IBM, Microsoft and Hewlett-Packard are all working on new software that will, hopefully, recognize voices in much the same way that scanners can recognize fingerprints. It would allow banks to check identities without requiring people to install new hardware. The only thing they would need is for people to speak into a phone.
As speech-related biometric technology joins the fight against identity theft, your credit card may start asking you to speak louder Most consumers, at one point or another, have thought about how easy it would be to steal an identity, particularly over the phone: You call your bank.
To verify that you are who you say you are, a clerk asks for a Social Security number, address, date of birth, or account number.
A number of companies, including IBM (IBM), Microsoft (MSFT), and Hewlett-Packard (HPQ), have recently developed new biometric software and devices designed specifically with the phone in mind.
Their solutions to the phone-security conundrum range from embedding detectors such as fingerprint scanners right into mobile phones and personal digital assistants to using a promising new biometric technique called voice verification.
En-masse deployment of voice-verification technology could happen within a year, with sales of related software and devices expected to rise from $45.9 million last year, to $224.6 million in 2008, according to researchers at International Biometrics Group, an independent industry researcher.
Here's how it works: A special sensor on the credit card stores its owner's previously recorded voiceprint in digital form.
Many telecommunications companies and banks are also looking at the other kind of voice verification, which requires no alterations to a phone.
Caller-identification technology that software maker Nuance (NUAN) unveiled a year ago is already used by Canadian telcom Telus and is being tested by several U.S.-based banks and credit-card companies, says Nuance CEO Chuck Berger.
It works on a simple premise: Customers make a short voice recording.
The next time they call, the technology compares their live voice's range and speed with the recording.
A 20-second recording collected using this new method, which is awaiting commercialization, could identify customers with what developers hope will be nearly unfailing accuracy, says IBM researcher Ganesh Ramaswamy.