Developments in Microphone Technology

What’s interesting is that this matchstick-sized microphone can be attached to drones.

Conventional microphones work when sound waves make a diaphragm move, creating an electrical signal. Microflown’s sensor has no moving parts. It consists of two parallel platinum strips, each just 200 nanometres deep, that are heated to 200° C. Air molecules flowing across the strips cause temperature differences between the pair. Microflown’s software counts the air molecules that pass through the gap between the strips to gauge sound intensity: the more air molecules in a sound wave, the louder the sound. At the same time, it analyses the temperature change in the strips to work out the movement of the air and calculate the coordinates of whatever generated the sound.

EDITED TO ADD (10/6): This seems not to be a microphone, but an acoustic sensor. It can locate sound, but cannot differentiate speech.

Posted on October 4, 2013 at 6:59 AM68 Comments


Tom October 4, 2013 7:20 AM

You (meaning any organization with the ressources and the will to fight this surveillance) need to get a few of these, put them in parks and busy streets, record people’s conversations or phone calls and stream it to the very people who’s conversations you recorded. Do this 24/7 on places like Times Square for a few months and you got the media attention we all need.

Snarki, child of Loki October 4, 2013 7:27 AM

“Microflown’s software counts the air molecules that pass through the gap between the strips to gauge sound intensity”

It’s hard for me to see how this design (if described accurately) will not be shot-noise limited.

Tom October 4, 2013 7:30 AM

You (meaning any organization with the ressources and the will to fight this surveillance) need to get a few of these, put them in parks and busy streets, record people’s conversations or phone calls and stream it to the very people who’s conversations you recorded. Do this 24/7 on places like Times Square for a few months and you got the media attention we all need.

Glenn Dixon October 4, 2013 7:38 AM

“as far away as 20 metres”

I guess this could work on really, really small drones, possibly used in hostage rescue or other law enforcement work. Not practical yet for Predator/Reaper usage.

Larry Smada October 4, 2013 8:06 AM

Glenn Dixon: “Not practical yet for Predator/Reaper usage.” – How do you know this?

I imagine that the NSA, along with several other of our spy agencies, have funneled quite a bit of money into this technology.

The 20 m – 25 m range is probably only what the company is allowed to disseminate.

121jigawatts October 4, 2013 8:33 AM

Should be easy enough to spot using a camera sensor tuned to color temperature of 473.15 kelvin, possibly on a home made drone.

Then trigonometric location based on,
1. more that one geographic perspectives with the same camera, or

  1. more than one geolocated cameras observing the same energy source resolved using spherical trigonometry.

Possible countermeasures.

Send a supersonic projectile to dispense ceramic insulation all over the sensor. Supersonic ensures the microphone will not catch the threat in advance of being doused. Foam insulation would not obviously work or it would melt.

Aim a directional random baseband white noise source at the sensor to increase its noise floor many orders of magnitude, there by degrading signal to noise ratio by the same magnitude.

A semi powerful LASER modulated with white noise could also apply direct thermal noise as heat energy to the surface of the electrodes, thereby jamming their thermal sensor.

While aiming dual directional beat frequency ultrasonic beams (one straight carrier and one sideband modulated with random white noise) can mix nicely together to jam diaphram mics using surprisingly small transducers, this would not work on a temp based mic, since there is no acoustic surface for these to mix on, and the device just needs to discriminate the ultrasonic air movement band out.

Develop your telepathic communication capability, which should at least buy you a few more years of privacy.

Neuromancer October 4, 2013 10:03 AM

having to heat it to 200 deg c seems to be a bit of a drawback for any long term covert deployment.

Wael October 4, 2013 10:21 AM

I have a hard time believing a single sensor without external support can detect the direction. A few sensors can detect the direction based on phase and amplitude. A few sensors can also detect direction by triangularization using delays. A single sensor, in the audio spectrum frequency band will need some form of a reflector to be “unidirectional”. I am not sure what the radiation pattern is of this sensor, but I think it’ll be close to omnidirectional, or perhaps a bowtie pattern. I think I’ll take this claim with a big bite out of Lot’s wife right arm!

