> Yikes, please don't do or encourage using these in public - there are many accessibility devices (hearing aids, cochlear implants, etc.) which depend on MEMS microphones to function.
> You could inadvertently make the world much worse for people who already have a difficult time of things. Imagine carting a cellular and WiFi and bluetooth jammer around outside of a Faraday cage - it's insanely irresponsible and inconsiderate.
It would be a loss if due to a critical need to protect privacy we end up inadvertently harming a group of people that are already at a disadvantage. I once knew a hearing-disabled person and the hardship it brought them affected nearly every interaction with them I had. For me that trade-off is not worth it.
If the best channel to help the deaf listen clearly is also the best one for letting eavesdroppers listen clearly, then this is a problem best not handled with such a heavy handed solution - possibly not even a technological solution at all.
> It would be a loss if due to a critical need to protect privacy we end up inadvertently harming a group of people that are already at a disadvantage. I once knew a hearing-disabled person and the hardship it brought them affected nearly every interaction with them I had. For me that trade-off is not worth it.
We already do this by making so many things practically require internet access but then also not making things easily accessible to eg screen readers or text mode browsers
The law has been well-received both by the public and by businesses, with very few actual fines being issued, but many bad pages have been soft-forced into improvement. So it's actually possible, but will probably be much harder in the US, in which business, government and taxpayers/customers seem less aligned.
Or any hearing aid? That said, hearing aids in particular employ circuitry specific tuned to only amplifier the desired sound and frequency range intended to be heard, modern ones having profiles for normal speech, congregations, music, etc.
Microphones pick up harmonics of the input frequencies. So if the ultrasound white noise is a multiple of the human voice range, it'll get let through.
The interesting parts are those 25KHz transducers which seem identical to the 40KHz used since likely forever in ultrasound remotes and more recently in collision avoidance sensors for robotics. I did a small search and found mostly high powered ones at that frequency, probably ultrasound cleaners spares, or smaller but a lot more expensive transducers compared to 40 KHz ones.
Does anyone know of a source for these transducers?
I also wonder if a simpler approach could be used since the purpose appears to be (can't understand the math) generating noise by driving randomly a number of oscillators around the transducers resonance frequency then induce subharmonic vibrations into the MEMS mics through etherodyne operations between these sounds. If that's how it works, then the DDS chips, the Arduino and the code might be swapped with a less random but likely equally functional set of dissonating oscillators modulated by LFOs (all doable with plain old logic gates); not unlike the old school way of generating cymbals metallic sound in analog drum machines. Here's the Boss DR110 relevant schematic as an example.
it seems to me that this is largely an attack on common preamplifier circuitry. would it be sufficient to ensure that the preamps implement low pass filtering? or is the issue more in the microphone element?
"For the above idea to work with unmodified off-the-shelf microphones, two assumptions need validation. (1) The diaphragm of the microphone should exhibit some sensitivity at the high-end frequencies (> 30kHz). If the diaphragm does not vibrate at such frequencies, there is no opportunity for non-linear mixing of signals."
The devices tested include hearing aids, smartphones, smart watches, etc, which are all likely to include small surface mount MEMS microphones. I doubt any of these techniques will work against a larger dynamic or condenser microphone, where the mass of the diaphragm makes the system inherently insensitive to ultrasonic frequencies. There's a reason the jamming signal is inaudible, and it's not because our auditory cortex contains an ideal lowpass filter.
Yes, I agree that the microphone has to be responsive to ultrasound. That doesn't seem to be the only requirement, though:
"When these tones arrive together at the microphone’s power amplifier, they are amplified as expected, but also multiplied due to fundamental non-linearities in the system"
"In practice, however, acoustic amplifiers maintain strong linearity only in the audible frequency range; outside this range, the response exhibits non-linearity."
That suggests to me that the nonlinear mixing isn't occurring in the MEMS structure, but rather the amplification stage. Perhaps the authors' language is imprecise?
