Natick, MA-based Vivox said today that it’s opening its new “Vivox Web Voice for Facebook” service to all Facebook members. The application—which allows Facebook users to set up free voice chat rooms and invite their friends to participate from within Facebook—is one of the first creations of Vivox Labs, a new R&D arm of the company, and had been in closed beta testing for the last few months. Vivox, which we profiled last month, is known mainly as a provider of voice communication services for massive virtual worlds and game worlds such as Second Life and Eve Online.

Comments | Reprints | Share:        

Vlingo, the Cambridge, MA-based startup that makes a suite of speech-to-text applications used by millions of iPhone, BlackBerry, and Nokia mobile device owners, is about to get a brain transplant of sorts. It said today that it will largely abandon a core speech-recognition engine developed by IBM and maintained by Nuance Communications in favor of a system from AT&T Labs in New Jersey.

As part of the shift, says Vlingo CEO Dave Grannan, Vlingo and AT&T have agreed to a long-term strategic alliance. Vlingo’s speech scientists will be able to modify and improve the source code for the AT&T technology, called Watson, while AT&T will take a minority ownership stake in Vlingo. All of Vlingo’s applications will be running on top of the AT&T speech-recognition system by the first quarter of 2010, Grannan says.

Vlingo’s own speech scientists have developed software that exploits information collected from users—the way a Bostonian’s pronunciation of a dictated phrase like “I parked my car” might differ from a New Yorker’s, for example—to build statistical models that help improve speech-recogition accuracy over time. These models provide supplemental input that helps to guide a core speech-recognition engine as it transforms speech sounds into text. Vlingo didn’t build its own core engine—it has long licensed that part of its system from IBM.

The switch from IBM’s engine to AT&T’s is a “best of all worlds” situation for Vlingo, in Grannan’s words. For one thing, he says, the Watson technology simply works better than the IBM recognizer. “Watson is superior on speed and base-level accuracy,” he says. Once the transition is complete, users of Vlingo’s iPhone, BlackBerry, and Nokia apps should notice fewer wrong guesses in the transcriptions of their utterances. Grannan says they’ll also see a few new features, such as automatic punctuation, that Vlingo can now add because it will be able to tinker with Watson’s innards.

But just as important, the switch will help Vlingo disentangle itself from its strained relationship with Nuance.

Burlington, MA-based Nuance (NASDAQ: NUAN) is one of the Boston area’s biggest high-tech firms, and it is the world’s largest specialized provider of speech-related technologies. It offers software for mobile speech recognition that competes directly with Vlingo’s. In June 2008, after losing out to Vlingo on a contract to supply Yahoo with speech-recognition technology for its oneSearch service, Nuance hit Vlingo with a lawsuit alleging that …Next Page »

Comments | Reprints | Share:        

For a long time, Second Life was stuck in the cyber equivalent of the silent-movie era: people communicated by typing, and their words showed up in little thought bubbles above their avatars’ heads. All of that changed drastically around 2007, when Linden Lab, the company behind Second Life, hired an obscure outfit called Vivox to equip its 3-D virtual world with a voice communication system. Now any Second Life citizen who has a headset connected to their computer can simply speak, and everyone whose avatar is standing nearby will hear them in living stereo.

For the Gloria Swansons of Second Life, like myself, the changeover from typing to talking was a bit traumatic—and indeed, 20 percent of Second Life citizens still abstain from voice communication. But the other 80 percent gab for a billion minutes every month, which is a rather convincing demonstration that most people inside 3-D computer environments prefer talking to texting.

And now Vivox, a four-year-old startup based in Natick, MA, is about to introduce its technology to three new communities that could vastly increase its user base. The company announced this morning that it has formed a partnership with Redwood City, CA-based Electronic Arts (NASDAQ: ERTS), the world’s largest entertainment software company, to add its voice services to several online EA games. First up is Command & Conquer 4, a continuation of EA’s hugely popular real-time strategy game that’s expected to launch early next year.

At the same time, Vivox is announcing the launch of Vivox Labs, an incubator-within-a-startup where the company is trying out different ways of delivering its voice services over the Web. And the first two Vivox Labs experiments are aimed at big targets: Facebook, where the lab’s “Vivox Web Voice for Facebook” application will allow members to invite their friends to instant Web voice conferences; and World of Warcraft subscribers, who will be able to use a new Vivox-powered website called Puggable to assemble teams of players for in-world campaigns. Both the Facebook and Puggable applications are in private beta testing and are expected to go public by January.

“We started the company about four years ago with the goal of making voice a seamless, natural part of every online experience,” Vivox co-founder and CEO Rob Seaver told me when I visited the company last week. “Our view at the time was that more and more human interaction would take place online, and the richest form of communication we have is talking to each other. So we thought there would be an opportunity to turn the Web from this silent, barren place into one filled with the warm sounds of human voices.”