121jigawatts October 4, 2013 10:59 AM

“Lots wifes right arm”
That was good…

I would agree that accurate directional discrimination is not possible without an array of sensors. The individual sensor is one level.

But on the streets, there is a higher level that analyzes the time of arrival of matching soundbytes between differently located microphones, taking into account and cancelling known pre calibrated acoustical echo signatures in the vicinity (which they do for existing gunshot systems). This is probably how people would be located.

At a range of 20 meters, this would suggest likely grid spacing of 20 to 30 meters in the real world for 1984 style coverage. If oriented vertically (vertically polarized), it might be closer to omnidirectional, even if in the horizontal plane is a bowtie pattern.

For directional use on a drone, several could be deployed so that sound arriving from the direction aimed adds in phase, but that coming from the sides cancel (like a shotgun mic).

These would be perfect ears for the terminator style robots being produced by Boston Dynamics.

Darron October 4, 2013 11:18 AM

@Wael It seems like it’s a “single sensor” that measures effects at different locations on the “strips”. So, in effect, one device is several sensors.

If you believe the article, the original purpose is detecting the direction of sounds and capturing recognizable speech is a recently discovered bonus.

Wael October 4, 2013 11:27 AM

@ Darron,

So, in effect, one device is several sensors.

I’d believe that if the device were large in comparison to the wavelength of sound[1]. Even if you have multiple sensors, they’d have to be appropriately separated in distance.

[1] I had a long discussion with a colleague once about the wavelength of sound compared to the wavelength of a GSM/CDMA RF signal. Discussion started with a claim that a shielded room for sound can work as a shielded room for RF. I forgot the details, but I’ll share them when I remember, in the right context.

MikeA October 4, 2013 11:28 AM


“If oriented vertically (vertically polarized), it might be closer to omnidirectional, even if in the horizontal plane is a bowtie pattern.”

So, a donut then? Sounds like a good match for LEOs.

On a more serious note, this reminds me of a homebrew wind direction/speed sensor I saw in the 1980s or so. Obvious much lower bandwidth, but what other “old” tech can be re-thought with newer fab techniques?

Wael October 4, 2013 11:33 AM

@ 121jigawatts,

Re: Lott’s wife…
There an evolution history behind that. Will share it later. The latest iteration was borrowed from @Clive Robinson, who consumed all of Lott’s wife, TWICE!

By the way, it’s spelled Gigawatts, and pronounced as you have spelled It:)

121jigawatts October 4, 2013 11:34 AM

Then it is probably not a linear distribution of sensor elements which is the first thing that came to my mind.

A 360 acoustic field of view should be either polar or cartesian array of elements distributed horizontally. At a thickness of only 200nanometers (which is one wavelength at far UltraViolet), they apparently have alot of latitude as to what form factor can be used and still keep these elements within the size of a match head.

Key word…should.

Alex October 4, 2013 11:41 AM

Could someone correct me if I’m wrong on this, but isn’t this similar to the anemometer (wind gauge) used on NOAA’s ASOS stations? It’s been ages since I looked them up, but this sounds very similar in design.

paul October 4, 2013 12:02 PM

If you were going for shock value, you’d distribute these in nominally-public areas near Wall Street or the Capitol.

Wael October 4, 2013 12:23 PM

@ 121jigawatts,

they apparently have alot of latitude as to what form factor can be used and still keep these elements within the size of a match head.

Wavelength of sound is between 1.7cm and 17 m

A thickness of only 200 nanometers is way smaller than a wavelength. Will not be able to detect direction, I think.

121jigawatts October 4, 2013 12:38 PM

@Wael Next thing your going to tell me is that a eeottabyte is actually spelled with a “y” right?

If this measures the unrestricted free flow of air, there is no bandwidth restriction imposed by physical limitations of a diaphram or ribbon, so is theoretically DC to supersonic threshold, and perhaps even beyond. Perfect for accurate wind speed and direction in one sensor.

