Cell Phones That Learn the Sounds of Your Life 121
An anonymous reader writes "Researchers at Dartmouth College have developed new software that uses the microphone on the iPhone to track and interpret a user's everyday activities using sound. The software, called SoundSense,
picks up sounds and tries to classify them into certain
categories. SoundSense can recognize completely unfamiliar
sounds and runs entirely on the phone. It automatically classifies
sounds as 'voice,' 'music,' or 'ambient noise.' If a sound is repeated
often enough or for long enough, SoundSense gives it a high 'sound
rank' and asks the user to confirm that it is a significant sound and
offers the option to label the sound. In testing, the SoundSense
software was able to correctly determine when the user was in a
particular coffee shop, walking outside, brushing her teeth, cycling,
and driving in the car. It also picked up the noise of an ATM and a fan in a particular room. The results [PDF]
of the experiments were recently presented at the MobiSys 2009 conference."
Privacy concerns? (Score:1, Informative)
FTA: To address privacy concerns, they designed SoundSense so that information is not removed from the device for processing. Additionally, the program itself doesn't store raw audio clips. A user can also tell the software to ignore any sounds deemed off limits.
I can think of lots of fun and interesting uses for this kind of monitoring. And I can't think of many reasons to limit what sounds my phone can hear. Total capture and processing of the sounds of my life could have remarkable implications which I'm sure Charlie Stross or somebody is already writing about.
I'm concerned over the privacy issues. And these are total dodges on the privacy front. By the time the useful information has been processed, listed, and categorized, the raw data doesn't count for much. It's the end result of the algorithm, not the data that goes into it, that is of interest to advertisers, nanny states, and Generally Evil People. Unless maybe you're trying to use the phone as a bug. Fortunately, I rarely say (or arguably write) much of interest to anyone but the voices in my head.
The crux of the privacy issue is this, for me: I want *control* over *who* gets to looks at the results of the reality mining algorithm. If it's going to an app that determines how best to advertise back at me, well, fuck that. Or some bureaucrat who wants to make sure I'm not engaging in troublesome activities; fuck that too. If that's the price, I'll forgo the experiment and just keep track of things myself, thanks.
Re:Privacy (Score:3, Informative)
GPS can tell you where you are within a few inches.
Your mapping software is what lacks in precision.
Re:Privacy (Score:1, Informative)
That's actually entirely incorrect, GPS is able to track Latitude and Longitude as well as Altitude.
Check the following link for more information:
http://gpsinformation.net/main/altitude.htm
Re:Privacy (Score:5, Informative)
Not only did you get it wrong, but most of the people replying to your post got it wrong as well.
GPS comes in several varieties. The common one used in cheap handheld units are using the C/A signal [wikipedia.org]. This gives you an accuracy down to about 5 meters (due to multipath and atmospheric interference). The accuracy is determined by the precision of the built in clock which in term determines the size of your unit. C/A gets you down to about 5 meters accuracy.
Add in DGPS [wikipedia.org] and you more or less eliminate atmospheric interference which can get you down to about half a meter. Technically 20 inches can count as a few inches when compared to the 200 otherwise.
P-code (the military) gives you down to about 2 meters accuracy by comparison. Not sure how much better they get with DGPS, but I'd suspect it'd get them to 1/10th just like the C/A does.
For the best accuracy you won't be relying on L1 and L2 directly (decoding the signals), but will be looking at the carrier phase change [wikipedia.org] which requires bigger and better antennas as well as a much more precise clock which is why most of these are big, bulky and used for surveying more than anything.
If you're moving around (airplanes pictures and very likely road surveying as well) you can get down to about half a centimeter, but expect from 5 mm to 10 cm). When not moving for a significant amount of time, you can get down to 2 mm to about 3 cm)
As for one of the replies claiming that GPS only gives you a 2D location, this is rubbish. You need a clear view of a minimum of four satellites to get a proper height as well as longitude and latitude. The accuracy of each of these coordinates varies as well . However the biggest inaccuracy you're likely to face when dealing with GPS is using the wrong datum [wikipedia.org] for your map. It's fairly easy to end up with coordinates several hundred meters from the correct one, merely by forgetting to switch datum when moving into a new area.
If you're not using the US GPS but instead rely on Glonass [wikipedia.org] (not done yet) you get some advantages. Since each satellite is running its own discrete frequency, you're essentially able to rule out atmospheric errors. This adds to the cost of the unit though, as it now needs a much more advanced radio receiver. I can't remember if Galileo [wikipedia.org] uses one or multiple frequencies. I think it's supposed to use two, but it's not that important.
Granted, it's been a while since I worked heavily with GPS theory, so feel free to correct me (if you can cite properly). I'm using Wikipedia because I doubt most of you are able to read Danish which is the language my text books on the subject.