The power of the spoken word
Published May 17, 2007 at 3:24 pm · Filed under Features
To mark the 10th anniversary of Dragon NaturallySpeaking, Business Info spoke to Nuance regional sales director Neil Grant, about the future of continuous speech processing
Speech processing software, which converts the spoken word into text up to three times faster than typing, has traditionally been used in vertical markets where dictation is critical, i.e. in the legal and medical professions. With the release of Dragon NaturallySpeaking (DNS) 9.0, it is now ready for a much wider deployment.
“Over the last 10 years Dragon has improved substantially so the use of speech and dictation as an input method is increasing, Grant explained. “Improvements in speech processing are partly to do with improvements in software and partly to do with advances in processing speed and general computing power.
Because office workers no longer need to train DNS to recognise their voice, Grant expects the take-up of speech processing to accelerate, but only where there are real business benefits in its adoption.
“We are not saying that speech recognition technology is the only way to get dictation into a PC. It is all about increasing productivity. Sometimes it is quicker to use a keyboard and mouse. But in a lot of cases it’s quicker to talk, Grant explained.
As an example, he cites ‘high wrap’ agents in call centres who spend four or five minutes entering data into a CRM system immediately after each call. “The potential savings in cutting one or two minutes off data entry times are immense. Shaving a minute off the average call time in a call centre of 10,000 agents could save 20 million a year, Grant claimed.
But it is not just at the desktop that speech processing has benefits. It is increasingly useful for mobile professionals too. Dragon Mobile, for example, allows mobile phone users to dictate SMS and email messages and search the web completely by voice. According to Grant this is likely to be “a key growth area in the coming years.
For general office use, Grant conceded that there were still cultural and psychological barriers to be overcome, particularly in open plan offices where people may feel inhibited about talking at their computer.
“I would liken the use of speech processing today to the time 20 or 25 years ago when people hated having telephone conversations in open plan offices. There was a psychological barrier then, with people feeling uncomfortable that others might listen in to what they were saying. But today people happily talk on their mobile phones on trains without any concern that they are being overheard.
“With speech recognition, there is a similar feeling that people might be listening in when you create a document. But I think that will diminish in time in the same way that our approach to telephone conversations has changed, he said.
One application of speech recognition technology about which people have fewer reservations is as a machine interface. Speech can already be used to control computers, navigate telephone-based booking systems and make hands-free phone calls. And we can look forward to further developments in this area, though Grant says that there are still significant barriers to be overcome before we can use our voices to control all machines. These include background noise levels, the requirement to recognise numerous users and the processing power needed for large vocabulary continuous speech recognition.
“One area where we are making great progress is in-car systems. A lot of that is being driven by demand for hands-free operation of entertainment systems and telephones. As the computing power within cars becomes more powerful, we are able to put more speech-based technology within vehicles. Ford has just announced a collaboration with Microsoft, for which we have provided the speech technology, that allows people to sit in the car and say ‘call my office’ or ‘call my home’; enter destination information into navigation systems by voice; and control MP3 players by saying ‘play the next track louder’, for example. So a speech interface with machines is becoming more widespread all the time.
He added: “We are there with the software we are just waiting for the hardware to catch up.
