Siri, the virtual assistant built into Apple’s iPhone 4S, seemed to catch competitors off guard. But all along, Google has reportedly been working on its own voice-controlled assistant for Android phones that responds to natural language.

The project is apparently codenamed Majel, and may see an initial release by the end of this year, Android and Me reports, based on unnamed sources. The codename is a reference to the Federation Computer in Star Trek, whose full name is voiced by Majel Barrett-Roddenberry.

Android already supports voice commands, but the syntax is rigid. If you want to play music, you must say “Listen to…”, and if you want directions, you must say “Navigate to…” What makes Siri so interesting is that it responds to natural language. So if you want to find a place to eat, you could say “I’m in the mood for…” or “Show me some nearby restaurants.” The program picks up on keywords to figure out what the user wants. Reportedly, Majel will have similar capabilities.

The initial release will be limited to search queries, Android and Me’s Taylor Wimberly reports. Presumably that means you’ll be able to find nearby points of interest or get answers to questions, but won’t be able to perform phone actions such as navigation with natural language. Also, Wimberly hedges a bit by saying that January or February is a more realistic time frame for the arrival of Majel.

Although we’re still in rumor territory, Google has previously shown an interest in Star Trek’s ideas. In 2010, when the company acquired Phonetic Arts, whose technology generates computer speech from small voice samples, Google’s Mike Cohen said “we’re confident that together we’ll move a little faster towards that Star Trek future.” More recently, Google’s Android product management director Matias Duarte likened Siri to the droids of Star Wars–essentially, comic relief–whereas his company was inspired by Star Trek, in which everything is voice aware.

From those remarks, we at least know that Google has a vision, but only in vague terms. If the latest rumor is accurate, we may soon get a better idea of what the company is thinking, and how it plans to seriously compete with Apple and Siri.

[This post republished from Techland.]

The other day I received a package, or I should say what in hindsight seems a waste of one: a large box, inside which lay a jumbo-sized cardboard egg, from within which I plucked a tiny rectangular piece of colored paper slightly larger than a business card. On the card, a picture of an iPhone, a greenish tongue of flame, and the words “Introducing…. Dragon Go!”

This is apparently someone’s savvy marketing idea to get my attention (or squander cardboard), perhaps hoping to conjure some latent connection to the dragon eggs featured in HBO’s recently completed (and as of today, multi-Emmy-nominated) first season of Game of Thrones. Intentional or no, I’m making my way through the HBO series now, and here I am, writing about Dragon Go!. Mission accomplished, outsourced PR person!

Dragon, as many of you may know, is the call sign for a suite of speech-recognition tools, the forerunner of which, DragonDictate, was released in the early 1980s for DOS. It’s since been recognized as perhaps the most accurate of the consumer-grade speech recognition utilities (at one point, employed as a computer systems engineer, I provided tech support to a quadriplegic who used Dragon Naturally Speaking—as it was called by 1997—to run his entire home office).

Enter Dragon Go!, basically a freebie iPhone version of publisher Nuance’s voice recognition technology, designed to let you conjure answers to natural language questions. Dragon says it’s partnered with 180 content providers, including AccuWeather, Bing, ESPN, Facebook, Fandango, iTunes, Last.fm, LiveNation, Milo.com, OpenTable, Pandora, Rotten Tomatoes, Twitter, Wikipedia, Yelp, YouTube and Yahoo. The idea’s that you ask the phone a question and it’ll not only recognize what you say, but understand the semantics of the question in terms of the most appropriate content portal.

“Say what you want and Dragon Go! will deliver your results within seconds,” goes the app-info sales pitch. But does it really? After fiddling with it a bit this morning, I can say it does, and with sometimes astonishing intelligence, though it still occasionally fumbles the ball.

When I tapped the record button, said “Harry Potter tickets,” then tapped “done,” the app—which displayed what I said accurately—threw up a list of Harry Potter movies. Okay, no problem. So what about “Harry Potter and the Deathly Hallows part two tickets”? The app got everything right save one word: hallows (it means “revered stuff”). Dragon Go! read this as “hollows,” then threw me over to Google, which—since the movie debuts stateside at midnight tonight—automatically corrected the misread and displayed the proper results for my query. One click later (on “use my location”) and I had both theaters and show times at my command. Even Flixster isn’t that fast.