That’s exactly what could happen if even more gaming, virtual-world, and social networking communities turn to Vivox’s services. Not bad for a company that started out as a wacky idea from Jeff Pulver, the founder of the company that became Internet phone service provider Vonage.

You’ve probably heard of Voice over Internet Protocol, or VoIP; it’s the technology behind Vonage and Skype, and the one that has turned the telecom industry upside down by transforming phone calls into digital data packets and routing them over the open Internet. Vivox’s system works on similar principles, except that …Next Page »

Comments | Reprints | Share:        

[Updated 9/3/09, 3:00 pm. See below.] Bellevue, WA-based VoiceBox Technologies, a developer of speech recognition systems for use in cars and mobile applications, has raised about $13 million from corporate investors in Asia over the past year. The investors include AutoNavi, Inventec, MiTAC, and the Morningside investment fund.

[An earlier version of this story cited a regulatory filing stating that VoiceBox had raised $7.4 million in equity and options out of an $18.6 million offering, and said the investors were not disclosed---Eds.]

Chief strategy officer Victor Melfi of VoiceBox says the company has raised a total of about $21 million to date, including investments from friends and family, and InfoSpace. He adds that VoiceBox is now looking to raise an additional $15 million from institutional investors, for which it has signed on Seattle investment bank Cascadia Capital. Melfi says VoiceBox is sensitive to customers in Europe and Asia—particularly China—and that it is developing technology for nine different languages. [This paragraph was added at 3:00 pm after speaking with Melfi---Eds.]

VoiceBox is developing what it calls “conversational voice search” software that lets you search, navigate, and discover content and services using natural spoken language. An example would be telling your car to give you directions to a particular location, pick a song to play, and adjust the temperature—all while you’re driving. Or telling your smartphone to search for a stock quote or other information online while you’re on the go.

Technologically, it’s a very hard problem. That’s because of ambient noise, differences between people’s accents and the way they make requests, and, fundamentally, the challenge of correctly understanding the meaning of what they’re asking for. Voicebox has partnerships with a number of companies including IBM, Toyota, and XM Satellite Radio to refine its software. The company also has an iPhone app for voice dialing.

VoiceBox was incorporated in 2001, and is led by its co-founder, chairman, and CEO Mike Kennewick, a former manager at Digital Equipment Corporation and then Microsoft. Kennewick previously founded Saros, a document management software company that was bought by FileNet in 1996. As of January 2008, VoiceBox had not taken any venture funding, but was considering taking a round, according to VentureBeat.

Comments | Reprints | Share:        

Nuance Communications (NASDAQ: NUAN), the Burlington, MA-based voice technology juggernaut that has already absorbed most of its East Coast competitors, reached west today, announcing that it has acquired Seattle-based Jott.

Jott, founded in 2006 by ex-Microsoft employees, started out as a free voice-to-text service that allowed users to record messages via telephone that were then transcribed into e-mails. Over time, the company transitioned to a paid business model, and expanded the capabilities of its service to let users create text messages, blog posts, appointments, reminders, and notes. The service has proved popular among mobile professionals, gaining hundreds of thousands of users, according to the company.

Nuance’s acquisition of Jott gives it a credible product in the area of phone-based voice-to-text services, where other companies such as Google, with its Google Voice service, and UK-based Spinvox have begun to encroach.

“Jott’s voice-to-text offerings have experienced a groundswell of adoption and positive industry recognition since the company’s inception, and we’re thrilled about the opportunity to expand our market reach and our voice services portfolio,” Nuance senior vice president Michael Thompson said in an announcement. “Together we will deliver a range of new services to our mobile operator and enterprise customers.”

Nuance isn’t saying how much it shelled out for Jott. The startup was funded by Bain Capital Ventures, Draper Richards, Ackerley Partners, and UK-based Atomico Investments; its last publicly divulged funding round, in 2007, amounted to $5.4 million. Jott may have needed a larger partner like Nuance in order to compete in its sector, given that competitor Spinvox, with some $200 million in venture cash, had far outpaced it in fundraising efforts.

Nuance and Jott said that Jott’s services, including Jott Assistant, Jott Voicemail, and Jott for Salesforce, will keep working as usual, with no interruptions in service. But as a result of the acquisition, Jott-like capabilities may come to many more consumers—Nuance says it plans to package Jott Assistant as part of the voice services it provides to wireless operators.

“Our combined expertise will bring innovative and differentiated voice services to a variety of markets with tremendous scale,” Jott co-founder John Pollard said in a statement.