If elements are spherically distributed around free space, it could even be used for wind based speed navigation (until enough crud accumulates on everything).

But there would be a world of physical design difference between wind speed and audio ranges, specifically as it pertains to the rate at which the platinum elements have to give up their heat with sufficient speed to detect and reproduce audio range waveforms from temperature differentials, and the rate restrictions at which that heat can get resupplied by the heat source so that it does not reset the differential faster than the audio information can be recovered. This probably has to do with the 200nm thickness they wound up using, taking into account the thermal conductivity of platinum, and the thermal conductivity of the substrate used to moderate or damp the heat source on the resupply side.

roadie October 4, 2013 1:02 PM

I am just amazed at all these smart people crawling out of the woodwork to build the framework for a secure future – MegaCity One.

maybe it was TV that killed the imagination in people, who knows.

Aspie October 4, 2013 1:05 PM

If I’m reading this right, this allows you to know that “a noise occurred at a particular compass point.”

Did I miss something? If it’s omnidirectional, how can you pinpoint a specific source and conversely if it’s unidirectional how can you separate sources? Moreover how can you then record anything useful beyond a cat fart that occurred “somewhere”?

(sound of scratching noggin)

121jigawatts October 4, 2013 1:09 PM

Awesome @Milo M.
I’m all set for a couple weeks.

Ezekiel 28 described a latter day regime as being wiser than Daniel, to the extent that there was no secret it did not know. This same being lifted up in its capability into projecting itself as a substitute for God. Many profiling tumblers click in as positives there too.

Aspie October 4, 2013 1:12 PM


What I mean is, I have ears and porridge that tell me where a sound occurred. What I want to know is what the sound was.

If that makes sense.

Milo M. October 4, 2013 1:13 PM

@Alex: “isn’t this similar to the anemometer”
@Mike A: “this reminds me of a homebrew wind direction/speed sensor I saw in the 1980s or so”

“Engineers have used heated, single-wire anemometers to study airflow for at least 50 years, says Rich Lueptow, a Northwestern University mechanical engineering professor. Microflown seems to be the first to have seen the promise of a miniaturized, double-wire probe, he says.”

SA story links to this:

Wael October 4, 2013 1:23 PM

@Milo M

In lieu of speculating, one might read some of the Microflown papers

Excellent idea. Thanks for the link and the suggestion.

121jigawatts October 4, 2013 1:25 PM

@roadie No framework bitches here. Think understanding as a means to counteract the FW instead.

@Aspie, omnidirectional field of view, with ability to resolve direction of a given soundbyte using post processing.

Alan Bostick October 4, 2013 1:32 PM

Snarki, child of Loki: It’s hard for me to see how this design (if described accurately) will not be shot-noise limited.

It surely is, just as the CCD array in your iPhone camera is shot-noise limited. It doesn’t stop either from being useful.

Aspie October 4, 2013 1:32 PM


Does that mean a (near enough) spherical locator?
You say “post processing”. Does that mean hoovering 360 by 360 degree sound and somehow not only pinpointing signals but dragging out clear enough content to use?
Oh and being able to give a 3D vector from the mic at the time they were captured?

(asking for a [severely] scratched noggin)

MarkH October 4, 2013 1:50 PM

On Using the Described Sensor to Localize a Sound Source

  1. By measuring the difference in sound-wave arrival times between the two strips, a single sensor can be used to measure the source angle about the axis between the two strips. Two such devices, mounted with their axes mutually perpendicular, can therefore be used to measure the spherical angular coordinates within some limited angular range, probably on the order of 2 to 5 steradians.

  2. Given that the sensor is described as matchstick-sized, I estimate that it can probably do a reasonable job of angle measurement for frequencies of approximately 10 kHz or more (accuracy increasing with frequency). [Note well that 200 nm refers to each of the platinum conductors — the relevant dimension for angle measurement is the distance between the conductors.]

  3. For the design application of the sensor — locating the source of explosive sounds (gunshots, detonations and the like) — a significant high-frequency component is presumably typical. If the explosion is near enough to reach the sensor as a supersonic shock wave, that would likely provide the best case for measurement accuracy.