Saying “coffee Ann Arbor Michigan” conjured Yelp, automatically selecting the sort by “best match” tab. And here’s where thing’s got really cool: Saying “best cup of coffee in Ann Arbor Michigan” auto-selected Yelp again, but this time Dragon Go! somehow parsed “best” intelligently enough to auto-select Yelp’s sort by “ratings” tab instead. And saying “closest Taco Bell” auto-selected Yelp’s sort by “distance” tab, using my current GPS-based location to estimate mileage. Will I now use Dragon Go! to find food on a road trip, say the nearest A&W? Heck yes.

“Latest news on budget deficit” translated correctly and brought me to CNN, but alas, CNN’s search engine failed: The top results for “news on budget deficit” were unhelpful links like “CNN Student News” and “Prime News,” both dated October 2009. I had to scroll down to the fourth link before I found a one-day-old story. Accurate as Dragon Go! may be, it’s only as powerful as each content portal’s endpoint technology.

You can sometimes stump it, of course. Saying “Palin versus Bachmann” took me inexplicably back to Yelp, with a “no results found” screen. And sometimes it punts on portal results: I said “play signed, sealed, delivered” and it brought up a bunch of Google search results (links to YouTube videos) instead of sending me to something more appropriate, like last.fm or Pandora (though to be fair, the tabs for these were just a click away).

This isn’t a formal review—I’ve only spent all of 15 minutes monkeying with the app—but it is a strong first-impressions endorsement. With Dragon Go!, Nuance seems to be giving us our first serious look at the on-the-go, anywhere-you-are semantic web, promised almost a decade ago.

(This post republished from Techland.)

With all the hundreds of TV channels available today, it can be harder than ever to figure out what to watch, Johnston observed. But through a new iRemote app currently in development, you can speak voice commands into a smartphone to get an immediate list of “all reality shows on Thursday night”–and other categories of TV programs small enough to easily digest — on your TV screen.

Other apps, such as Google Voice, can already do voice search, Johnston conceded. But unlike Google, AT&T is focusing its voice search on contents of databases, such as TV listings and other directories, as opposed to the entire Web. Through years of research, he contended, AT&T’s voice search is fine-tuned  for delivering very accurate results in these kinds of voice searches.

“We think that Internet TV is the future,” said Junlan Feng, a researcher in a neighboring booth. On flat panel TV screens there, AT&T showed how conglomerations of tweets from Twitter can be organized into new kinds of categories for display to end users and companies.

AT&T is studying how consumers might tweet their responses to TV shows and then compare their own thoughts to other tweets. As a consumer, you could switch out of a TV program like Meet the Press to look at Top Tweets, Recent Tweets, and Most Positive and Most Negative Tweets, and Popularity of TV Show, measured by number of tweets per day.

But why would TV viewers care about these aggregated Tweets, anyway? “People always want to know what others are saying,” says Feng. After all, she noted, people are already commenting on news stories on the Web and responding to each other. Feng predicted that collected responses from social media might even replace conventional polls like those used for Nielsen ratings some day, since tweets are instantaneous.

Businesses are interested in Tweets, too, to keep on top of public response to “events” such as product recalls. In a demo of its Sonar (Social Network Analysis and Reporting) project, AT&T showed line graphs of tweeted “public sentiment” over its own decisions to first pull–and then reinstate–food programming on UVerse last week.

But research like this calls for development of new ways of analyzing and categorizing data — involving the application of principles of language syntax and semantics, for example — to “give you the meaning of the tweets,” she said. Tweets raise special challenges, I was told. Tweets are short in length–with an upper limit of 140 characters–and are “often ungrammatical.”

In a demo of Talkalytics, AT&T showed how sophisticated algorithms have already been built for quick analysis across large volumes of recorded customer service phone calls. Phone calls are first turned into searchable text through TTS (text-to-speech) conversion. After issues like service outages have been highlighted by computer systems, humans step in to do further trouble-shooting.

AT&T Mobility has been using this kind of technology internally for about a year now, I was told. (By the way, users are informed in advance that their calls might be recorded and monitored.)

AT&T is also working on search technologies which combine voice with gestures for so-called “multimodal search.” One already available app like that, AT&T’s own Speak4it, is aimed at helping users of Apple iPhones and iPads find nearby restaurants, stores and other places.

Although the app is enabled for voice commands, you can also use a touchscreen to draw a circle around a geographic area–or a line pointing in a particular direction — if you have a basic idea of where you’re headed but aren’t sure of the exact address, for instance.