They do say immediately after the last bit:
"The diaphragm also exhibits similar behavior [non-linearity]."
Is just the diaphragm's nonlinearity sufficient for the effect?
Ok, that's actually a lot more interesting than I was giving them credit for. I assumed they were just overwhelming the LPF or exploiting some side lobe to create aliasing in the audible range, but that's discussed in section 2.
All I can contribute is that nonlinearity in audio systems is also known as distortion, and it's impossible to eliminate entirely because every system, whether electrical mechanical or even digital, goes nonlinear when it reaches its amplitude limits. Some more gracefully than others.
Seems like it would be difficult to tease out the relative contribution of the MEMS element and the preamplifier because the preamps are typically implemented on ASICs in the same package. So that might be speculation on the authors' part.
This is really neat work. I'm curious to see if smaller and more portable versions of this can be made - obviously a prototype is always going to be bigger and be relatively limited. Now that the research has been surfaced, this appears easy enough to follow so it'd be neat to see how electronics enthusiasts run with it.
Do cell phone speakers operate at this frequency range as well? I could imagine an app specifically focusing on responding to the words "to be safe", by turning on the white noise maker.
I think this is a neat idea, but I suspect it would be more useful as a standalone device than as a wearable. Devices like this could be installed in secure rooms or deployed on the fly in discreet locations with a high rate of success, I'd guess. Arrays of them could work together for better coverage.
Still, this may be the only real option in public spaces (i.e. outdoors). If you're okay with people knowing that you're trying to avoid being recorded, then this would probably be fine.
They make it a wearable to counteract the nulls caused by the multiple speakers. Random movement causes the nulls to not stay in one place for very long, limiting how much of a word or sentence a microphone could pickup. You could build a phased array of these that could quickly move a hotspot of ultrasonic noise all over the room. You could them position on the ceiling with a fixed radius between them to make sure that the highest pressure occurs at about waist level, where phones in pockets, smart watches on wrists, and smart speakers on tables would reside. Another idea would be a ball of these in the center of the room and have it move up and down to get the best average coverage around the room.
Smart assistants are usually not recording really high quality audio, it takes more time to process it and more time to send it back home so they are going to a lower sample rate than typical voice recorder app would use. Siri uses a 16KHz sample rate (Fs=16KHz) which is enough to put the whole human vocal range in the 1st Nyquist zone (less than Fs/2). Playing a sound at 26KHz (3rd Nyquist zone, >Fs but <1.5*Fs) is going to cause a reflection across Fs. So the 26KHz tone, sampled at 16KHz, creates a tone at 10KHz which could be enough to confuse a naive implementation of a smart assistant. Ideally, you want fix this by either installing an analog filter so the ultrasonic noise can never reach the ADC or sample the whole range (up to 44.1KHz is a good start) and filter digitally.
There is a paper called DolphinAttack [1] where they attempted to use the ultrasonic audio band as an inaudible attack vector. You could play an ultrasonic noise that no one can hear except for the smart assistant.
This only defends against off-the-shelf mikes, and adversaries will just get better mikes so not particularly useful in the security space.
It is illegal public and most commercial spaces as most organizations need to follow anti-discriminatory legislation. Spaces unusable by people with hearing aids, hearing dogs or just very good hearing is not going to happen.
They mention that the wearable component was a conscious design choice to overcome certain limitations of standalone devices:
(On standalone device limitations)
>(2) They rely on multiple transducers that enlarge their jamming coverage but introduce blind spots locations were the signals from two or more transducers cancel each other out. If a microphone is placed in any of these locations it will not be jammed, rendering the whole jammer obsolete.
>To tackle these shortcomings, we engineered a wearable jammer that is worn as a bracelet, which is depicted in Figure 1. By turning an ultrasonic jammer into a bracelet, our device leverages natural hand gestures that occur while speaking, gesturing or moving around to blur out the aforementioned blind spots.
