Start-Up Has Software to Turn Your Computer Into a Conversation Piece
Just what you always wanted: a computer that talks back.
Computer scientists have struggled for years to develop a voice-recognition system that would respond to human speech patterns so flawlessly that all you have to do is say the right word. Then, like magic, your computer will fetch the information you need, or your television set will turn down the Rolling Stones video, or your cell phone will automatically dial your boss.
Progress has been made, but with current systems you have to sound a little like a Martian who has just learned how to speak English, picking only terms the computer will recognize, and avoiding similar-sounding words that have a very different meaning.
But an upstart San Diego company claims to have developed software that will allow the use of conversational English--or a bunch of other languages--to control almost any type of electronic device. And it has joined with two of the giants in the field, IBM and Philips, to spread the technology around the world.
“I wanted to write software that my mother could use,” said Dean Weber, founder and chief executive of One Voice Technologies in San Diego.
Weber says his mom would have no problem using IVAN, or Intelligent Voice Animated Navigator, which he plans to begin distributing for free within the next couple of months to anyone who needs to talk to their computer.
At 39, Weber is already a software veteran and the creator of IVAN, which works as an adjunct to IBM’s Via Voice and Philips’ Speech Processing voice-recognition systems. What Weber brings to the table is artificial-intelligence software that allows the computer to do such things as ask questions if it is confused, or suggest better search terms.
“We have developed the ability for natural language processing,” Weber said.
For example, if you ask your computer where to go to buy a car, it will probably ask what kind of car, whether you want a new one or a used one, and how much you want to pay. And Weber says the more you use the system, the more comfortable your computer will become with your speech patterns, learning what you really mean when you utter a sentence that to anyone else might seem incoherent.
The system has been demonstrated to potential clients--Web sites with big bucks that are expected to provide the revenue stream--and one firm has signed contracts with One Voice. Autobytel.com, which specializes in automotive e-commerce was the first to sign, and other companies are expected to be signed to deals soon.
That’s a critical milestone in the company’s effort to build credibility in an arena littered with disappointing voice-recognition products from electronic giants. You might want to compose a note to your spouse on IBM’s Via Voice, for example, but you’re not likely to create a literary masterpiece.
A major Web site would have to cough up $300,000 to $500,000, plus a monthly “maintenance fee,” for Weber’s engineers to install software that will allow the site to communicate verbally with its visitors.
Consumers will be able to get the software free, or buy an advanced version for about $40, by downloading from One Voice’s Web site (https://www.onevoicetech.com). The firm is following the business plan of such successful Web ventures as RealNetworks and MP3.com. If they can get millions of users around the world, Weber said, then major sites will sign up for the service.
It’s the old “build it and they will come” philosophy, but how many consumers are going to go to the trouble if very few Web sites are voice-compatible?
Initially, Weber said, One Voice will “spider” its software onto sites accessible from the company’s home page. That can be done without the active participation of the site itself. When you ask IVAN a question, it will check One Voice for sites, and channel the data through the piggybacked software so two-way chatter can begin.
And because IVAN is also a browser, it can direct consumers preferentially to One Voice’s paying customers. If you want to buy a book, for example, it would send you to Barnesandnoble.com, if that company has subscribed to the service, instead of Amazon.com, if Amazon isn’t with the program.
So if it succeeds, highly competitive sites will have no choice but to climb aboard, Weber says.
Weber hopes to have the top 100 Web sites available through One Voice by the time the software is released.
But there’s a catch. If the sites don’t ante up the dough in a reasonable period, and have the software installed directly on their site, they will be dropped.
So we’ve got a chicken and egg scenario here, and both components have to work for this thing to fly. If there aren’t enough users, sites won’t buy into it; and if there aren’t enough sites, why would anyone bother?
Which brings us to this: Does anyone really need this stuff? Is it all that hard to click a mouse?
The product is aimed primarily at household users, and people who may not be all that comfortable sitting in front of a computer. A mother who is getting her kids ready for school, for example, could just ask IVAN if it’s going to rain today.
It will take awhile for all this to play out, so it’s too soon to pass judgment on One Voice’s claimed “breakthrough.” But the system has been demonstrated across the country, leading publications such as Forbes to call the technology “impressive stuff.” Popular Science called it a potential “dream come true.”
By partnering with IBM and Philips, the company has opened the way to a wide range of applications. IBM, of course, is big in the computer market, and Philips is one of the world leaders in electronic consumer devices ranging from VCRs to flat screen television. Philips, incidentally, plans to offer the program in 16 languages.
Devices will be able to switch between voice and keyboard, allowing greater flexibility for users. That could come in handy.
*
Lee Dye can be reached at leedye@gci.net.