To get voice search and other speech apps out the door faster, AT&T is now collaborating with third-party developers. Vlingo, one partner, has already released a multiplatform smartphone app with hands-free operations that include letting you speak instead of text if you want to do messaging while driving a car.

AT&T plans to launch a Web portal in 2011 that will open up its speech API for wider third-party development. The portal will support creation of new apps for iPhones, BlackBerries, and any other mobile phone environments supported by AT&T at the time.

For me, the announcement served mainly as a reminder that GOOG-411 existed in the first place. Sure, I knew about it, but owning a smartphone pretty much renders 411 services obsolete. Looking up a business on the Internet is faster, and delivers better results.

For kicks, I just used GOOG-411, which won’t shut down until November 12, to look up my favorite bagel shop in Los Angeles. The automated directory took a minute to deliver results, and Google didn’t even have the eatery in its listings. Yelp does, so when I conducted a voice search with Google’s mobile app, finding the information among plain old search results took half the time.

Now, I understand that not everyone owns a smartphone or a feature phone with a data plan. For these folks, there’s still 1-800-FREE-411. And you can still ping Google by text message (“466453,” or “GOOGLE”) to get listings as text, which is arguably more useful than spoken directory listings, anyway.

The reason I’m actually happy, and not just indifferent, to see the end of GOOG-411 is in line with Google’s explanation: They want more resources to work on the next generation of speech-enabled services. Recently, Google introduced Voice Actions, which allow Android users to send text messages, punch in GPS directions and listen to music, all with speech.

I imagine that the initial purpose of GOOG-411, to harvest phonemes for other voice services, is now better fulfilled through these new services and Google Voice. If the end of GOOG-411 speeds the delivery of more speech-enabled features like Voice Actions, or perhaps makes Google Voice transcriptions less awful, I’ll shed no tears for Google’s Old Yeller.

Lesson learned: Disastrous rollout of Google Sixth Sense ends the same way as Wave and Accelerator. [Source]

Lesson learned: Following awkward confessions by voicemail, Google Voice injects a pinch of ambiguity. [Source]

Lesson learned: Don’t cross Google with bogus rumors about acquiring Twitter, Michael Arrington. [Source]

Lesson learned: Google Voice excels at generating sweet band names. [Source]

Lesson learned: Strange threats abound when the world learns you’re a motor vehicle. [Source]

Lesson learned: Porky Pig takes over transcription when all servers are busy. [Source]

Lesson learned: Split personalities will never divulge your contact information to each other without permission. [Source]

Lesson learned: Before introducing herself, Irmalina Babies always pauses for dramatic effect. [Source]

Lesson learned: For a computer algorithm, Google Voice is really bad with numbers. [Source]

Lesson learned: And that’s how babies are born! [Source]

Lesson learned: Rollout of Google Polygraph is an overnight success. [Source]

Lesson learned: For giggles, Google Voice might get you flagged as a terrorist. [Source]

Lesson learned: Hey, it beats Maserati wienerschnitzel. [Source]

Lesson learned: Sometimes, e-mail and weed just aren’t enough. [Source]

Here are a few sent in by readers:

Lesson learned: No one rides to ass pens for free. [Thanks, Erin!]

Lesson learned: Typecast out of Hollywood, Jaleel White found spirituality as the head of his own telco business. [Thanks, Casey!]

Lesson learned: The easiest way to show courtesy is to avoid talking about catacombs and death during business. [Thanks, Marques!]

These last ones are from Harry’s own Google Voice inbox:

Lesson learned: Do not underestimate Technologizer’s influence.

Lesson learned: Google Voice does not support separation of church and embarrassing transcriptions.

Lesson learned: Google Propaganda, another new service, works flawlessly.

Google didn’t announce any of that stuff. But it did roll out two new app/services, one of which it first previewed back at its Google I|O conference in May.

The brand new item is an addition to Android’s voice-recognition features called Voice Actions. The OS already lets you talk to perform Google searches and dictate text into any app that accepts keyboard app–and now you can speak commands to send texts, pull up maps, make phone calls, send yourself notes in the form of e-mails addressed to yourself, and more.

