Smart speaker system uses white noise to monitor infants' breathing | Hacker News Reader

Smart speaker system uses white noise to monitor infants' breathing

techxplore.com

54 pointsdnetesn38 comments

Yup. So another excuse for the "smart speaker"[0] to be always listening and always transmitting data.

And honestly, I'm not against a microphone that's always on. But I am against a microphone that sends what it hears to the cloud, to be owned, stored and processed by some third parties in unknown ways. Ambient computing is a very fine concept on its own - but what it needs is to be pried out from the hands of commercial entities driving it. It should be implemented in such a way that the user owns the data, storage of that data is entirely decoupled from the software that's being run on it and from the cloud that brings the two together. That is, I should be able to choose software that operates on a data I own, in a place I want it to, and that software should not exfiltrate that data.

I guess it won't happen without a long legal fight that's still ahead of us. But until then, there's no way in hell I'm going to let such a device be anywhere near my kid.

[0] - It's funny how the name emphasizes the speaker aspect, whereas the core competency of such devices is the microphone.

I'm not sure where you're getting that this requires transmitting all the microphone data to the cloud. All of this could in theory be done on the device without cloud processing at all.

It could be, but it won't. The article mentions mainstream "smart speakers", none of which does it, and whose producers have every incentive to not do it.

Yet the photos from the article clearly show it running on a laptop.

> To make things easy for new parents, the team made a system that could run on a smart speaker that replicates the hardware in an Amazon Echo.

> "Smart speakers are becoming more and more prevalent, and these devices already have the ability to play white noise," said co-author Shyam Gollakota, an associate professor in the UW's Paul G. Allen School of Computer Science & Engineering and the director of the UW computing for health group. "If we could use this white noise feature as a contactless way to monitor infants' hand and leg movements, breathing and crying, then the smart speaker becomes a device that can do it all, which is really exciting."

They are clearly aiming to get this running on smart speakers.

I came to the comments just to voice a similar concern. We need to take ownership of our data, whether it is biometric, environmental or social. And once we have established this for ourselves, we should make it as accessible as the commercial offerings. This should be the new open source movement. Projects like OpenHAB and homebrew sensor networks with low power open source schematics and drivers.

Hey, we are already heading there. It’s just the masses that will stay ignorant for a decade.

> "We start out by transmitting a random white noise signal. But we are generating this random signal, so we know exactly what the randomness is," said first author Anran Wang, a doctoral student in the Allen School. "That signal goes out and reflects off the baby. Then the smart speaker's microphones get a random signal back. Because we know the original signal, we can cancel out any randomness from that and then we're left with only information about the motion from the baby."

Reading between the lines of this, it sounds like these folks found another clever use for adaptive filtering. I wonder how well this plays along with whatever adaptive filtering is already going on inside an Amazon Echo?

As I've learned more about the guts of voice speakers' signal processing, I've come to realize that adaptive filtering plays a huge part in making modern acoustics and telecom possible. It's a super cool subset of signal processing if you're interested in that sort of thing.

https://www.mathworks.com/help/dsp/ug/overview-of-adaptive-f...

It seems like the equivalent of a DSSS radar, but with sonar.

Not sure, but I wouldn't be surprised. I don't understand much of how radar works, but my understanding at a really basic level is that you're sending out some known modulated signal and then extracting some information from the magnitude and delay of that returned signal.

In that case, this is the same idea, but in the adaptive filtering case, it's actually the errors in the received modulated signal that you care about, because the error contains all the signals you didn't originally transmit - things like room sounds, or audio multipath.

I will admit, however, that I don't know enough about radar to know that we're not talking past one another. XD

I can see this being useful in the case of extremely premature babies, or those significantly more vulnerable than usual. But for the average new parent this is just another way of increasing anxiety. In my humble non-medical-expert opinion, you really shouldn't get one unless you're told to by someone qualified.

Technology always leads to new norms and standards of living. We have medicine that cures infections that would have absolutely killed us a hundred years ago. We have machines that can route blood around your heart while doctors replace it with another one. Here we have a technology that has the potential to reduce the occurrence of SIDS(sudden infant death syndrome) at a cheap cost once the tech is commoditized.

I understand your hesitance in promoting anything that can exacerbate the anxieties of people unnecessarily(our society is already bad enough at this), but as a parent, that anxiety was already there for me, and I think it's there for most other parents too. I used to wake up and check my daughter in the middle of the night because she hadn't cried in a while. This would have brought me peace of mind.

Same here. As new parents, we followed all guidelines for preventing SIDS but still were always checking on our son all the time.

I highly doubt we would have trusted any device 100%. But it would have brought a little bit of peace of mind which new parents desperately need.

New parent here - there's a learning curve and once you are over it (ie have balanced your anxieties with the realities that your child is probably ok seeping at night) you will have traded off your privacy for a device that to your own language you wouldn't trust 100%. It feels like this is just a push for positive stories on privacy infringement to make it easier to buy the products. I.e. "We got these spy devices in our house, but for the first 3-6 months of our new child being in the house, it monitored them for breathing"

The 'trust' in the post wasn't about privacy but about the breathing monitoring I'm pretty sure. I.e. they would have still worried about their kid some even with the device.

...your child is probably ok seeping at night

"probably ok" is exactly the level of confidence that left me with anxiety around this issue. It's not exactly a convincing statement to make for people worried about whether their child is still breathing or not. Regardless of the rationality of the concern.