One area where Nuance’s technology may help Jott is in automated speech-to-text software. While the basic user interface that Jott users encounter when they call the service is driven by speech recognition software, users’ recordings are actually transcribed by humans working in large processing centers. Replacing those humans with advanced speech-to-text software, similar to Nuance’s Dragon Naturally Speaking line of dictation software, would be an obvious way to make Jott’s service more efficient and scalable.

Comments (2) | Reprints | Share:        

The human voice is making a comeback. For a while, it looked like e-mail, instant messaging, blogs, RSS, and all of the Internet’s other texty goodness might permanently eclipse the old-fashioned phone call and other voice-driven forms of communication. Even the spread of cell phones hasn’t halted the tide of text—more than a third of mobile phone owners use their phones primarily to send SMS text messages rather than making actual calls, according to research from Cambridge, MA-based Vlingo.

But a stream of new mobile-device applications designed for voice input might be restoring the balance. This month I’m excited about two examples in particular: the new Voice Memo app that showed up with Apple’s iPhone 3.0 operating system, and AudioBoo, a nifty audio recording app for the iPhone with a surprising origin: Channel 4, Britain’s publicly funded alternative television network. Along with several other programs, these apps are turning the iPhone into a handy platform for “personal podcasting,” an emerging genre of amateur digital publishing that’s as convenient and spontaneous as Twitter but, because it’s actually a person talking, feels more human.

[You can click here or skip to page 3 to hear an AudioBoo version of this article.]

No apologies, by the way, to non-iPhone owners. With iPhone 3G now priced at $99 and the 3GS starting at $199, there are fewer and fewer excuses for not trying out Apple’s marvelously powerful uber-gadget.

First, a word about Voice Memo on the iPhone. Many mobile phones come with a voice recording function these days, so it wasn’t a surprise to see Apple add one when it updated the iPhone operating system last month. It’s fairly basic: it lets you make new audio notes and review your old notes, all of which get copied to your iTunes library whenever you sync. There’s also a basic editing feature that lets you trim a voice memo by lopping time off the beginning or the end. Best of all, there’s a “share” button that lets you send out copies of voice memos via e-mail.

I really like the sharing feature, which is great for sending people quick voice messages, and has two advantages over conventional voicemail. First, the sound quality is far superior. Voice memos are monaural, but they don’t get compressed the way your voice does when you’re leaving a message for someone over a cellular voice network (compression that’s redoubled if the recipient is retrieving their voicemail from their own cell phone). Second, e-mailing a voice memo is a non-sneaky substitute for voicemail for those times when you want to leave a voice message but you don’t want to risk actually talking to the person. (Slydial offers a similar capability by connecting you directly to someone’s voicemail—but it’s not foolproof, as it sometimes makes their phone ring anyway.)

In a pinch, you can also use the iPhone Voice Memo app to record audio for publication on the Web. It clearly wasn’t designed for this purpose, as the app records memos using the relatively voluminous .m4a audio format, and doesn’t allow you to transfer memos over a certain size by e-mail. (I’m not sure what the limit is, but I was unable to send a 5-minute, 12-megabyte file.) Also, it buries the synchronized copies of your voice memos deep in the iTunes folder of your computer, where it’s difficult to find them. But as a test, I located one memo—a few thoughts that I recorded on a drizzly afternoon at the Japanese Garden at Boston’s Museum of Fine Arts—and used iTunes to convert it from .m4a to the more compact .mp3 format, which made it small enough to post on my personal blog at Tumblr.

But if you really want to use your iPhone as a tool for audio publishing, there are much simpler options.

For a long time, my favorite iPhone audio recording app was iTalk, the coolest feature of which is that …Next Page »

Comments | Reprints | Share:        

Back in March, I wrote a column about Google Voice, the reincarnated version of a voicemail unification service that Google acquired from Grand Central back in 2007. The free service gives you a single phone number for life that isn’t tied to any particular land line or cellular device—instead, calls ring through to whichever phones you specify. Voicemails get stored online and (if you want) transcribed into text e-mails. In my column, I called Google Voice “the end of the phone as we know it,” since a Google Voice number resembles an e-mail address more than an old-fashioned phone line. It goes with you everywhere, can be managed entirely through the Web, and literally turns your voicemails into e-mails.

Google Voice was, and still is, in a private, invitation-only, beta testing phase. When I checked with Google early last week, employees were still saying the service will be available to the general public “soon”—which is the same thing they were saying back in March. But the big day may be approaching. While rumors circulating last week about the service’s impending launch turned out to be false, Google Voice product manager Craig Walker did state, via his public Twitter stream, that “We’re cranking 24/7 to get there.”

In conjunction with my March column, Google kindly provided 100 Google Voice beta invitations for Xconomy readers—and not surprisingly, all of the invitations were snapped up within an hour after we publicized the offer. So in anticipation of the public launch of Google Voice, I decided to ping the lucky 100 beta account winners last Friday to find out how the service has been working out for them, and whether they’d recommend it to others.

