This lesson was driven home most recently when Nuance Communications acquired Swype, Inc., a small, company whose single product is software makes it simpler and faster to use your finger (or a stylus?) to enter text on a smartphone. According to this SEC filing and this comment by my associate Greg Sterling on Internet2Go, Nuance is committing something on the order of $100 million to merge Swype into its operations, presumably to add Swype capabilities to its Flex T9 interface.

In his post, Greg questions why Nuance needs to buy Swype, even as Flex T9 already outperforms Swype in some respects. He presents a list of possible answers to his question, including what I would call “the three P’s,” patents, positioning and personnel. But I would add a fourth: “Packaging.” Swype’s market proposition was simplicity itself: “Swype is a faster and easier way to input text on any screen.” That’s it. And it won over a following.

As for Nuance, it is following its time-tested modus operandi for constant improvement and refinement through acquisition. It’s been doing to bring specific refinements to its product portfolio automated speech processing – including voice biometrics, directory assistance/auto-attendant, text-to-speech rendering, dictation, medical transcription and other domains that benefit from solving specific problems. The core T9 technology came with the acquisition of Tegic from AOL-Time Warner in 2007 for $247 million. At the time I called it part of Nuance’s “Pay It Forward” strategy of constantly improving a mobile subscriber’s ability to take control of his or her device and the services provided through it.

For Nuance, it adds to its portfolio of technologies that support highly personalized mobile services. For Swype, it provides a windfall, liquidity event that satisfies its founders and investors. For those of us into pattern recognition, this acquisition is similar to the one that brought GroupMe – a year-and-a-half old, single-product company – into Microsoft/Skype’s pantheon of products and services. A journalist at the International Business Times-Australia accurately referred to GroupMe as “a super-small, zero-revenue tech company founded only last year at the Techcrunch Disrupt Hackathon.”

So why did GroupMe fetch a multi-million price (rumored to be between $65 and $78 million) from Microsoft? The resemblance to the Nuance/Swype deal is striking. First, GroupMe solves a known problem confronted by millions of mobile users as they try to form small groups on the fly. Or as GroupMe puts it on its Home Page: “Group Messaging from Any Phone.” Second, it’s not an “app,” it’s part of the user interface. It can be generalized across multiple modes and media. As we’ve learned with Google+ and, with some difficulty on Facebook or LinkedIn, we humans benefit from group-forming utilities. GroupMe’s core product may be the result of a mere eight hours at a TechCrunch Hackathon (using Twilio’s tools on Amazon Web Services EC2, incidentally) but it fulfilled a recognized Gap in Skype’s mobile user interface.

The acquisition enhances the Microsoft/Skype e-services genome. It’s a grand slam home run for GroupMe. From the mobile subscriber’s point of view, it improves his or her ability to take control of their device and the services it offers. See the pattern?

Readers who are into pattern detection should see that these companies comprise small group of companies taking leadership in defining a better user experience that regards better speech recognition and language understanding as key to satisfactory fulfillment of each individual’s expectations.
Other participants in Interactions funding are North Hill Ventures, Cross Atlantic Capital Partners, and Updata Partners.

Interactions achieved notoriety of sorts (at least for the speech-enabled customer service crowd) by publishing survey results depicting “overwhelming dissatisfaction with IVR.” The study, conducted by Liel Leibovitz, New York University Assistant Professor of Communications, and commissioned by Interactions, contained no real surprises. Respondents indicated that they found IVRs hard to use and that two-thirds prefer assistance by live agents. Interestingly 15% listed IVRs as their preferred channel for communications with their selected vendors.

The study validated the shortcomings of today’s IVRs, which Interactions and its investors has interpreted as a “experience gap” between live agents and IVRs. The products and services offered by Interactions seek to close the gap by providing what Opus Research has been calling “assisted self-service” – meaning that they introduce enough human input into the IVR experience to enable the system to recognize the intent of a call more quickly and respond most appropriately. This should lead to a higher success rate and correspondingly lower rates of dissatisfaction with the voice channel.

The company’s stock-in-trade is to apply its research and development energy toward “filling gaps” in the current speech processing technologies. CEO Ehud Halberstam had worked with the team of engineers at Microsoft who launched the Windows Phone 7 operating system and attendant services (including its voice user interface). Nir Segal, VP of Marketing and Business Development, managed the user experience development team for MediaMind Technologies, a global company whose core product is a platform for multichannel, rich digital advertising campaign management.

In a wide-ranging conversation this morning, we discussed how building a “delightful hands-free experience” should not necessarily depend on such “trigger words.” Indeed, as the Kinect system from Halberstam’s old employer Microsoft demonstrates, the addition gesture and video serves to improve the system’s ability to understand a user’s intent without resorting to the use of a keyboard or physical mouse.