I envision a variation that sits in the corner of a room, similar to how a fan operates, gradually turning the speakers. However, as others mentioned, I'm curious over the effect it would have on hearing aids, etc. Secondly, if there a variation of this device that could straight up disable WiFi?
Right, that'd be my concern - that if their implementation could be modified in a similar manner to create jammers that would be quick to deploy but difficult to locate.
You could just have a simple Wifi deauth beacon. You can do it with a simple ESP8266 and could either kick everything off the network or target blacklisted MAC address ranges. It's usually illegal to do this to someone else's network just like running a real Wifi RF jammer.
It's only "wearable" if your definition of "wearable" doesn't take into account style. I mean, come on! The space age design wouldn't pair well at all with my wardrobe. Can you imagine that clunky thing over the cuffs of my Ralph Lauren suit jacket? Rating: 0/10. under no circumstances would I don that monstrosity.
Edit: sorry, I went against my better judgement and decided to respond sarcastically to the other comment. I was being facetious.
I can see this being useful for blocking your smartphones and Alexas, but it seems like devices specifically designed for surveillance could start being designed to fix this exploit.
Currently wearing noise cancelling headphones and wondered what would happen with them, but hearing aids would be a much bigger concern. I would wonder if since hearing aids are so focused on human hearing that they would have filters that would block out any sounds outside the human hearing range.
A similar effect occurs when using infra-red LEDs near the face to prevent video recording. The problem is that many phone cameras now have a UV/IR filter - I can see microphones having a similar setup to improve sound quality in the future.
All CMOS cameras have an IR filter, for reasons that have nothing to do with people using IR LEDs to fool (some) cameras. In fact, removing the IR filter and replacing it with a visible light filter is a cheap way to turn a cheap webcam into a (crappy) night vision camera.
This might result in jammed audio for human listeners, but recovering the original audio seems like a fairly mundane signals extraction problem subject to the standard signal/noise ratio issue.
It's not an advancement thing, its a cost thing. They made the filter good enough to block out the stuff people can't hear but this device is super loud so you would need additional layers of filtering to completely block it. Smartphone mics need to be tiny so additional analog filtering is going to take up space and resources. You could also run a higher sample rate on the ADC that converts the sound to a digital signal and run a digital filter to cut off the ultrasonic band but that requires more power and chip resources and the ADC might need to be swapped for one capable of the higher sample rate. The tools are all there to defeat this but it's a matter of reducing power, cost, and computing resources.
Perhaps, though not easily. You could be looking at getting harmonics on the physical components of the microphone from the ultrasound which would bleed all the way through the normal sound ranges. Plus, with enough sound pressure, the microphone may be physically maxed out (that is, physically/electronically unable to register any further sound pressure).
Seems like countering it might be a use-case only important to specialty buyers who want to get around this kind of protection. Even if it wasn't that difficult to counter, there might not be enough incentive for most off-the-shelf mics and specifically Alexa/phone/etc to do so.
Would it? I'm not aware of any laws about sound that wouldn't bother the human ear. The closest thing I can think of is laws about RF jamming but this clearly isn't a radio frequency.
Hearing aids, emergency telephones and phones used by medical and security personnel are already covered by laws in many places. Also messing with law enforcement devices, like police body cams. The people developing this could technically be liable right now if they tested the device in a public space. It will never get as far as the University library. The main use of this research will be in selling better mikes to the military and police.
It seems a little irresponsible for an academic publication to make a claim like
> always listening, recording, and possibly saving sensitive personal information
without any evidence to support it. I get that they're just setting up context for their device, but they're also making some pretty serious (and widely disproven) accusations.
All of these devices have been reverse-engineered and packet sniffed to death, and no one has ever produced any evidence that they're doing anything other than what it says on the box.
While I do appreciate that the cited sources in their paper, I would have appreciated actual information security papers rather than mainstream media articles.