121jigawatts October 4, 2013 1:58 PM

If the omnidirectional device has awareness of its lat/lon such as

stored coordinates of fixed installations, or
varying GPS coordinates of mobile platforms

until I read through Milos doc references, I do not see how a single sensor will be able to sense both direction and distance of a sound source. For a given direction, regardless of the distance, the timing and angles of arrival will be the same for all practical purposes between all elements of a given single sensor.

Add some spacing between two sensors, any you can resolve a source to a 50 percent probability it is in one direction or the other of the resolved angle but without determining distance.

With three spaced sensors, spherical trigonometry referenced to the device coordinates could be used resolve the targets coordinates (meaning az and distance in lat/long format) whether from fixed or mobile platform, using time of arrival analysis.

Is it possible for all these elements of this latter case to exist on an area the size of a match head with such fine powers of resolution so that so only a single sensor will do it? I will have to see what the docs say cuz I will surely be scratching my head if it can.

Mindbuilder October 4, 2013 2:31 PM

A partial basic explanation of this I’ve read is that as the air flows over one wire and then downwind to the second wire, the warmed wind cools the second wire less than the first. The difference in cooling between the wires varies depending on the direction of the air movement. The sensor has not just one pair, but three pairs of wires oriented orthogonally, and in addition must have a traditional pressure sensor microphone in close proximity to the wires.

Bryan October 4, 2013 3:21 PM

Looking at the slideshows on the website, while the sensor itself is small, its mount is a bit bigger than a matchstick — look at the pictures of the small 1 meter UAV, and it is a probe that sticks out in front. It looks like the sensor itself has a diameter of approx .7 cm and a length of 1 cm.

The deployed units pictured on jeeps and on the ground and elsewhere look like they’re squashed hemispheres slightly larger than 20 cm in diameter (wider than a man’s head nose-to-back) and 12 cm high. I imagine this is for a power source, DSPs, etc. The inter-unit spacing for border security is about the same as Bluetooth Class 1 range, or more probably ZigBee.

The sensor in these hemisphere units is inside and somewhat protected, and difficult to get at unless you’re facing it. However, maybe a squirt gun loaded with salt water would neutralize it.

Garfinkel October 4, 2013 3:54 PM

@Harry Caul
Some day we may be legally required have a mic in every room.

[Channeling Eric Schmidt:] Only reason you would not want that is if you have something to hide.

(Which co-incidentally brings to mind that there are many valid non-criminal reasons to want to “hide” some thing. But perhaps for Google such a sordid act only makes that “thing” the more interesting, who knows.)

Anyway…could it be that all these disclosures about surveillance capabilities are making some folks go cuckoo?

I am saying this because some sources claim that Miriam Carey (the woman that was now shot by the cops in DC) believed Obama was stalking her. And supposedly Aaron Alexis (the DC shooter a couple of weeks ago) had heard “voices”.

(Or is it that we are fed stories like this to make it look like only mentally ill folks believe these things?)

iPhone October 4, 2013 4:52 PM

@121jigawatts: “For a given direction, regardless of the distance, the timing […] of arrival will be the same for all practical purposes between all elements of a given single sensor”

No, it won’t be the same, you will notice a time delay: sound speed is 300 m/s, the delay is then of sin(angle).7e-2/300 = sin(angle)2.3e-5 s, assuming a distance of 0.7cm.

2.3e-5 makes 20000 cpu cycles for a CPU of one GHz. Sounds very practical.

@Harry Caul: “Some day we may be legally required have a mic in every room.”

The NSA will just ask smartphone importators on US soil to upgrade their mic; the public and/or the ads will notice better conversations quality, because of directionnal discrimination done by this device.

Maybe a NSL has been done about that years ago. Time to open your phone and see if the mic is that hot.

121jigawatts October 4, 2013 5:52 PM


I’m trying to wrap my head around what you are saying, I am wondering if I mis-communicated.