Today, Nuiti employas core speech processing technology from Microsoft and have put together two families of solutions. FireSay InPage supports hands-free command and control of Web sites (a la KitchenPC.com). FireSay VoiceAds provides a voice based mechanism for delivery of advertising and promotional messages across multimodal channels.

Earlier this year I coined the term “No rest for the RESTful” to dramatize how new tools and API’s to support agile programming would accelerate application development and service delivery. As we enter 2011, the mantra appears to be “Speech is Sexy,” as evidenced not just by Apple’s “Help Wanted” posting, but by Google’s recent acquisition of Phonetic Arts and rapid-fire refinement of mobile user interfaces from Nuance, Vlingo (powered by AT&T’s Watson Engine), Google and Microsoft/Tellme.

The big difference in 2011 is that speech is getting more pervasive while, at the same time, it is being subsumed into multimodal user interfaces. Microsoft, for instance, continues to call speech recognition “foundational” to its user interface but, with the introduction of Kinect, already puts much more emphasis on accurate recognition of gestures. Last year Google’s Mike Cohen explained his objective of making speech as an alternative “every time” a keypad or keyboard is used on a mobile device.

2011 will be a year for smoothing out some of the rough spots in speech enabling the user experience. Candidates include better (more accurate) recognition, noise cancellation, more “human sounding” text-to-speech rendering, speech-to-speech translation, low-latency interaction with dynamic data “in the cloud,” and (to keep things safe, secure and personalized) voice biometrics-based authentication or ID proofing.

But there’s been a fundamental change. I used to write about the “Voice User Interface.” In the coming year attention will be on the “Mobile User Interface” that includes voice. That’s why it should not be a surprise to see that the the speech software engineer at Apple should be prepared to work on a “team” to accommodate the “rapidly changing priorities.”

There’s a video demonstration embedded in this blog post on Vlingo’s site. As its author explains, the Action Bar uses the same “predictive intelligence” for both voice recognition and typed commands. This allows users “to initiate actions when it’s not as convenient speak to their phones – like on a noisy bus or quiet conference room.”

Well said, and much needed. As voice-enabled applications mature, developers have come to recognize that all mobile commerce apps must support multi-modal input. This positioning is especially important on Android-based devices where, as we noted back in August, Google has made significant investment in rapid, seamless access to speech recognition for command and control of a multiplicity of apps, features and functions on the Android, as well as dictation for populating search boxes or messaging utilities.

Mobile subscribers are the beneficiaries of the efforts by Google, Vlingo, Apple (which owns Siri), Nuance and a handful of other firms who are goading each other on to improve the mobile user interface. Improving recognition accuracy, as well as predictivity (which I know is not a word) is the task at hand and these technology companies are continually raising the bar in what is much more than “hands-free to hands-free” combat.

The service is described in greater detail in this blog post by Amir Mane the product manager along with Glen Shires from the Google Voice technical staff. Google has made personalized search into an “opt in” feature of the new Google Voice Search app. Recognizing that there is a fine line between personalization and invasion of privacy, it has taken a further step of providing a mechanism for the personalized voice profiles to be “disassociated” from information in your Google account (which, for many people could span Gmail, shared documents, contact lists and other sensitive information).

The app is available from the Android Market, making personalized voice search a few clicks away. The allusion to “improved name recognition and speed” is almost nostalgic. For many years, Amir Mane’s name was synonymous with automated Directory Assistance, an enclave of the speech processing and information processing world that long-ago began to tackle the challenges of rapid recognition and response to the most challenging utterances – names of cities, towns, streets and people.

Google is about to find out how many people expect to see sufficient benefits from Personalized Voice Search” to justify their decision to “opt-in.” My suspicion is that the numbers will be fairly small at first because people generally have to be given incentive (preferably financial, but often merely gratifying) to “opt-in” to just about anything. The promise of better speech rec may not fill the bill. Regardless of the percentage, Google has a large enough sample of subscribers to learn who takes the step into personalized Voice Search.

After overcoming the opt-in hurdle, Google will learn which of its users are sophisticated enough to go deep into the administrative layers of Google Voice to “disassociate” their voice profiles from the rest of their Google account. My suspicion is that the number will be pretty low. After all, people are sharing their location, their check-ins and other activity streams routinely. The idea that some “bad actor” might benefit from associating audio files (or the metafiles derived from them) with other publicly available information seems pretty remote. But I’m always surprised at what self-described privacy advocates decide to address as communications, search and transaction processing technologies move forward.

