Speech recognition: less talk, more walk - Technology Information

The very public failure of Lernout and Hauspie is only one side of the story for voice technologies. Others say the industry is ready for big growth.

The speech recognition industry thinks it is poised to reap the benefits of the rise of third generation mobile networks. But before it can do so fully, say experts, quality issues still need to be addressed.

According to Analysys Ltd., of cambridge, England, a fairly basic automated speech recognition technology will function well where voice input is predictable. However, when no context is provided to distinguish between similar words or phrases, the technology can suffer from inaccuracy.

Although a question mark hangs over the future of Lernout & Hauspie, of leper, Belgium, a leading speech technology company that filed for chapter 11 bankruptcy protection in November 2000, few players believe that company's situation will affect the wider market, and most see its early potential finally coming to fruition.

Picking up where WAP left off

content companies have been among those quick to recognize the potential of speech technologies. InfoSpace corp., of Seattle, Washington is keen to embrace the opportunity. The content aggregator recently announced a definitive agreement to acquire speech technology firm Locus Dialogue Inc., of Montreal, Canada, last November. InfoSpace also signed a deal last month with U.K. mobile virtual network operator, Virgin Mobile, to provide its wireless content technology platform and mobile data services.

"There's been a surge in voice-based content because WAP has stuttered and stalled a bit," said Nick Henry, director of broadband and wireless services, Europe, at InfoSpace. "I can see tremendous opportunity in delivering sports results or stock tickers using voice."

But Frederic Gavignet, research and development engineer, speech technologies department, France Telecom research and development, of Lignon, France, believes there is still much work to be done in the field of mobile speech technologies. "Speech recognition is quite acceptable with a fixed line. but there are still improvements to be made on a wireless line. It's not so much to do with bandwidth as background noise."

France Telecom is pioneering work on speech-based Web interactivity and has created an experimental voice portal that follows the model of its Voila Web portal. "We are in a new phase. Next-generation technologies, particularly UMTS, will help to drive use and development of services that use voice," said Gavignet. "[But] users...should keep in mind the system has been conceived to recognize [only] a subset of any language and is limited."

Fueled by several sources

Others say it's not just the wireless arena that's driving renewed interest in speech technologies. "It's a ripe area. In the next year we'll see a lot happening," said chris Hart, managing director of SRC Telecom, London, a consultancy and provider of speech-based services and hosted applications. "It's difficult to distinguish between fixed and mobile. On the corporate side there's often no distinction but with many corporate applications more of the access is fixed line because more people call in from fixed lines."

Anthony Stevens, managing director of 20/20 Speech Ltd., of Malvern, England, a speech technology development joint venture between NXT plc and the U.K. Defence Evaluation and Research Agency, has identified three main markets: mobile services and the entertainment and education sectors. "In the near future people will need and want to be hooked up to a lot of information," he said. "I can see a raft of ideas in the education and entertainment markets. For example, the use of speech in multimedia games opens huge opportunities."

Others agree. "The new focus is on what you can do rather than the technology itself," said Mike Hartnett, director of business planning, Wildfire communications Inc., Lexington, Massachusetts "There are some places where speech is the natural interface. It should only be used where appropriate. I think as networks offer new services they become more complicated. Speech is an easy way to expose people to new services."

COPYRIGHT 2001 EMAP Media Ltd.
COPYRIGHT 2001 Gale Group