The readers who’ve written back so far have been lavish with their praise—at least, the ones who have actually been using their accounts. Several have admitted that they never signed up, or that they signed up but found that Google Voice wasn’t what they expected, or that, as one reader put it, “I would have liked to [use it] but then work (life?) got in the way.” More about the potentially high barriers to adoption below.

Readers who’ve used Google Voice seem to like the way it lets them give out a single phone number to everyone, rather than separate office, home, and cell numbers. Several readers said they like the (somewhat sneaky) feature that lets users listen to callers as they’re leaving a voicemail, and break in if they want to talk to that person directly. And if there’s one feature everyone loves, it’s the automatic transcription of voicemail messages into e-mails—a Google invention that wasn’t part of the original Grand Central service. While Google’s speech-to-text technology is far from perfect, readers say it’s good enough to get the gist of a message across, and that it saves them from the universally dreaded task of actually listening to all their voicemail. (You can browse readers’ detailed comments below.) Xconomy’s CEO and editor-in-chief, Bob Buderi, has been using Google Voice since March, and he also cites voicemail transcription as his favorite feature.

Readers report surprisingly few technical glitches or other difficulties using Google Voice. The problems they do cite tend to be ones that are baked into the service’s design. Most people said it’s too much trouble to make outgoing calls through Google Voice, since users must either call their own Google Voice number first, or go to the Google Voice website. Which leads to another frequent complaint—the caller ID problem. Unless you place all your outgoing calls through Google Voice, then the people you call will see the number of the device you’re calling from, rather than your Google Voice number. That means you have to train everyone not to store your device’s number in their contact list, but to call you back on your Google Voice number instead. That’s plain confusing for everyone.

Asked to say whether they’d recommend Google Voice to a friend or a family member, quite a few readers said “Yes, but…” The “but” was that they’d only recommend it to people who are technically adept—”power users,” in one reader’s phrase. As another reader put it: “The person who is going to use [it] needs to be a bit of a techie (not super technical, but my wife who is not technical would get lost in the concept)…[there are] lots of configuration options which I enjoyed learning and setting up.”

How much have Xconomy readers actually used their Google Voice accounts, in the end? That varies. Some say they’ve made their Google Voice number into their main phone number, and that they use the service extensively every day. Others say …Next Page »

Comments (12) | Reprints | Share:        

It was a relatively busy week for deals in the Northwest, with plenty of action in software, biotech, and alternative energy.

—Seattle and San Francisco-based Twilio, a startup that provides cloud-based tools for building voice applications over the phone, raised its first institutional round of funding from Founders Fund and computing pioneer Mitchell Kapor. The amount was not disclosed. Twilio co-founder Jeff Lawson talked with me about the deal and how it will help the company expand its services.

—Seattle’s Principle Power, a wind energy startup, is in the process of raising $20 million to develop the world’s first floating wind farm in the deep waters off the coasts of Oregon and Portugal. Principle Power’s CEO, Alla Weinstein, wouldn’t say who the company’s strategic investors are, but she did say that no VC firms or Seattle investors are involved. The deal is expected to close in the second quarter of this year.

—Luke broke the exclusive story of Merck’s Stephen Friend raising $5 million in anonymous donations to pursue the vision of open-source drug development at a new Seattle nonprofit called Sage. The idea is to provide an open database of patients’ genomic profiles that researchers, doctors, and drug companies can access in order to make better drugs.

—Bellevue, WA-based Ignition Partners led a $10 million investment in Zenprise, a mobile-management software startup in Fremont, CA. Existing investors Bay Partners, Mayfield, and Shasta Ventures also participated in the round. Zenprise makes automated software to help businesses fix IT problems with smartphones.

—Seattle-based Arzeda, a University of Washington startup that designs custom-built enzymes, has secured commitments from OVP Venture Partners and WRF Capital to anchor its $12 million …Next Page »

Comments | Reprints | Share:        

Bedford, MA-based vCentrix, which hosts voice-over-Internet services for businesses, is becoming part of Momentum, a provider of white-label digital voice services based in Birmingham, AL, according to an announcement from Momentum today. “In such a fragmented market, both companies feel that hosted voice applications are entering a period of consolidation, and Momentum is well-equipped to be a leader in this coming shakeout,” Momentum CEO Alan Creighton said in the announcement. Financial terms of the deal were not disclosed.

Comments | Reprints | Share:        

Marlborough, MA-based ThinkEngine Networks, a telecommunications company whose media servers unify traditional circuit-switched and Internet-based voice signals, has filed for Chapter 7 bankruptcy protection in U.S. bankruptcy court in Worcester, MA, according to reports today in the Boston Business Journal and the Boston Herald. The American Stock Exchange delisted the company last March due to underperformance.

