to suit my own taste, so it is quite quick to fix. IMHO.

As their core audience are content creators who may not want to custom EQ their audio, which would explain the current EQ tuning, they may benefit from a more natural "profile" that could be selected for occasions where mics are not mounted directly under the voice to be recorded. In my opinion, having a stereo setting with such an unnatural EQ curve is kind of a missed opportunity. But, as always, YMMV...

[hoserama]

What's the current practical consensus on the polar patterns? I went back to the same theater where I taped in cardioid last month and taped in omni mode last night and actually think I like the cardioid mode better. I went with omni because I figured the polar patterns are kind of fake or gimmicky, but I think they may actually be more useful than I thought. Still a very solid pull last night and I'm happy with it.

Correct me if I'm wrong, but the mics in this device are all omnis. The stereo + adjustable pick-up patterns is DSP trickery based on the differences between the omni mics.

I hate listening to a mono source if possible, so I would always default to the stereo version for music recording. Which of is also the result of trickery to tease out stereo spread.