Let me rephrase my assumptions:

the standard rate of sound through air is constant.

the angle of arrival does not vary.

the delay would be common to either distance.

how will you determine the distance of someone talking 6.6 meters from the sensor versus the same source standing 20 meters distant along the same axis using just one sensor?

I am not saying it cannot be done, just that I am not reconciling it.

If you feel you have nothing to hide, that is fine. A certain whistleblower now living in Russia, who will remain unnamed to protect his identity proved someone has LOTS to hide from Americans and the world. Maybe they should be the ones to take your advice.

121jigawatts October 4, 2013 6:05 PM

Sorry @Garfinkel, you were quoting Eric Schmidt so disregard that first line. the last line should read
Maybe they should be the ones to take Erics advice.

As for your you questions in bold, as the unnamed Russian refugee has revealed… better err on the side of security than to rely on an unenlightened benefit of the doubt.

Feelin' Lucky October 4, 2013 7:54 PM

@Garfinkel RE: “voices”

Posted on Bitmessage:

“I have a theory about the tinfoil hat phenomenon.

Everyone is familiar with the NSA and their supercomputers. They are “the fast spies”.

Some of the data that they compile is used to conduct operations against individuals.

Some of those operations are psychological shaping operations that can take years to accomplish.

The people who conduct these types of psyops are “the slow spies”.

They are invisible because they operate so slowly that their movements are not noticed.

They can do seemingly normal things that taken in isolation do not register as manipulation to the victim.

Mental destabilization can be cultivated by altering the impressions that a person is exposed to over time.

Voices in the head is how that mental destabilization is experienced by the victim.

To explain the presence of the voices, the idea follows that perhaps they are implanted by microwaves.

Making a tinfoil hat is a response to the expectation of microwaves.”

Figureitout October 4, 2013 8:34 PM

The other night my dad was talking about how a tiny solar panel and a little IC can recharge a tiny battery many times and lose either little or no capacity after all those recharges; and it’s used for extremely low power transmission bursts. Forget the company, but they were previously reserved for the military and I think developed in Oak Ridge, TN and they could power all sorts of little sensors.

Coyne Tibbets October 4, 2013 8:41 PM

Another link: Source Localization Using Acoustic Vector Sensors (Washington University). Per that article, in 2009 they were using it in 4-microphone linear arrays that can detect the direction to a sound source.

I predict it will be used in match-head sized phased microphone arrays that can “focus attention” to record a specific conversation taking place in any desired direction (probably many conversations in many directions at once). Hardware or software to filter such directionality out of phase space should almost be off the shelf these days.

I also suspect it will be much more sensitive in such a form than is indicated in the main article: I think it likely that such a device, mounted on a street corner telephone pole, should have no problems listening to any conversation (indoors or out) within 100 meters ( more than 1/2 block). It would only require an improvement of 2-3 orders of magnitude in sensitivity; that may already have been accomplished.

MarkH October 4, 2013 11:21 PM

Yes, see what Tibbets wrote. “Calculate the coordinates” does not (and in practice, can not) mean coordinates in space: it is a question of the direction from which the sound comes.

Remember, this gadget was developed for a purpose: when somebody is shooting at you, you want to know what direction they are shooting from, ASAP. The information doesn’t have to be very precise, in order to be very useful.

I received a brochure in the 80’s from a company offering acoustic gunshot direction finding systems for use by ground troops.

Garfinkel October 4, 2013 11:48 PM

No probs about the last line in your earlier posting, and I agree with you about our Russian refugee. I would probably stay where he is if I were him.

One thing that I disagreed with was Bruce’s point that balanced power is more ‘stable.’ One semester of International Relations and a general knowledge of WWI leads me to believe that balanced power is actually unstable. What’s stable is overwhelming power.

In my opinion you are correct, and I am sure the strategies followed by the US military and its various departments (such as NSA) are fundamentally based on this notion.

I would also suppose that this will be more evident in the future. I mean if the current developments continue, the US military will eventually likely have some kind of laser equipped swarm-capable high-speed AI drones capable of cloaking themselves in visible light.

Or something like that, although currently I am just calling it a hypothesis.