Comments | Reprints | Share:        

Back in 2006, Cambridge, MA-based Art Technology Group (NASDAQ: ARTG) spent north of $48 million in cash and stock to acquire eStara, a maker of software that allows Web surfers to open voice-over-Internet connections with sales agents by clicking on Internet ads. Yesterday the company announced that it’s extending eStara’s capabilities to Flash-based video ads. “While consumers are more comfortable watching video clips online now than ever before, direct-response advertisers—many of whom are small, local businesses—have struggled to find a video advertising model that is cost-effective and measurable,” said Ruth Habbe, vice president of ATG’s eStara Business Group, said in a statement. “eStara Video Connect solves this problem by making video ads actionable, and by giving media companies a rich tool for tracking consumer actions and leads generated from each and every video ad.”

Comments (1) | Reprints | Share:        

Updated Nov. 5, 4 pm PST: See below for details on funding

TelCentris, a new San Diego company, says it has released a beta version of its free consumer service called VoxOx, which combines Voice-over-Internet telephony and other types of communications into a single screen on a computer desktop. The company says its universal communicator service is the first in a coming generation of consumer software products that combine voice, video, instant messaging, text, e-mail, fax, and social networks into a single screen on a desktop.

While competitors such as Skype offer VoIP and rival Digsby aggregates messaging, social networking and e-mail onto a single screen, the company says that VoxOx combines all these features into a single product—with an iPhone-like graphical user interface. The name VoxOx is a play on “voice over X,” meaning the system can send voice over any type of network.

TelCentris says it has targeted communications-overloaded GenXers and “Millenials,” those born between 1980 and 2000. That’s a different tack than rivals that have been trying to get corporate customers to commit to their hardware and software for Voice-over-Internet telephony. It might be an easier sell, but Voice-over-Internet telephony is rapidly becoming a commodity, meaning the competition is getting fiercer and players are looking for ways to break out of the pack.

TelCentris was founded by CEO Bryan Hertz, his brother, Kevin, who is chief technical officer, and father Bob, chief information officer. The CFO is Michael Faught, who has 25 years experience in finance, management, and technology commercialization, according to the company’s Web site.

A spokeswoman for TelCentris says the founders bankrolled the company themselves, with some additional angel investment. She added that TelCentris has been generating revenue by powering small telephone companies with its Unified Communications Service Delivery Platform, as well as providing Hosted PBX service to dozens of small-to-medium businesses.

Comments (1) | Reprints | Share:        

Cambridge, MA, startup Vlingo today launched a drastically improved version of its voice command and speech-to-text interface for RIM Blackberry smartphones, first released in June. The free vlingo 2.0 software, which works on Blackberry Pearl, Curve, and 8800 series phones, not only allows users to compose text messages and e-mails by speaking into their phones, but also lets them create status updates for Facebook and Twitter and launch built-in Blackberry applications such as the address book, the camera, the memo pad, or the calendar and third-party applications such as Google Maps, the Opera browser, or Viigo.

Comments (1) | Reprints | Share:        

Jingle Networks, the company behind  the advertising-supported directory assistance service 1-800-FREE411, is returning to its Massachusetts roots. The venture-funded startup, which was founded in Burlington, MA, in 2005, announced yesterday that it’s moving its corporate headquarters to Bedford, MA, after several years in Menlo Park, CA.

The move coincides with a change in leadership at the company. Former CEO George Garrick has ceded the top spot in the company to founder and former CTO Scott Kliger, a veteran of rich-media Internet advertising firm Narrative Communications and Excite@Home. Garrick, who lives in the San Francisco Bay Area and will remain on Jingle’s board, said in a statement that the company needs a CEO who is based on the East Coast, where most of its technical team is located.

Along with its change of headquarters, Jingle announced that the 1-800-FREE411 service reached “per-call profitability” in the second quarter of this year. It’s not clear exactly what that means, or how close the company is to actual profitability. But it appears to be an indication that Jingle is earning more for the 10-second audio advertisements that each caller must listen to before getting their listing than it costs to host a call.

Jingle said that in the next quarter, it would pass the 500-million-call mark and serve its 1 billionth advertisement. Since its creation, the company has collected almost $75 million in funding from the likes of Hearst and Goldman Sachs; investors seem excited by its business model, which is to undermine the existing providers of 411 service (who charge about $1.25 per call) and to set up a “voice ad network” that could deliver audio ads to any company that regularly keeps large numbers of callers on hold.

In that respect, Jingle Networks is competing directly with another Massachusetts company, North Adams-based VoodooVox, which is building what it calls an “in-call network” that functions similarly to an online advertising network. In place of banner ads on a website, the company inserts audio spots into the “call streams” of its “publishers,” meaning companies with high call volumes. “That’s why we say phone calls are the new page views,” the company says on its website. The response rate for in-call audio ads—the analog of the click-through rate on Web ads—is 12 percent, far higher than the rate for most online ads, according to VoodooVox.