Such as:

“Send text to Mike Morrisey I’ll be there in ten minutes”

“Note to self: pick up chicken wings”

“Go to YouTube”

“Navigate to Daphne’s Greek Cafe”

“Listen to Frank Sinatra”

As with Android’s other voice features, the recognition is done on powerful remote servers, not on your handset. It works pretty well when there’s not a lot of background noise. (Other people talking around you hurts the accuracy, which is an issue: When Google explained where the feature would be most useful, it mentioned bustling venues such as airports and train stations.)

There are a total of thirteen Voice Actions (including Google search). They can be surprisingly smart–when I mentioned local businesses, for instance, the phone knew about them. But in other instances Voice Actions are kind of rudimentary. For instance, the music command only works with third-party apps like Pandora and Rdio, not with Android’s built-in music player.

“Go to [Web site]” seems to work off a canned list of major sites. I can deal with the fact that it doesn’t understand “Go to Technologizer”–it takes you to a Google search for “Technologizer,” not to the site. But I don’t understand why the Voice Action isn’t smart enough to understand “Go to Technologizer.com.” (That too performs a Google search.)

Voice Actions are fun, and they’re a logical extension of Android’a existing voice features. But they’re most interesting as a possible salvo in a new front in the Android-iPhone wars. At the moment, iPhones only have voice recognition courtesy of third-party apps such as Dragon Dictation and Siri. [CLARIFICATION: I meant with recognition of all kinds of words in all kinds of situations--beyond the basic voice commands built into the OS.] Considering that Apple now owns Siri, though, you’ve got to think that the chances are very high that iPhones will get built-in voice, too–maybe in the iOS 5 upgrade. I’d love to see the voice competition get intense.

As for Google’s other news this morning–well, a lot of the Android fans who attended my live coverage already seemed to be well aware of Chrome to Phone. A side project by Google engineer Dave Burke, it uses a Chrome extension and an Android app to let you instantly shuttle information from your browser to your phone.

The extension puts a phone icon in your copy of Chrome. Click it when you’re on a garden-variety Web site, and that site will instantly open on your phone. Click it when you’re in Google Maps in Chrome, and it sends the address to Google Maps on the phone. Click it when you’ve highlighted a phone number, and your phone will dial the number.

For now, Chrome to Phone only works with Chrome and Android phones. But it’s open source, so other folks can build their own versions–there’s already a Firefox to Phone extension. And when someone asked at this morning’s event, the Google reps said that the company might build an iPhone version.

Another audience member asked a question I was curious about: Is there any way to send something from Chrome to phone for future reference rather than immediate use (a la Instapaper)? Nope, not now–but the Googlers said that they’d like to implement a history feature at some point.

Both Voice Actions and Chrome to Phone are available for download as of today, but require Android 2.2 Froyo, which means that they’re not compatible with most of the Android handsets on the planet just yet. (Hey, I thought that anyone who thought Android was fragmenting was hallucinating.) They come preinstalled on Verizon’s Droid 2, which goes on sale today, work on other 2.2 phones, and will be ready for various handsets that are supposed to get 2.2 real soon now (such as my original Droid).

Oh, and Voice Actions, at least, is apparently incompatible with at least some Android phones which are running nonstandard ROMs. The commenters following my live coverage discovered that while the event was still in progress. Like I said, they were a really serious bunch of Android enthusiasts…

A quick rundown of its features:

Web search. Google’s app, you will not be stunned to learn, only lets you search Google. Vlingo lets you search both Yahoo and Google. (Yahoo is the default, but it’s easy to switch between engines on the fly.)

Voice dialing. You can speak the name of a contact to dial his or her number, a feature that lots of folks wish the iPhone had built in. Google doesn’t do this.

Maps. You can search the iPhone’s Maps app from Vlingo to locate businesses (classic example: pizza).

Social networks. You can update Twitter and Facebook statuses by talking. Google doesn’t do this.

Google Mobile App’s signature feature is the ingenious (if apparently questionable) way it notices that you’ve lifted the phone to your ear and accepts voice input automatically once you have. Vlingo doesn’t do that–you need to press down an on-screen button while you speak. It’s less entertaining, but works reasonably well. (I do wish there was more real estate between this jumbo-sized “Press + Speak” button and the smaller ones below it for jumping between features–I occasionally jumped between modes by accident.)

How’s the quality of the voice recognition? In my completely unscientific tests, not quite as good as Google’s–the main problem I encountered was that Vlingo sometimes added an extra syllable or two which I hadn’t actually spoken to the start of my input. But it’s not bad, and it might get better with practice. And Vlingo, unlike Google, sometimes shows you several possibilities to choose from when it’s not sure if it understood you correctly.