The problem with this sort of development is though that hundreds of years ago, when peasants got tired of the ruling kings, they eventually got together in sufficient numbers and overthrew the tyranny. I do not think it will be possible in the (not so distant) future because the difference between the weapons of the population and the weapons of the government.

That is, if this hypothesis becomes reality.

In which case I think we will at that point need to send Bruce Schneier to the past to rescue the future…

iPhone October 5, 2013 3:52 AM

@121jigawatts: “I’m trying to wrap my head around what you are saying, I am wondering if I mis-communicated.”

@Wael: “You lost me somewhere in the math…”

Sorry, I misquoted 121jigawatts. I should have quoted: “Add some spacing between two sensors, any you can resolve a source to a 50 percent probability it is in one direction or the other of the resolved angle but without determining distance.”

If you have two such devices, the only possibility for two sources to be undistinguishable would be two sources on the line passing through both microphones: most unlikely.

If you have only one such device, I agree that there is a small probability that two sources are in the exact same direction. But if these sources and/or the device are moving, it is most unlikely that they will stay aligned, and some advanced maths will be able to separate both signals.

People from Occupy should start wearing necklaces that speak out random words loudly. They will then be able to communicate by whispering.

US people should start by fighting back: by sending such microphones in the TLA headquarters, by influencing congress and/or mounting them on cockroaches. If US people do no more have privacy, why should the TLA have privacy ?

Snowden October 5, 2013 4:13 AM

@Jarkko: “Monarch Programming is a method of mind control used by numerous organizations for covert purposes.”

Sorry, I will wait for revelation from Snowden about Monarch program before believing it is reality.

Some defense used by lawyers is not sufficient for me.

I know that Neuro-Linguistic Programming is real and is used.

Xelandre October 5, 2013 2:07 PM

Intriguing, but how well can it work?

It’s also nothing really new. This thing has been apparently bouncing around since the early 1990s, with the PhD work of the inventor Hans Elias de Bree.

The basic principle is explained in international application <a href=″>WO9935470, which matured into a number of national patents, including US5959217. These are due to expire in the coming months, if they’re not dead already.

Reference is made in the preamble of the WO document to a paper by one R.O. Fehr in the Journal of the Audio Engineering Society, April 1970, Vol. 18, Nr. 2, pp. 128-132 entitled “Infrasonic Thermistor Microphone”. 40 years ago. The applicant differentiates his invention from that prior art by the dimensions of his sensor, which is defined as being slightly smaller than Fehr’s 330um.

Microflown was founded in the closing years of the century, and has attempted to obtain further patents since, generally unsuccesfully, e.g. WO9935470.

I’m still skeptical as to whether this is amenable to high performance à la “The Conversation”. My RF designer brain says that you must have aperture to obtain gain and directivity. Here you achieve at least the latter objectives by concentrating on individual molecules. I have to mull this over

Xelandre October 5, 2013 2:11 PM

For some reason the first URL in my first comment didn’t go through.

I’ll try again:

Is there a limit of one URL per comment?

Coyne Tibbets October 5, 2013 8:03 PM

@MarkH: Locate four of these microphones in a 1 foot tetrahedron and direction becomes coordinate in real space. The coordinate easily gives you the range for the return shot.

MarkH October 6, 2013 1:01 AM


I didn’t follow the logic of that post. To me, “coordinate in space” means that the distance must be known, not only the direction.

Setting SONAR to one side, measuring distance with microphones is impractical unless the distance is very short, n’est ce pas?

Prinz van der Schemering October 6, 2013 1:04 AM

I keep wondering what the lifetime of these microphones will be before weather and pollution do them in. Likewise how inclement weather makes them inoperable. Indeed, what tools could be used to make them inoperable.

Wael October 6, 2013 1:58 AM

@ Milo M.,

Read the papers…
It’s not a microphone per se, as the linked article that Bruce included, claims. It’s an “Acoustic Sensor” that is capable of detecting directions. It also senses sounds from rifles and guns — high velocity projectiles that create a detectable “temperature potential difference”, and not from a “conversation”. Unless I missed something, such sensor is not suitable for collecting a conversation from a crowd, or even detect the direction of normal sounds. It can’t be even classified as a “microphone”.

