Voice recognition is already used in call centres and is set to
become an indispensable remote working tool
People have been working on the development of voice recognition
software for the past 30 years but only fairly recently has the
technology reached a point where it can be more widely used, writes
Ross Bentley.
"The underlying technology has come on in leaps and bounds in the
past few years," says Paul Barnes, chief executive at Fluency Voice
Technologies, a UK voice recognition specialist.
"Specifically, the technology now has the ability to understand
regional accents and to react with speed," he says.
"There has been a lot of work done with phonemes, which are the
building blocks for voice recognition.
"Once the technology reaches the point where it can be used
commercially, this acts as an incentive to develop the technology
even further."
According to Barnes there are three areas where voice recognition
is already having an impact. "It is being used extensively in call
centres and for virtual personal assistants," he says. Virtual PAs
enable people to remotely dictate memos, access e-mails, arrange
conference calls and "do all the stuff you can do with Outlook
using voice commands".
"Telematics is another area which we are seeing come alive. This is
voice-activated functionality within the car, where the driver
needs to do things hands-free, such as route planning and making
phone calls.
"We are also looking at adapting these systems for boat operators,
while Nasa is looking at incorporating the technology into the
space shuttle," says Barnes.
Microsoft is pushing the development of voice recognition with Salt
- speech application language tags. It says that Salt is a more
viable alternative to voice XML.
"Microsoft is saying that it plans to prove that anything you can
do with a keyboard, you should be able to do using voice
recognition," says Barnes, but he sees many other uses for voice
recognition software over the next few years.
"If, for example, you go on a spending blitz with your credit card
it may be seen as out of the ordinary and set off an alert at the
credit card company," he says. Unusual changes in spending
behaviour help credit card companies to spot fraud.
This alert would activate a phone call to the card holder, where a
computerised voice would ask for verification of the amount spent.
"If there is a problem the call can be patched through to an
operator at a call centre," says Barnes.
"This is one of the first proactive uses of voice recognition where
the software actually initiates a call.
"In tests we have found that most people don't mind being called by
a computer if the call is effective and doesn't waste time."
Barnes says translation software is an area where there has been
real improvement. "Translation functionality pushes the technology
to its limits.
"While there's nothing out there that really works as a broad
translation module because there are so many things you can say,
translation software does work when applied to niche areas."
Fluency recently took part in a project funded by the Japanese
government to provide a specialist translator to translate medical
questions from Japanese to English and vice versa for Japanese
doctors.
"Translation software was able to work here because of the limited
number of words used," says Barnes.
"There is a limit to the amount of phrases needed, it is tight and
defined, such as, 'Do you have a headache; do you have
palpitations?' - things critical to diagnosis."