New ways to say 'hands free'
Local company, Nuance, acquires quite a reputation (and market share) with its voice-recognition products
By Scott Kirsner | January 20, 2008
The biggest player in speech recognition is just across Route 128 from the Burlington Mall. Parked in the firm's visitors lot is a Mercedes sedan that relies on its technology to allow the driver to control the navigation system and switch CDs with spoken commands.
The company is an avid acquirer of other speech-related companies - it scooped up seven last year alone. Its software enables users of Palm and BlackBerry smartphones to dictate text messages to their phones and send them without hunting for a single key.
When it comes to controlling a mobile phone, car stereo, desktop PC, or GPS device by voice alone, software from Nuance Communications Inc. is fast becoming the equivalent of Intel Inside. In terms of the breadth of its products, and the number of employees it has dedicated to speech recognition, Nuance looms over its bigger rivals, IBM Corp. and Microsoft Corp. But on the local and national tech scenes, Nuance is far from well-known.
"They've taken a fragmented industry and rolled it up into one company," says Daniel Ives, an analyst at Friedman, Billings, Ramsey & Co., alluding to CEO Paul Ricci's passion for picking up smaller speech recognition companies. (Friedman, Billings makes a market in Nuance's stock, but hasn't done any investment banking for the company.) "Speech recognition is still a green field opportunity, and I view them as the 800-pound gorilla in the space."
But if Nuance hopes to shoulder speech recognition into the mainstream - and get credit for doing so - it'll first have to grapple with marketing and technology challenges.
The technology challenge is simple: If you don't have the crystalline diction of a Julie Andrews, or you happen to be ensconced in a noisy room, trying to use a voice-driven system can be frazzling. We've all tried to explain to an airline's phone system that we're flying from Boston, not Austin, or enunciate the difference between a "v" and "b."
"People's response is invariably, 'I hate those things,' " says Walt Tetschner, president of the Concord research firm Voice Information Associates Inc. Such an attitude certainly doesn't compel consumers to invest in speech-driven technology of their own. In fact, local tech entrepreneur Paul English even created a website, GetHuman.com, that explains how to circumvent speech- and keypad-driven phone systems and talk to a human instead.
Less-than-seamless speech recognition is still holding back Nuance, and the industry as a whole. The new TomTom 920 GPS, introduced late last year, is the first sold in the United States that allows users to speak addresses rather than key them in; Nuance's software helps the $549 device decipher what you're saying.
But the unit couldn't seem to find or understand the Globe's address in Dorchester when Nuance vice president Peter Mahoney spoke to it. The Nuance software on Mahoney's desktop computer, Dragon NaturallySpeaking, also had trouble opening up a new e-mail when Mahoney asked nicely, using a wireless headset to talk to the PC.
Mahoney laughed it off, saying demos have a tendency to go awry when a reporter is in the room. He had better luck showing how his BlackBerry could pull up the latest news from the Web, simply in response to the command "Top Stories;" that service is called Nuance Voice Control.
Such demos make it clear that speech recognition feels like it is 95 percent of the way there. The last 5 percent will require a concerted campaign.
(Continued in next message) |