Comments (1) | Reprints | Share:        

Cambridge, MA, startup Vlingo announced back in February that it had built its speech recognition software into Yahoo’s oneSearch mobile search portal. Thanks to that collaboration, users of Blackberry smartphones could speak search terms into their devices rather than having to type them using the keypad. It worked great in our tests. And now Vlingo’s taking the next step, offering free software that lets Blackerry owners use voice commands to do virtually anything they want with their phones, such as launching applications, dictating e-mails and text messages, searching the Web, and, of course, making calls.

The application, which is available for download here starting today, is designed only for the latest Blackberry models, including Curve, Pearl, and 8800-series devices. But Vlingo CEO Dave Grannan says the company is working on versions of the software for other smartphones, including Windows Mobile devices, the Apple iPhone, and eventually devices running Google’s Android mobile operating system.

“What our company is really about is solving both the discoverability and usability problems,” says Grannan. What he means is that the tiny screens and navigation buttons on mobile phones sometimes make it tricky to find and open the application you want, such as the address book or the text-messaging interface. Then, because many phones lack a full QWERTY keyboard, you have to double- or triple-type inside those applications to get anything done.

But using the new Vlingo interface, you can open an application just by holding down your Blackberry’s side button and speaking a command such as “Open address book.” Sending a text message is as easy as holding down the button and saying something like “Send message to Brian: Remember to pick up batteries on your way home.” And sending an e-mail is only slightly more complicated; you have to include some pointers for the software, thus: “Send e-mail to Brian, Subject: Batteries, Message: Be sure to pick up some fresh batteries at the store.”

You can’t yet use voice commands to do things like creating new address book entries or calendar appointments; Vlingo would have to work directly with the developers of those applications to make that possible. But that’s exactly what Grannan hopes will happen: Vlingo’s business rationale for giving away the Blackberry voice application is that giving smartphone users a taste of its convenience will inspire the companies that make those individual mobile applications to license Vlingo’s speech-recognition system for their own applications. “This is one step better than what we have done already with Yahoo oneSearch,” Grannan says. “But the Holy Grail is doing everything with voice.”

Vlingo is moving full-speed ahead with its product development plans in spite of the patent infringement lawsuit filed against the firm last week by Burlington, MA-based speech applications company Nuance Communications (NASDAQ: NUAN). Nuance says Vlingo’s technology for improving speech-to-text transcription over time—technology used in both the Yahoo oneSearch interface and the new Blackberry software—infringes on one of its own patents. Vlingo, for its part, says it’s using completely different technology. “It’s pretty black and white—the lawsuit is groundless, on the merits,” says Grannan. “They’re trying to bleed us to death and make our big customers nervous. What works in our favor is that Nuance has a history of doing this, and everyone knows it. It’s a business strategy for them.”

Vlingo loaned me a Blackberry Pearl so that I could play around with the voice application. For the most part, it works as advertised. And when it does run into difficulties getting the right transcriptions, it learns pretty quickly. In my case, the device had the ill fortune to be tested on some uncommon names and words. It had a tough time with “Wade”—when I commanded the phone to “call Wade” (i.e. my personal cell phone) it kept trying to call Grannan, since “Dave” sounds a little bit like “Wade” and his name was already programmed into the address book. After about four tries, with me correcting it each time, it understood.

The system does get a lot of messages right on the first try. For example, it had no problem with “I’m running out of cereal.” When I said “E-mail Wade, Subject: Battlestar Galactica, Message: I wish the next season wasn’t so far away,” it at least got the subject right. (The fact that it knows exactly how to transcribe “Battlestar Galactica” is undoubtedly a measure of how many sci-fi geeks like myself have been using Vlingo’s voice applications.) Unfortunately, the message itself came out a bit garbled—as the software was perhaps swayed by the earlier training I’d given it on my own name: “I wish text season wade so far away.”

And perhaps it’s mean of me to harp on it, but Vlingo still has a lot of trouble with one of our favorite words around here, “Xconomy.” After about six tries and corrections, it continued to make wretched guesses like “x connie” and “ex con me” and “text tommy.” So, Xconomy readers, a plea: Since the Vlingo system learns by aggregating audio samples and corrections from many users, we need you to help train it to recognize our name. If all of you Blackberry owners out there wouldn’t mind downloading the Vlingo software and spending 10 minutes speaking the word “Xconomy” into your device, then correcting the transcription as necessary, we’d be much obliged.