...as a parent, that anxiety was already there for me, and I think it's there for most other parents too. I used to wake up and check my daughter in the middle of the night because she hadn't cried in a while. This would have brought me peace of mind.

This echoes my own experience in this area.

And what would you do should this device mislead you? Sue the company making the nonmedical device?

Might as well just attach a sensor directly to the kid instead. Something tiny, like wireless EMG electrode or MEMS accelerometer. Much less likely to get a false positive. (The problem is batteries, as usual.) Heck, a pulseoximeter.

If you tell people to do what "someone qualified" says, you're not offerring an opinion, humble or otherwise.

It's a nice to think there are qualified people who can advise parents on this, but they don't exist. SIDS strikes even healthy infants. It's caused by a weakness in an osciallator network in the infant's cardiovascular system (modulo some percentage of infanticide). There's no known test for it but there is a simple solution: allow it to entrain to the mother's breathing by placing the infant near the mother.

Another way to feed your paranoia as a new parent. There are other monitors already which purport to monitor infants breathing, and I haven't met anyone yet who used them who recommends them. Inevitably they give false positives, and it's hell on your nerves. After a while you learn to just ignore them so you don't go insane. Then you take out the monitor because what's the point.

This is the internet, though, so people like me will always exist.

We bought a fairly cheap wearable monitor early on, and it works. Both in the sense that it has given us a couple of false positives when it got out of position, and in the more important sense that it has only done so a handful of times and the rest of the time I sleep much more soundly because I know it would go off if our child stopped breathing.

Honestly, this is not a truly hard problem to solve, and I'm kind of amazed that the APA hasn't figured that out yet. On the other hand, obviously the effect on different parents may vary.

What use is it to know when your child stops breathing? Is there anything you can do about it?

CPR, artificial respiration, call an ambulance . . .

I wonder what the survival rate for infant CPR is. I know for adults it's shockingly low, particularly without an AED. Pediatric AEDs exist, but I doubt many concerned parents have one.

Don't get me wrong, I support people learning CPR. It's better than nothing.

as I understand, SIDS is sometimes (not always!) literally that an infant is sleeping so deeply that it forgets to breathe. Sometimes simply rousing them from the deep sleep would be able to correct the situation if it could be detected.

Oh wow. You reminded me of a mildly traumatic memory of my infant daughter literally turning blue before my eyes before taking a gasping breath and going back to normal. It was no more than a couple of seconds (babies turn blue fast), but it certainly sent my heart racing . . . I think she was crying very aggressively to get herself into that state; I forget the exact details.

> This is the internet, though, so people like me will always exist.

Touche. Definitely experience will vary. In my circle of friends, we were using angelcare monitors about 10 years ago, which to be fair are more about sensing movement than trying to directly detect sleep apnea. Nobody liked them, everyone stopped after a while and decided the very small risk of SIDS wasn't worth the suffering involved in trying to detect it.

That said, one of my friends did have a son pass away (on his first birthday, no less, which is pretty late for SIDS IIRC) during the night for unexplained reasons, so I can't really blame parents for being paranoid. You have to find what works for you, and for everyone that's going to be different.

Im curious to know what cheap wearable monitor you used. I will be having a boy 1st of the year and would like something like this.

We use a baby monitor in conjunction with specifically patterned clothing. The monitor observes the pattern and assesses whether the baby is breathing - if not, it will alarm on both the phone and the crib monitor unit.

Is it strictly necessary? No. But it helps me sleep, and that means the baby sleeps better because he's not being visited and monitored by a human.

We've had one false positive in 6 months. That's an acceptable tradeoff, in my mind.

Soon to be parent here. Could you share which device you've used with good results?

This is the Nanit monitor with the Breathing Wear band. The cost is significant - $300 for the monitor, $20+ for the band, and an annual fee after the first year. With that said, an IP-enabled monitor lets you do things like spend time outside the range of a non-IP camera (e.g. on vacation, go to the hotel pool while the baby naps), things I find extremely valuable. Naturally, there is a privacy impact that should be carefully considered.

Thank you. This is exactly what I was looking for!

Relevant to SIDS tech:

https://tricorder.xprize.org/prizes/tricorder

https://www.cringely.com/2013/01/22/qualcomm-tricorder-x-pri...

I wonder if there's much preventing such tech from being reapplied for, say, furtively monitoring keystrokes or other forms of questionable eavesdropping, in a kind of acoustic Van Eck phreaking sort of way.

Can anybody help me try to understand it? I imagine they play a white noise and use their directional mics to get the result of interference of white noise and breathing sounds. But "noise is reflected back" makes it seem like radar. Why is it not sonar/radar then?

And with "With this skill, called BreathJunior", it makes it sound like existing smart speakers can do it. This has nothing to do with smart speakers right?

Tangentially similar breath and heartrate monitoring by magnifying changes in video https://www.youtube.com/watch?v=3rWycBEHn3s

Or you could just sleep with your infant. Note the dystopian photo at the top.

Or you could be awake with your infant. I'm pretty sure both my infant and I sleep a lot better when we're in different rooms.

But yeah, that's a sort of creepy photo.

It’d be crazy to rely on something like this. I think you’d want the analogue to be much more robust if you’re gonna rely on it keeping your child alive...

Reply on news.ycombinator.com