Wael October 6, 2013 2:17 AM

@ 121jigawatts,

As promissed… [There (is) an evolution history behind that. Will share it later.]
I think I started by “exaggerating” the “pinch of salt” expression

Wael • June 17, 2012 3:08 PM
I have to give the word “assumed” in “assumed to be trusted” a higher precedence, and take the whole trust thing with a pound of salt – a grain of salt is not enough to enhance the taste here, but a pound still makes it difficult to swallow – if you know what I mean 😉

Then @ Clive Robinson “out-did” me by being more creative.

Clive Robinson • August 12, 2012 2:22 AM
So you have to take “the estimates” with a pich of salt around the size of “Lot’s Wife”.

Then again (the “TWICE”)

Clive Robinson • April 5, 2013 5:05 PM
I would take that with a huge pinch of salt atleast as big if not bigger than Lott’s Wife [1].

That is my recollection, maybe the expression or a similar one was used before on this blog…

Reg Gigawatts, I was being “cute” 😉

Coyne Tibbets October 6, 2013 10:07 AM

@MarkH: These microphones are intrinsically able to determine direction-to-source of a sound. Take two of these microphones separated by, say, a foot, and a sound before them originating from a point at some distance. The angle between the sound and each microphone varies by the location of the source, allowing triangulation, which gives not only a verification of the actual direction, but also distance.

For two microphones you have two problems: The triangulation only works in 2-D, and there will be problems triangulating sources that lie near a line passing through both microphones.

To cure the second problem, arrange 3 microphones on a plane, parallel to the surface. Two of the microphones will always be able to triangulate well, regardless of direction. So now we have 360 degree triangulation in 2-D. Now add a fourth microphone centered above the plane of the other 3 microphones, creating a tetrahedron. Now we can triangulate in 3-D, at any angle vertically or horizontally, along with determining distance to source. Anywhere within a sphere surrounding the microphones.

MarkH October 6, 2013 12:11 PM

Yes, but direction is not position: “coordinates in space” means (x, y, z), or (r, θ, φ). Such an array won’t give the range (r in spherical coordinates).

Coyne Tibbets October 6, 2013 12:58 PM

@MarkH: Read up on triangulation, and how it can be used to compute distance or spatial coordinates.

Then consider that each pair of these microphones can provide the direction needed for one input to triangulation. Distance is a consequence of the triangulation.

With 4 microphones set on the vertices of a tetrahedron, there are 6 paired combinations of microphones. Therefore, for any sound source, 6 triangulations can be performed. Some will fail for some source directions, but for any source we should see at least 3 successful triangulations on differing planes. Taken together, those should produce a very accurate distance or, as desired, full spatial location in XYZ space.

MarkH October 6, 2013 2:24 PM


I won’t go any farther down this road.

I leave as an exercise to the reader, to estimate the maximum range that can be measured to within 20% error by any array of microphones that is most 30 cm in extent.

PS If acoustic triangulation is effective on a short baseline … why do navies go to the enormous complexity and expense of active (pinging) SONAR?

Water October 6, 2013 3:28 PM

@MarkH: “why do navies go to the enormous complexity and expense of active (pinging) SONAR”

Because this new “microphone” does not work in water ?

Because rocks do not emit much sound, but can crush a submarine ?

Coyne Tibbets October 6, 2013 4:53 PM

@Water: Microphones work just fine in water. The reason submarines use an active ping is that underwater rocks do not give off sound by themselves. To have sound from a rock to use to triangulate it, a submarine must provide the sound, and listen for the echo.

Which is also why it is a ping: In theory a continuous sound could be used, but in practice the echos cannot be heard unless the submarine is quiet part of the time (between pings).

But up here in the real world, gunshots and people talking give off their own sound. Triangulation works just fine on anything that gives off its own signal. Which brings us around to submarines again: During war games or actual battles, the active ping must be turned off and the ship be operated as silently as possible…because the enemy (and his torpedoes) can triangulate your active ping (or any other sound you make) quite easily.