Comments (1) | Reprints | Share:        

Nuance Communications (NASDAQ: NUAN), the Burlington, MA-based provider of speech-related software technologies and services, including the Dictaphone system used by hundreds of thousands of doctors to dictate clinical notes over the phone, said yesterday that it will purchase Needham, MA-based eScription. The company makes automated speech-to-text software that turns dictated medical notes into rough first drafts, which are then polished by professional medical transcriptionists. Nuance said it will pay $363 million for eScription—$340 million in cash and $23 million in Nuance common stock.

Comments | Reprints | Share:        

Would Henry David Thoreau have been a blogger? I think he might have been. And if he’d had a cell phone and a voice blogging service like Utterz—launched today by Maynard, MA, startup RPM Communications—he could have blogged all of Walden right from his little cabin in the woods.

Thoreau is on my brain because I’m finishing this column on my MacBook from the shore of Walden Pond, a hundred yards from the site of said cabin. I can hear the MBTA commuter train rumbling past in the distance. The same rail line was active when Thoreau lived here from 1845 to 1847. Obviously, there weren’t any MacBooks, electrical plugs, or batteries around here in Thoreau’s day. But imagining that he’d had a way to record his own spoken thoughts and get them out to the world instantaneously (at www.walden.org, of course), I bet he would have loved it.

That’s what Utterz does, and after a few days of pre-launch testing I’m already taking a liking to it. As someone who blogs both professionally and personally, I know that I’m sometimes moved to sit down with my laptop and write something serious and contemplative. Other times I’d just like to post something short and spontaneous, without even bothering to find a computer or an Internet connection where I can log into my blogging service (Wordpress for Xconomy, TypePad for my personal blog). With Utterz, I can simply call up 712-432-MOOO (432-6666) and talk for as long as I want. The service will automatically post the audio clip to the blog or social-networking site of my choice, in the form of a widget that can be embedded as a sidebar or directly in a post. (Click here for a widget with some sample Utterz I recorded here at Walden Woods.)

While the voice blogging feature is the centerpiece of Utterz, it can also handle text, video, and photos. When you register with Utterz you program it with your cell phone number and e-mail address. Then if you mail a photo from your phone to go@utterz.com, it will automatically match it with your most recent audio clip and display them together. All of this works seamlessly. A phone call or e-mail is literally all that’s needed to broadcast what you’re thinking or seeing to the world.

“We’re leveraging your cell phone to let you create, access, and manage content on the Web and share it with friends and with your social networks from anywhere, at any time, regardless of your carrier or your type of phone,” says Michael Bayer, CEO of RPM. Today’s a big day for Bayer’s company, which also announced that it has closed a $4 million series A funding round, led by Menlo Park, CA-based Morgenthaler Ventures.

The company makes money by collecting a small fee from cellular carriers each time a customer calls 712-432-MOOO. “It’s good for the carriers because it’s minutes used,” says Morgenthaler partner Greg Blonder. “And if you wind up using more minutes than you would normally, you might buy the next plan up.”

Will Utterz catch on? It’s hard to say. Tools for mobile blogging, or “moblogging,” have been around for years. TypePad and other blogging platforms, for example, offer users the ability to post directly to their blogs via e-mail. Audioblogger, launched in 2003, was one of the first services that let you call a phone number and make an MP3 recording that would then be posted to your blog, where visitors could listen to your recording online or download the file. Audioblogger was discontinued in 2006, but other sites such as Gabcast, Hipcast, Gcast, and Jott offer similar capabilities.

Yet most blogs are still 99 percent text, perhaps with a few pictures thrown in. Given that most everyone who blogs probably also owns a perfectly good moblogging device (i.e., a cell phone), there are only two obvious explanations for why audio moblogging hasn’t really taken off as a sub-genre.

One is that people view blogs as a primarily text-based medium, and that they don’t have a burning desire to create or consume audio posts. This explanation could turn out to be the right one. Unfortunately, it’s impossible to tell without first eliminating the other potential explanation, which is that the existing tools for getting an audio clip onto a blog are too clumsy and complex.

Utterz should provide a good test of the second explanation. It’s so easy to use that if it doesn’t take off, that will be a pretty good indication that the companies offering audio blogging tools have miscalculated, and that there just isn’t much of a demand for this particular service. But I’m betting that it will succeed, at least among the subset of people who keep or read blogs but also enjoy more personal and immediate forms of communication such as instant messaging, SMS text messaging, Twitter, and picture-mail on cellular carriers like Verizon and Sprint.

Thoreau may have gone to the woods to “live deliberately, to front only the essential facts of life,” but he was no hermit, and I think he would have found something to like about the Internet. If we’re lucky, some modern Thoreau is blogging (or Uttering) right now from the slums of Kolkata or the slopes of Kilimanjaro.

Comments (3) | Reprints | Share:        

It’s the technology journalist’s downfall: The hot technology that you suspect isn’t quite ripe but you can’t help writing about anyway.

