BETA
This is a BETA experience. You may opt-out by clicking here

More From Forbes

Edit Story

Life After Siri: Nuance's Uphill Climb To Being Your Digital Assistant

This article is more than 10 years old.

This story appears in the Septemper 1, 2013 issue of Forbes. Subscribe

-----

In the gleaming Silicon Valley branch office of speech-recognition firm Nuance Communications , a small room has been made to look like a homey den.

Product executive Sean Brown holds up an Android smartphone to show off Nuance's next big thing: an invisible virtual assistant that was more than a year in the making, called Wintermute.

"How's Paul Pierce playing?" he asks the phone about one of the NBA's top scorers. A woman's voice responds in computerized staccato: "Paul Pierce has 13 points and three assists in today's game."

Wintermute has stored all of Brown's queries on Nuance's servers, so when he later sits down on the mock living-room couch and tells his Wintermute-enabled TV to "throw on the game," it checks his profile and opts for the Celtics.

This is how Nuance intends to change the way we interact with machines, collecting data about us in a cloud-based brain to both understand speech commands and better anticipate our needs from machine to machine.

You've talked, slowly and loudly, to Nuance before. Known for its Dragon Dictation products, its speech technology is also behind the iPhone's virtual assistant Siri, Samsung's S-Voice, automated call centers and virtual online assistants for airlines and top-tier banks, and it's on the majority of the world's smartphones.

The company earned $556 million in 2012 on revenue of $1.7 billion, and net profit has grown at a 39% compound rate since 2007. However, Nuance recently cut its profit and revenue forecast for 2013, citing contract delays with unnamed mobile customers. While competition intensifies, Nuance is determined to maintain its license fees. Nuance has also been putting a "significant" chunk of its $275 million annual R&D budget and many of its 400 research staffers on Wintermute.

Trouble is, Google and Apple are following the same evolutionary path for speech technology. Apple recently set up a research center in Boston (15 miles from Nuance's main offices in Burlington, Mass.), where former speech engineers from Nuance are working on a Nuance-free version of Siri.

Potentially, it could boot Nuance off the iPhone the same way it booted Google Maps as the default map service. That would be a major blow to Nuance's shares, already down 14% since the beginning of the year and trading at a relatively expensive 41 times 2012 earnings.

Google's speech-plus-artificial-intelligence technology, best seen through Google Now as a vocal way to search the Web, may be more promising. Nuance's speech technology is based on statistical inference methods that look at phonemes (syllable sounds) and context to recognize words.

Google, by contrast, is moving toward so-called deep learning technology pioneered by artificial intelligence hero Geoffrey Hinton, whom Google hired in early 2013. Hinton's approach promises superior results, and, even worse for Nuance, Google gives its voice-recognition software to Android app developers for free, while Nuance charges a licensing fee. Google has also poached Nuance staff, including its cofounder Mike Cohen to become its speech technology director in 2004. The company is meanwhile moving towards greater personalization of its search results.

Nuance's CEO Paul Ricci, in the job for 13 years, is prepped for the coming battle over the voice-assisted Web. Ricci is nothing if not a street fighter, notorious in the Valley for closing the deals he wants. Of the 60 acquisitions during his tenure, several followed the threat of a patent lawsuit. (Nuance has filed eight, officially.)

"I don't accept the characterization," says Ricci, who sits with arms and legs crossed in a spartan conference room at Nuance's offices in Sunnyvale, Calif. "There are a lot of satisfied sellers out there who built a lot of wealth through the acquisitions they do with us." He adds later: "The only thing you can do when running a company is to focus on creating long-term value."

If Wintermute isn't the bridge to the future that Nuance hopes, Ricci's choices are going to get very narrow very soon. "[Nuance] is at a stalemate," says one investor. "They should have been bought by now."

Apple was rumored to have made overtures toward Nuance in May 2011, which were rebuffed by Ricci. Even Google considered buying the $6 billion market cap Nuance a few years ago, a source involved in Google's internal discussions says, before deciding to develop its own technology.

Carl Icahn, renowned for taking large stakes in companies like Dell and goading them to improve or sell, has amassed 16% of Nuance. He now claims to have taken a large position in Apple too, sending its stock 2.4% higher on Wednesday. The billionaire wouldn't comment, but analysts expect him to start demanding this summer that Nuance sell its small but historic imaging division, or split Ricci's iron grip on the CEO and chairman roles, or sell outright to a neutral vendor like IBM.

It could be a while before Wintermute generates much revenue. The living room demo given to FORBES in June used the same phrases as its January 2013 demo at the Consumer Electronics Show in Las Vegas. "It looks like a science project right now," says Nandan Amladi of Deutsche Bank. "It's unclear how it would make money."

Nuance depends on the health care industry for half of its revenue. Thousands of doctors and hospital workers use its software to transcribe notes. Mobile is only a quarter of sales, and changes in how Nuance charges handset vendors for its software have limited its growth there. While global smartphone sales grew 43% in 2012, according to Gartner, pro forma revenue at Nuance's mobile division grew 29%.

For Wintermute to be a real success, Nuance has to cut a passel of complex licensing deals with potential rivals to enable the technology to follow its users around from home to car to office. "We have to work very closely with the television and smartphone manufacturers," says Ricci, so that "the technology will disappear into the system."

He expects that to take a few years. Won't it be hard to convince vendors like Samsung and Internet giants like eBay to open their systems to Nuance so it can cross-reference their data? Nope, says Nuance's chief creative officer, Gary Clayton. "We're Switzerland," he says. "Neutral. We play with everybody. Everybody is going to get on board with this."

From the outset eBay has been dubious. "We do not share our data externally," a spokesperson said when asked if the e-commerce giant would consider sharing data with something like Wintermute.

Daniel Gross, cofounder of digital assistant app Cue, which uses Siri to take the occasional speech command from users, says the world of intelligent personal assistants is extremely difficult and uncharted. He's worked on the problem for three years with 14 engineers.

"The largest problem I have in my business is it's very hard to focus," says Gross. "You can come up with a brilliant road map to selling this stuff as 'white-label software' or an even more amazing road map for a consumer product that can change the world. But the less you do the more likely you are to succeed."

Just improving Nuance's original speech technology will take some time. Ricci thinks it'll be another 10 to 20 years before the act of talking to computers is widespread and the technology is good enough to anticipate our intent. "How annoying will that be?" he jokes.

Ricci in person is blithe and unhurried by the prospect, lingering after his interview to reflect on the current state of Silicon ?Valley, where he has lived much of his adult life. He worries the place has become "more frenzied" in the last five to ten years. Considering what Google, Apple and Cue are up to, investors might want him to hurry up if Wintermute is going to create that "long-term" value Ricci's looking for.

Follow me on Twitter: @Parmy