RobertT October 6, 2013 8:38 PM

Why so much interest over a microphone?
Cheap small and sensitive microphones are trivial to build check out any modern phone and you’ll find several tiny microphones. No exotic electronics are needed and they can be made to have extremely low power by adding some passive Acoustic signal strength indicators.

You can already buy commercial products that incorporate Microphone, Mic Amplifier, and low power transmitter. Commercially they call these Bluetooth headset BUT they make great low power bugs, tey even have a frequency hopping Tx builtin. Add a small solar cell and you can deploy these in public spaces or just have them self powered for use in conference rooms, hotel rooms wherever.

vas pup October 7, 2013 9:21 AM

SSSS- technology (Google for details) provide the option when sound is applied/transmitted directionally to one person out of many standing around the target making impression that target is hearing voices because nobody else around is hearing nothing.
@121jigawatts provided list of possible countermeasures against lates microphone.
Q: Is it possible to detect in real time source/location of SSSS transmitor and separate those persons affected by such technology and those having mental problem hearing voices?

vas pup October 7, 2013 11:11 AM

Does SSSS is utilized for that? See below:
“The Stasi perfected the technique of psychological harassment of perceived enemies known as Zersetzung (pronounced [ʦɛɐ̯ˈzɛʦʊŋ]) – a term borrowed from chemistry which literally means “decomposition”.

By the 1970s, the Stasi had decided that methods of overt persecution which had been employed up to that time, such as arrest and torture, were too crude and obvious. It was realised that psychological harassment was far less likely to be recognised for what it was, so its victims, and their supporters, were less likely to be provoked into active resistance, given that they would often not be aware of the source of their problems, or even its exact nature. Zersetzung was designed to side-track and “switch off” perceived enemies so that they would lose the will to continue any “inappropriate” activities.

Tactics employed under Zersetzung generally involved the disruption of the victim’s private or family life. This often included psychological attacks such as breaking into homes and messing with the contents – moving furniture, altering the timing of an alarm, removing pictures from walls or replacing one variety of tea with another. Other practices included property damage, sabotage of cars, purposely incorrect medical treatment, smear campaigns including sending falsified compromising photos or documents to the victim’s family, denunciation, provocation, psychological warfare, psychological subversion, wiretapping, bugging, mysterious phone calls or unnecessary deliveries, even including sending a vibrator to a target’s wife. Usually victims had no idea the Stasi were responsible. Many thought they were losing their minds, and mental breakdowns and suicide could result.

One great advantage of the harassment perpetrated under Zersetzung was that its subtle nature meant that it was able to be denied. That was important given that the GDR was trying to improve its international standing during the 1970s and 80s.

Zersetzung techniques have since been adopted by other security agencies, particularly the Russian Federal Security Service (FSB).[28]”

Fred Oyster August 26, 2014 5:01 PM

It’s an old post, but just to avoid this collection of hyperbole and supposition being cast in internet history, a few comments.

There have been microphones flown from aircraft for decades, not very successfully. The best example to date is NASA’s YO-3A variant that’s used to measure the noise from nearby helicopters. The turbulent flow from the airframe and propulsion system makes a lot of noise, orders of magnitude larger than anything the microphone would hear from the ground, except maybe gunfire. The Microflown, being an acoustic vector transducer rather than a pressure transducer, is particularly unsuited for a moving application. It can indeed be used as a microphone — I have done it, and heard speech through it quite clearly. The Microflown is simply two tiny hotwire anemometers aligned along one axis, being able to sense air particle velocity and sign rather than pressure. The actual area that’s heated is tiny and requires a very small amount of power. The significance of the Microflown idea is its ability to discriminate direction of sound arrival.

Leave a comment


Allowed HTML <a href="URL"> • <em> <cite> <i> • <strong> <b> • <sub> <sup> • <ul> <ol> <li> • <blockquote> <pre> Markdown Extra syntax via

Sidebar photo of Bruce Schneier by Joe MacInnis.