Neither Vlingo nor Google does well at all with proper names. Here’s what I got when I tried to search for “Robert Scoble.”

A few other Vlingo quibbles:

It doesn’t use GPS to determine your location. If you don’t speak a location for your map searches, they take place wherever the iPhone’s Maps app happens to be at the moment. (Google knows where you are and uses that information to speed searches.)

It doesn’t have Google’s other local search features. Such as the ability to get movie times and weather.

It doesn’t auto-capitalize or punctuate your social network status updates. unless you speak punctuation marks your updates will look like they were written by ee cummings

Both Vlingo and Google Mobile App suffer from one usability gotcha that only Apple could fix: They launch Safari and Maps in some cases, and while it’s easy for an iPhone app to send you to another application, there’s no smooth way to get back to the original app without relaunching it. The iPhone could definitely use a “return to first app” feature of some sort.

And ultimately, voice input would be much cooler if it were simply built into the iPhone and available everywhere, rather than something you need to launch a specific app for. What if the on-screen keyboard had its own “Press and Speak” button which you could use wherever you’d otherwise tap? I have no idea whether that idea intrigues Apple as much as it does me, but the fact that the company hasn’t gotten around to doing even voice dialing yet may not be a good sign.

Ultimately, I don’t think I’m going to be a regular user of voice input in either Vlingo or Google Mobile App. (I haven’t launched the Google program once since I wrote about it.) I’m comfortable enough as an iPhone typist (tappist?) that voice isn’t a clear time-saver for me. But one of the best things about both apps is this: They’re free, so trying them out is painless. Fun, actually–whether or not you stick with them for the long haul.

Herewith, a few more Vlingo screens…

]]> http://technologizer.com/2008/12/03/vlingo-a-new-way-to-talk-to-your-iphone/feed/ 1 Harry McCracken vlingo vlingo-home vlingo-google vlingo-dial vlingo-maps vlingo-twitter http://technologizer.com/2008/11/17/hands-on-with-googles-voice-search-for-iphone/ http://technologizer.com/2008/11/17/hands-on-with-googles-voice-search-for-iphone/#comments Tue, 18 Nov 2008 07:59:39 +0000 Harry McCracken http://technologizer.wordpress.com/?p=4064 ]]> First, it made a splash last Friday, complete with a New York Times story. Then it failed to arrive when folks thought it would. But now the new Google Mobile App for iPhone with voice search is live in Apple’s iTunes App Store. I’ve been playing with it and enjoying the experience–and while it’s by no means revolutionary, it is pretty darn clever.

Google really wants you to try the voice features: The first time you launch the new version of the app, you get a cutey-cute animated intro explaining how voice search works, and an invitation to watch a training video…

Voice search is indeed simple–you hold the phone up to your ear, wait for a beep, then say your search query. The Mobile App chugs away, for as little as a couple of seconds and as long as twenty seconds or so in my tests…

…and then delivers results for your search.

That’s assuming, of course, that it figures out what you said–the first time I said “Hotel Lucerne New York,” I got this:

And occasionally, the app just throws up its hands, figuratively speaking:

Overall, Mobile App’s success rate was good enough to impress me: It understood me close to ninety percent of the time. And when it didn’t, it was generally with searches that involved names of people or products that aren’t household names. (It pains me to report that I couldn’t get it to understand “Technologizer.” Lord knows I tried…)

Voice Search is at its most obviously useful when it combines your spoken query with its knowledge of your location–thanks to GPS in the case of the iPhone 3G–to give you local results. I happen to be blogging from Cambridge today:

In the end, Google Mobile App’s Voice Search is about one thing: saving you the effort of tapping out queries on the iPhone’s on-screen keyboard. Typing isn’t that hard, and Mobile App doesn’t even necessarily save you time, since it can take a while to parse your vocal input, and you don’t know if the app heard you correctly until it starts to return search results. But it’s nice to have options, and the voice feature does what it does well: Until voice recognition reaches 100% accuracy, it’s hard to imagine anyone executing it much better than Google has.

Voice search will undoubtedly become Google Mobile App’s signature feature. But the program does other stuff, too: You can search by typing if you prefer, and there are convenient links to such Web-based Google services as Gmail, Google Calendar, and Google Reader. If you use Google and you’ve got an iPhone, you need to give this free program a try. And if you do, let me know what you think…