My suggestion: If you have an Android phone running version Froyo or above, it will be worthwhile to upgrade to Personalized Voice Search. As Amir’s post notes, the improvements will be subtle at first, but they will be beneficial. Thanks to advancements in microphone technologies (like putting multiple microphones on devices to identify and cancel out background noise), as well as acoustic modeling and filtering mobile devices are getting much better at supporting person-to-machine conversations. Personalized Search moves along a different vector to try to provide a more accurate way to recognize spoken commands or search terms consistently.

My take is a little different. Last month I wrote this post about Google’s “Home Field Advantage” as it provided a number of widgets to invoke its apps with the touch of a single button on the “home screen”. This is not about how “deeply integrated” voice is with the core platform or mobile operating system. It is, instead, a matter of limiting the number of clicks it takes to invoke speech enabled applications (or “Speechable Moments”, as I like to call them).

I’m not sure what the attrition rate phone users as they click through an icon on the some screen to invoke an application or feature. I remember that Web-based e-commerce experts used to say that you lose one-half of your customers each time they have to click on a link to get to a check-out. That was Amazon.com’s key motivation to launch “one-click check-out”.

The 3:1 ratio of speech enabled search on Droids compared with iPhones conforms to that rule. Users can’t invoke Voice Search unless and until they launch the Google App and then tilt the phone to its proper position. For Bing, voice search begins after opening an app and pressing a button. They don’t take a lot of effort, but they pose sufficient barriers to have a measurable effect. A number of speech enabled services (Greg mentions a few) are deeply integrated with the call-flows and API’s to support the a range of services, including search, text entry, Twitter and more. They would definitely get more use if they were easily or automatically invoked from the home screen or idle screen.

Speech apps work best when they appear to be the product of serendipity. I was recently given a new Plantronics M1100 “Savor” bluetooth headset to try. I was pleasantly surprised upon reception of the first incoming call. A synthesized voice (or stored recording) announced an “incoming call” and provided a prompt to the effect that I could say “answer” or “ignore” the event. I didn’t have to set up the feature or “train” the unit to recognize my instructions. Upon pressing the command button on the earpiece, the unit also calls “Bing 411″ (Tellme’s Directory Assistance) in response to the command “Call information.”

By contrast, I’ve been told that the unit also supports Plantronics’ a number of other speech-enabled services, with the help of its proprietary, hosted service called “Vocalyst”. Registered users can configure the system to support a reminder service (based on Evernote); e-mail review, reply and origination (supporting Gmail, Yahoo! and AOL); Twitter; and review and origination of text messages (on Blackberry’s and Android-based phones). One of these days I will configure Vocalyst but, in the mean time, I’m very happy with the information I can get from Bing 411 and in the normal course of using my wireless phone.

Parakeet does not represent any technological leap forward. The program displays text versions of utterances for which the recognition engine has the highest confidence levels while it simultaneously displays several other renderings which carry lower degrees of confidence, sort of a range of so-called “nBest” candidates. Users can then indicate their actual utterances by using their fingers on the keyboard.

The originator of a spoken message performs first level triage by making sure that the message that is transmitted reflects what was spoken with great accuracy. I think this should be incorporated as an option for every mobile speech application.

First of all, it is wonderful to find an instance where the abbreviation “UC” does not stand for “Unified Communications”. In the case of Redstart Systems, it refers to its flagship prduct “Utter Command“. Redstart’s UC is a set of applets that interworks with Nuance’s Dragon NaturallySpeaking. Together they provide PC users with the ability to use their voices in ways that truly replace keyboards. They provide a consistent set of spoken commands to do such things as open programs, prepare the header and address information for an email (including cc:’s), go directly to a Web site, move a cursor, find keywords or phrases in any program or document and even change tracks in a media player while working in a different program.

The company’s developers based the design to solve problems that PC users encounter every day. For desktop use, NaturallySpeaking can be considered the speech-to-text rendering engine while Redstart enhances with a wrapper of concise commands. The $295 licensing fee may seem a bit steep, with the required Dragon NatruallySpeaking “Pro” already carrying a $625-$900 retail price.

In spite of the steep entry price, we see some real promise in the approach taken by Redstart. It uses Dragon as a “core” speech-to-text rendering engine, which it has proven to be very good at. It then acts as a developer/partner that creates a market for dictation augmentation. We are already witnessing similar partnering structures taking shape in the mobile world. As Igor Jablokov or Yap mentioned to us, in two recent partnership announcements (SHAPE Services GmbH and MyCaption.com) Yap is the speech-to-text rendering engine while the to partners deliver the workflow and scripting that provide a complete solution.

These constant changes, refinements and, ultimately, “mashups” are both the causes and the outcomes of the advent of “Recombinant Telephony”.

Advisories are available to registered users only.

For more information on becoming an Opus Research client, please contact Pete Headrick (pheadrick@opusresearch.net).