Keyless cellphones and voice as a means of personal authentication are not far away, says Tim MacNamara of voice recognition (VR) specialist Nuance.
Vodafone, in collaboration with Nuance, which produces VR middleware, and Procom, which makes devices, is already working on a phone to be operated entirely by voice. It will save the awkward transfer of the phone away from a user’s face to key numbers then back to the user’s face again to talk.
In Australia, too, it could mean the difference between a trouble-free car trip and a police ticket; it is illegal there to operate the keys of a conventional cellphone while driving.
MacNamara, Nuance’s Sydney-based sales director for Australia and New Zealand, was updating a Tuanz audience last week on the progress of VR technology.
The voice, MacNamara suggests, is a more seamless way of operating internet-type services through a phone than is the usual WAP thinking of keying and clicking text menus or web pages on a small screen.
Another of Vodafone’s projects, in cooperation with Ford, is to make the car itself a phone and voice recognition device, using microphones in the door window pillars.
Telecom in New Zealand is starting on voice-based information services with WordUp, a “voice portal” launched quietly earlier this year.
It currently provides a small range of information services, such as news, sports news, Lotto results and horoscopes, operated through voice menus.
The killer application, says Telecom spokeswoman Katherine Scott, will be email.
However, in the present form of the service this only uses voice recognition for commands. Received emails are read out, but replies are simply recorded as .wav files and played to the recipient.
The potential of even the straight information service is demonstrated by US voice portals TellMe and BeVocal. TellMe, for example, has a worldwide list of restaurants; so if you are travelling to Silicon Valley, you will be able to track down a good spaghetti meal in Palo Alto in advance — then the same phone service can put you through directly to the restaurant. But you have to cultivate a reasonable imitation of an American accent to be understood.
Nuance has constructed a mathematical model of the Australian-New Zealand accent, says MacNamara. This means voice recognition technology can now be deployed reliably for telephone and online sales and service in the two countries, catching us up to a technology already being used with notable success in the US.
And yes, the New Zealand and Australian accents are really very similar, says MacNamara, apart from a few nuances.
The model of both accents consists of 46 basic sounds, each of which can be modified by the flow between it and the sounds preceding and following it.
The unreliable VR technology of a few years ago does not stand comparison with present-day technology, which is based on totally different thinking, he says.
He claims an 85% success rate in enabling the Queensland TAB to take bets directly on the phone, without the intervention of operators to write them down.
The TAB in New Zealand is understood to be trialling VR.
MacNamara adds the voice recognition systems at the TAB and US airlines are even adept at understanding the voices of tired people or those slightly the worse for drink.
And voice recognition systems, he says, can be used as a “biometric” identification device — no more hard to recall PINs and passwords.
Already, it has been practically demonstrated that a telephone banking system can authenticate individual customers by voice, with no possibility of fraud, he says.