Millions of Voiceprints Quietly Being Harvested

An anonymous reader sends this excerpt from The Guardian: Businesses and governments around the world increasingly are turning to voice biometrics, or voiceprints, to pay pensions, collect taxes, track criminals and replace passwords. "We sometimes call it the invisible biometric," said Mike Goldgof, an executive at Madrid-based AGNITiO, one of about 10 leading companies in the field. Those companies have helped enter more than 65M voiceprints into corporate and government databases, according to Associated Press interviews with dozens of industry representatives and records requests in the United States, Europe and elsewhere. ... The single largest implementation identified by the AP is in Turkey, where the mobile phone company Turkcell has taken the voice biometric data of some 10 million customers using technology provided by market leader Nuance Communications Inc. But government agencies are catching up.

Comments

Stephen Hawking look out.

[dwywit]

I can see a rapid increase in the customer base of synthetic voice software

Re:Stephen Hawking look out.

[RabidReindeer]

User Identified.

Welcome, Stephen Hawking.

Are they collected by hot women

[Anonymous Coward]

with 80s hairstyles? If yes, here's mine: My - voice - is - my - passport.

Re:Are they collected by hot women

[pr0t0]

Don't mind the young haters. I came to type the same thing. Would that I had karma to give.

please - verify - me

Re:Are they collected by hot women

[neorush]

http://www.youtube.com/watch?v=-zVgWpVXb64

Always use voices.

[Anonymous Coward]

Donald Duck gets customer service satisfaction and no BS.

US Customs

[Anonymous Coward]

I recently returned home from an international trip. I don't travel outside the country very often, and this was my first encounter with the new kiosks that replace the old paper form asking where I went, why I went there, and what I brought home with me.

I was also fairly sure that the reason the Customs agent asked me to look directly at him and state my full name was that he was collecting a voice sample for future use. I think this article confirms that either this is already happening, or will very soon.

Re:US Customs

[AHuxley]

re "Well, speaker recognition dates as far back as facial recognition does."
It was big in the 1980's to find interesting people using different phones and very early cell phones in South America by the US mil/gov.
The UK enjoyed using it "Spy-in-sky patrols over British cities in hunt for Taliban fighters" (3 August 2008)
http://www.dailymail.co.uk/new...
"They are attempting to identify suspects using ‘voice prints’" ... "Traffic’ intercepted by the equipment on board is analysed and processed, probably at the GCHQ spy centre in Cheltenham, searching for voice matches"
The only real change is the cost of collection, cost of sorting and ability to build on public, telco, private and mil databases been shared with the Five Eye nations and friends.
Expect every arrival chatdown to be recored and indexed with your face, passport and the usual biometrics details.
Expect every car rental, duty free, cafe airport chatdown to be recored and indexed with your face, passport and the usual biometrics details.
Get you talking, keep you talking, its not just about the car rental use or been friendly to the wider travelling public ;)
Domestically the telco and trusted brands Interactive voice response (IVR) will record the rest of the wider populations over years.
That 1 or 5 min chat with Bob or Sally in telco support is not just kept for training purposes or quality control :)
Your billing details have been matched with something globally unique.

Re:US Customs

[Neil Boekend]

With the youtube videos a computer has difficulty determining who is in the video.
At the airport you have to identify yourself. Your name is entered right with the voice data.

Over the phone?

[TubeSteak]

The Vanguard Group Inc, a Pennsylvania-based mutual fund manager, is among the technologyâ(TM)s many financial users. Tens of thousands of customers log in to their accounts by speaking the phrase: âoeAt Vanguard, my voice is my passwordâ into the phone.

The problem with biometrics is that you can't ever replace them if they are stolen.

So naturally they want to use a password that you have to announce publicly. :facepalm:

Re:When is enough enough?

[torsmo]

It feels so hackneyed at this point to try to describe the dystopia we are headed toward (already in?)

I dunno, it seems to me that an advanced technology-enabled dystopia is not universal in its spread. Poor third-world countries do not seem to be headed that way. Corruption may have something to do with it, but I'm too tired at the moment to form a logical construct to validate that thought.

Re:When is enough enough?

[Anonymous Coward]

Third-world countries are waiting for the run-off-the-mill cheap Chinese equipment. They have the rest of the dystopia either working, or ready to go on a short notice.

Re:When is enough enough?

[AHuxley]

Not much people can do about tame Interactive voice response (IVR), calls kept for kept for training purposes over the years and passed onto gov/mil.
The good news is people now know more and know of the public, private , gov and mil sharing of tech like voice prints and the low cost of huge generational databases.
What was once used to track high ranking Soviet officials in realtime and people of interest in South America is now at home, cheap and for 'legal' domestic use.
With the added features on International Mobile Subscriber Identity catchers and the next gen IMSI-catchers expect the voice print tech to be in the hands of city and state officials.
Driving your car near any protests with powered cell tech and city parallel construction kit might just log all.

This is going to backfire in an ugly way

[duck_rifted]

My voice sounds different on every single device I've ever heard it played back from. That's in addition to it sounding different based on the time of day; bass in the morning, flat at night. On top of that, it sounds different based on my mood and health. So, this has a high potential for false positives and false negatives.

Then there's the matter of reproducing voiceprints. People have done that for decades for practical jokes, comedy routines, and more. It's not only possible; it has been done already and can readily be done by anybody who puts a little effort into learning how.

Finally, there's the matter of fraud. Combine the two above observations, and your bank can forge your "voice signature" and then play back audio if you can even afford to take them to court. Viola, the banks literally own absolutely everything and nobody has property rights.

Brilliant.

Using tech like this to improve voice recognition and speech synthesis is useful. Using it to verify identities is problematic and should be banned before it causes any serious problems, destroys lives and livelihoods, and wastes resources and time. This is quite possibly the worst, most easily abused application of technology I've ever heard of any government or institution being idiotic or corrupt enough to try.

Re:This is going to backfire in an ugly way

[coofercat]

About 20 years ago, Nuance did a demo where they played a recording of Margaret Thatcher, which the system identified as being her. They then asked an impersonator to do the same thing, but the system could tell they weren't Mrs T.

I'm sure these sorts of demos are part truth and part smoke-and-mirrors, but the point being that it was a long time ago and something they claimed to be able to do. You can bet they've got it sorted now so that it's considerably more accurate. I have no idea if they have a means to stop a recording being played back though - that's presumably a harder problem to solve.

Is this one way?

[Skylinux]

What happens when someone steals the db. Will they now have the data required to emulate the owner's voice against other sound based authentication methods?