In 2003, when I was a senior editor at MIT’s Technology Review (and, in the interest of full disclosure, Bob was editor in chief), speech recognition and natural-language processing were firmly in that category, yet we went ahead and published a cover story I’d written called “Computers That Speak Your Language.” Exercising a little more enthusiasm than usual, I predicted that computers equipped with speech recognition software “may soon be able to interpret almost any conversation, or to retrieve almost any information a Web user wants.”

But as is well known to anyone who’s been stuck in the telephone “death spiral”—the actual term experts use for one of those interactive voice-response conversations where you can’t get the frakkin’ computer to transfer you to a real person—most voice interfaces are still a long way from maturity. To the charges of overeagerness and a little gullibility, I plead guilty.

And yet—as hesitant as I am to go there once again—I am compelled to report that speech recognition technology may be taking a real lurch forward. Last week, the folks at Harvard Square startup Vlingo gave me a demonstration of a cell-phone-based voice-recognition system that not only works, but gets more accurate as more people use it.

Vlingo, formerly called Moebius, is officially emerging from stealth mode today and unveiling Vlingo Find, an application that users of Sprint mobile phones can download at www.vlingomobile.com. The software, which works over the phone’s data connection, allows users to search for business phone numbers nationwide simply by speaking the business name or category and the city name—for example, “Thai food in Cambridge, Massachusetts.”

If that sounds like a glorified 411 service without the human operator, that’s what it is. But that’s more remarkable than it sounds. After all, there is no human operator—nobody listening to the words and typing them into a computer somewhere. Instead, software runs the digitized sound of the user’s voice through statistical language models to make its best guess at what the user said, then sends that text back to the user’s screen for verification. If it’s correct, the user can press “search” to get the phone number. If there’s an error, it can be corrected by using the phone keypad cursor buttons to scroll back to the mistaken word and speaking it again.

The system works—in fact, it works so well that Vlingo officials who wanted to show me how easy it is to correct an error couldn’t get the system to make a transcription mistake in the first place. And Vlingo Find is just a taste of what Vlingo’s planning. The company is in talks with mobile phone manufacturers and cellular carriers to apply speech recognition to any task that normally requires tedious triple-typing on a standard mobile phone’s 12-button keypad, such as entering a URL in a Web browser or searching an online music store for a specific MP3. Vlingo “unlocks access to the mobile Internet with the power of voice,” says CEO David Grannan.

If you’ve had any experience with older speech recognition systems, you know that they achieve accuracy in one of two ways: either by drastically constraining the context so that only a few specific words might come up (”For billing inquiries, say 1, for technical support, say 2″) or through a long training process in which the speaker reads a prepared text and the software learns to recognize that speaker’s (and only that speaker’s) speech patterns.

If there were a Nobel Prize for software, a speaker-independent speech recognition system with a truly unconstrained vocabulary would win it. Vlingo’s system doesn’t do that, but it comes close, using an “adaptive” approach pioneered by the company’s technical founder, Michael Phillips. When I interviewed him last, for my 2003 story, Phillips was the principal scientist at interactive voice-response company Speechworks (which was later acquired for $132 million in stock by Scansoft, which renamed itself Nuance after acquiring a Speechworks competitor by that name). Speechworks’ technology was both grammar-based—loosely put, it expected words to occur in a certain order, such as subject-verb-object—and statistical, in that it used machine-learning techniques to comb databases of phrases for likely matches.

Today, Vlingo’s system uses both techniques, but with an ingenious addition: the ability to improve over time using the corrections users enter into their phones. “We are totally into adaptation,” says Phillips. “It’s the only way to make this work.” Vlingo’s so-called “adaptive hierarchical language models” are far too complex to run on mobile processors, so the heavy lifting actually takes place on servers accessed over the cellular data network. Remarkably, this doesn’t slow things down much—Vlingo can send back a transcription in only a fraction of the time it takes to complete a search once the query is actually submitted.

Vlingo is staffing up using $6.5 million in Series A funding from local venture firms Charles River Ventures and Sigma Ventures. “The technology has come a long way,” says Izhar Armony, general partner at Charles River, which was also an investor in Speechworks. “Mike’s brilliance was to realize that you can have the best of both worlds, user independence and an unrestricted grammar, if there is collective learning inside the network. The system gets better and better as more people call in. This is why we were so excited [about funding Vlingo].”

Of course, some bootstrapping using canned training information was needed to get the system up and running, so the system may still be raw around the edges. But Phillips says he expects it to improve quickly once people starting using Vlingo Find. Eventually, Phillips says, Vlingo could put an end to triple-typing. “The goal is to get carriers to buy into this as a standard part of the software ecosystem on their handsets,” he says. And that might lead to a generation of phones that really do speak your language.

Comments (1) | Reprints | Share: