VoiceRecognition
Latest
Would you buy a voice-controlled camera, or perhaps a DSLR with touchsceen?
Do you talk to your digital camera? Perhaps stroke its glossy LCD? If a pair of recent patent applications are any indication, those mildly creepy gestures might one day actually do something. Sony's just laid claim to a DSLR touchscreen that can be manipulated by thumb even while the rest of one's face is smushed up against the viewfinder, and Canon's got its eye on technology that lets shooters activate advanced camera functions using simple voice control. The latter wouldn't be limited to "fire," but could potentially be directed to switch modes, stops and even zoom in and out of the frame. It wouldn't necessarily substitute for a remote as there are just two modes, "close-talking" for speech uttered when using the viewfinder, and "non-close-talking" when you line up shots on the LCD display. Neat as they are, these alternatives to physical controls make some at Engadget HQ quite sad, but we understand that minimalism is the word of the day.
Quad-copter responds to your voice, isn't coming to a toy store near you (video)
Hey, remember Taiyo's voice controlled toy helicopter from last year? This is way cooler. Like, way cooler. It's an autonomous quad-copter created by the Robust Robotics Group at MIT. It hovers and flies a bit like the AR.Drone that had us smitten at CES this past January, but this one has a lot more brains. It response to natural (though slowly delivered) language voice commands that look to be processed on an iPhone before being sent to the helo. In the video embedded below you'll see it responding to the command "Fly past room 124 then face the windows and go up." Sure enough, it does as instructed, and while we don't know how many takes that particular feat of robotic subservience took to pull off, we're suitably impressed. Mind you, this is a research project and not any product ever destined for retail, so after watching that video a few more times we'll just go back to crashing our Picco Zs into the walls -- and each other.
Sexist computers: male voices are apparently harder to recognize than female ones
Researchers up at the University of Edinburgh have determined that the male voice is harder for voice recognition software to pick up and understand than its female counterpart. This conclusion was reached after telephone conversation recordings were run through a battery of tests, which revealed that men seem to say "umm" and "err" more often, while also identifying that the greatest difficulties arise with words that sound similar and can arise in the same context, such as "him" and "them." Equally troubling is the first word in a sentence, as it comes without context and therefore doesn't benefit from any predictive assistance. Done in partnership with Stanford, the study was aimed at identifying and overcoming the major hurdles to producing usable and reliable voice recognition -- something Google's universal voice translator phone is also aiming to achieve. Let's hope somebody figures out how to get around all our hemming and hawing, eh?
Vlingo 2 adds more voice power to iPhone
In the burgeoning market for iPhone voice recognition apps, the free Dragon Dictate clearly holds the top spot. In fact, this paragraph was dictated using DragonDictate (with some minor edits after the fact). However, the application doesn't always shine when it comes to getting your text quickly and easily into different places where you might want to use it. You can send an e-mail or a text message if you want to, but you can't update your Facebook status or send a Twitter message without copying and pasting. Performing a search requires an entirely separate app. To make your dictation process easier -- at least when it comes to getting your text to go where you want, when you want -- there's the latest update to the Vlingo app, which we first reviewed last June. Vlingo 2.0 is trying to up the ante for text recognition by putting all the 'next step' options in one convenient place. You can search (Google, Yahoo, or Bing), find map items, dial your contacts (with optional contact name upload to Vlingo's servers to improve recognition) and update your Facebook or Twitter status. For email or SMS, you have two in-app purchase options to extend the free app's capabilities; it's $6.99 for either SMS or email action, $9.99 for both. Vlingo has gotten a thorough UI overhaul in this new version, and it's quite a bit easier to use than it was. The dictation button can be used in hold-down or tap-to-talk mode, and the app can be set to recognize speech on launch for maximum speed. You can specify what action you want by speaking it: "Email Joey, Subject how about some coffee, Message Got time to meet me at Starbucks? You're buying!" will create and address an outgoing email for you, all in one step.
Uh oh! Google releasing free turn by turn GPS app for Android
Just when you thought it was safe to buy a nav app for the iPhone, Google goes and announces a free app for Android, and says it's going to be available for the iPhone if Apple will let it into the App Store. The app, called Google Maps Navigation, will ship with phones running the Android OS 2.0 and includes search by voice, search for points of interest by voice while in route, satellite and street view, and support for a hardware dock for 'certain devices.' This won't be great news for TomTom, Navigon, or any of the rest. Google says the app is U.S. only for now, and will be seen first on the Verizon Droid which will appear in November. Talk about disruption! With the FCC watching it will be most interesting to see if Apple allows the app onto the iPhone and, whether yes or no, the effect on Android sales. I can't wait for all the fur to fly. You can watch a video of the app in action in this YouTube clip. Admit it. You want this!
Lighthouse SQ7 MID brings voice recognition to Facebook, Twitter
If you're savvy enough to want to use the Facebook Mobile app, we're guessing you're savvy enough to get it up and running on your phone or portable -- so buying an additional device for responding to friend requests seems a little silly. Hopefully, AdelaVoice (a company that develops voice-enabled apps for the Internet and social networks) has more plans for its Lighthouse SQ7 than what we see here. Based on the Linux-powered SmartQ7 Internet tablet, the heart of this bad boy is a UI that allows you access to Facebook, Twitter, the web, and an image slideshow. But that ain't all! Taking advantage of Webkit's integrated voice recognition, this thing will let you input text through the mic -- perfect for posting to Facebook or tweeting sans stylus. Perhaps for some people these features might be a big deal, but we've been asking around -- and we have yet to find anyone who'll 'fess up. We have to admit, though -- what this thing does do, it seems to do pretty well. Tell you what, AdelaVoice -- add JDate and Classmates.com and we'll give the thing another shot, okay? Video after the break. [Via Best Tablet Review]
2010 BMWs boast improved Nuance voice control system
BMWs aren't exactly lacking for technology these days, but it looks like the new 2010 models come packing a little something extra to make your life even easier. As Nuance Communications is all too happy to announce itself, the new vehicles boast an improved voice control system from the company that promises to allow for a "more conversational dialogue between drivers and BMW navigation and entertainment systems." That, naturally, gets paired with BMW's own iDrive system, and includes some Sync-like abilities to search for music by voice, as well as a new "One-Shot Destination Entry" feature to let folks enter a destination address in one simple voice command. It can even apparently recognize several different language in parallel, letting a French-speaking driver, for instance, search for a German song title or English album title.
Video: High-res image enlargement tech for PS3 and PSP takes extreme closeups to a new level
The GigaPan has opened new doors in ridiculously high-resolution imagery, creating pictures so huge our feeble human minds can't hope to comprehend them. Sony's getting in on the action, not so much with the idea of capturing historic events at gigapixel resolutions, but providing the same sort of effect by linking multiple documents and items together and allowing you to zoom from one to the next. The video after the break shows a number of demonstrations, including what looks to be a release calendar that, inside each cell, contains high-resolution photos and even videos of whatever is releasing that day. Hidden advertising is also possible, like a mosquito on a man's face tagged with an ad only apparent after you've gotten uncomfortably close to the poor man's face. The adult entertainment opportunities are endless. Also demo'd is a new voice recognition library for the PS3, dubbed PS3VR, which can take you straight to your Cindi Lauper albums even if you're not a girl and are seemingly disinterested in having fun. That tech isn't quite so interesting, but it and the high-resolution libraries are intended for release to developers... eventually. Read - High-Resolution Enlargement TechnologyRead - PS3 Voice Recognition
Video: ASUS EeeTop PC ET2002T gets touched up
We might already have all the specalicious details on the new EeeTops from ASUS, but it never hurts to see one in action, right? Taking center stage here is the voice-activated widgetry ASUS has added to its Windows 7 install, allowing you to check the weather, play music, and even initiate Skype calls solely via the awesomeness of your voice. Expected to retail at €599 in Europe, the 20-inch ET2002T definitely looks sleek, though its screen is glossy enough for the camera girl to wave at us surreptitiously a number of times. Check it out, but be warned, there's a small bit of salty language midway through, which is as inappropriate for the office as it is hilarious. [Via Notebook Italia]
Recent Apple patent filing speaks of stealth biometric security on iPhone, other insane insanity
Recent patent fillings by Apple that AppleInsider has dug up point to the company adding several new methods of insuring the security of its devices for users. The new filings cite biometric authentication methods that would excite Ethan Hunt -- including installation of a hidden sensor behind the screen that would recognize the user's fingerprint when touched, and / or a front-facing camera for retinal recognition. The filing also suggests further possibilities, such as the device being capable of recognizing the user's voice, or collecting DNA samples for recognition via genetic code. We've heard plenty of semi-speculative tech tales in the past, so we always take this stuff with a grain of salt, but Apple seems to be moving forward at least into the research phase of such endeavors. To the future we go! There's one more fantastic scribble after the break, hit the read link for more details.
ASUS promises voice-controlled Eee products by end of the year
For a company that gets so much (deserved) guff for endless, boring product iterations, ASUS also manages to sneak in a good amount of crazy functionality -- some of it more gimmicky than others. ASUS has been doing multitouch trackpads since before they were hip, is toying with dual touchscreens and is even crazy enough to build the Fold / Unfold computer. Next up for the firm is voice control, with a whole team dedicated to voice recognition, and plans to ship Eee PC and Eee Top products by Q3 or Q4 of this year. According to ASUS CEO Jerry Shen they're working with third parties in this effort, so we're guessing they're not going to reinvent the voice recognition wheel. Still, it's a promising development for a line of computers that seem to suffer most when it comes to input methods -- we'll wait and see how well they pull it off.
Voice dial on us: TUAW Vocalia giveaway
In product shootouts between the iPhone and other smartphones, the iPhone usually gets dinged for not having voice dialing built in (although you can get around that with services like Jott). Vocalia, from Creaceed, is a voice-recognition dialing app that has been around for about six months. Vocalia (click opens iTunes) displays a HAL9000-like glowing "eye" after being launched. When the eye appears, speak the name of a contact you wish to call. Vocalia analyzes your speech, then displays the closest matches to what you said. If there's more than one contact listed, you then speak the number next to the correct one. Are there multiple phone numbers for a contact? Just say "home", "mobile", "work" or whatever to dial. Creaceed and TUAW would like to give away seven free promo codes for Vocalia.What do you need to do to enter? Just leave a comment on who you'd like to dial using Vocalia. Good luck, and we look forward to seeing your creative answers! Open to legal US residents of the 50 United States and the District of Columbia who are 18 and older. (Sorry, we know our international readers want to participate, but promo codes don't transfer outside the US.) To enter leave a comment telling us what person you'd like to call using Vocalia. The comment must be left before February 28, 11:59PM Eastern Time. You may enter only once. Seven winners will be selected in a random drawing. Prize: Promo code for free copy of Vocalia for iPhone (US$3.99) Click Here for complete Official Rules.
MacSpeech update something worth talking about
MacSpeech Inc. unveiled a new update to its well regarded speech recognition application for the Mac today. The release of version 1.3, available immediately by download, has some significant enhancements and fixes. To test this new release, I've dictated this review into our browser-based content management system without much trouble. I had to correct a couple of uncommon words, but that is easily done in the spelling mode.One of the interesting features of the new version is a 'cache document' command that will allow MacSpeech to read a document's contents in order to navigate that document, and to enter and edit text. This improves MacSpeech Dictate's ability to work with any document, including documents not created with the program itself. MacSpeech has also greatly improved the help system and added several visual enhancements that will allow users to see the status of the program and the mode it's operating in.The new 'press the key' command will input a specific keyboard key. This allows using modifier keys, such as Control, Option, and Command. MacSpeech Dictate has always had a strong speech recognition engine, but was sometimes weak in the software that surrounded it. Early on, the company distributed some bad install disks, but that was quickly remedied.MacSpeech Dictate requires an Intel-based Macintosh computer with a minimum of OS X 10.4 .11 or 10.5 .2 or higher. Here is a link to more details of the update.MacSpeech Dictate had its last update in November of 2008, when version 1.2.1 was released. Existing users of the program will be prompted to do the update the next time MacSpeech launches. In my case, the update worked well, and without incident.MacSpeech Dictate is US $199.00 direct from the developer or at retail including the Apple online and brick and mortar stores.
Zumba Phone: the Steorn Orbo of cellphones?
Ready for this? The BBC is reporting on a "top-secret phone" developed by a 40-person strong ejector-seat technology company -- IA Technology -- from Hereford England. The Beeb calls it, "the world's first fully accurate voice recognition system for mobile phones." A bold claim (to say the least) from seemingly nowhere. The report starts with a dramatic lead-in, "It's a secret world, much of which we can't film," before introducing us to the Zumba phone. A phone so secret that the BBC can't even show us how it works. They do manage to demonstrate how it's worn. A retro slug-on-the-ear type earpiece is removed from a thin slab sporting a display and circular dialer reminiscent of Bang & Olufsen fashionphones. The device relies upon a "100% secure" connection to an "Internet portal" (apparently called Zumba Lumba) that holds all your contacts. If the phone is lost or stolen "it is instantly useless to anyone else." A lot of hype to be sure, especially with plans to deliver the Zumba to shops before Christmas. Hit the read link for the video and let us know what you think.Update: Ha! The zumbalumba.com site just lit up with an unreadable, elementary school design and no detail other than a whispy concept called "Natural Connecting." Some sleuthing and cross-referencing by our readers, however, seems to link Zumba CEO, Dean McEvoy, to a group of party promoters. Good luck with this Dean, the whole internet is about to be up in your shiznit.[Thanks to everyone who sent this in]Read -- BBC VideoRead -- IT Technology website
Google is listening, and iPhone users are talking
According to the New York Times and the Google iPhone page, the big G is about to add voice recognition capabilities to its iPhone search app (iTunes link), with the new version showing up on the App Store any minute now. You'll be able to ask questions (the kind Google can answer, of course; someone else will have to tell you if those pants make you look fat) and get speedy search results returned from the web or your own address book. [As of Saturday afternoon 3:30pm ET, the new version is still not in the App Store; interestingly, the Google page no longer has the video embed of the voice recognition demo, nor does it show the new feature listing. Perhaps there was a last-minute delay? Update, Sunday AM: Techcrunch is reporting an anticipated Monday release for the new version. No details given on the reason for the delay. -Ed.] Update, Monday PM: The app isn't listed as updated in the App Store, but removing the old Google Mobile app from the phone and reinstalling it brings it back with the voice search capability. Thanks to everyone who sent this in! The voice recordings from the application will be sent to Google's servers for processing "in the cloud." Given the size of Google's infrastructure and its ability to transfer gargantuan amounts of data, it may have the advantage over existing voice-driven search apps such as Yahoo's oneSearch. Google execs seem hesitant to quantify the accuracy of the voice recognition software, but here's hoping it at least tops some of the iPhone voice-dialers I've been playing with. You can see a sample video of Googlers testing the feature in the continuation of this post. [via Truemors / blogtopten]
Moshi's IVR Alarm Clock accepts voice commands, harbors own primary directives
The Moshi IVR (Interactive Voice Responsive) Alarm Clock is a chatty one. Instead of those perfectly reasonable physical buttons which have been providing us with additional five minute segments of sleep since time immemorial, the Moshi IVR wants to talk you through your morning. To activate a command, just say "Hello Moshi." Moshi will most likely mutter something threatening under its virtual breath, and then ask you "Command Please," to which you can instruct the clock to tell you the time, set the alarm, inform you of the temperature and a bunch of other things. Sounds great, and we doubt Moshi will be able to murder you in your sleep without any life support systems or pod bay doors under its dominion, but we still fear the prospect of attempting to reason with a too-smart-for-its-own-good alarm clock during one of those dreaded "before noon" hours of the day. Moshi IVR is available now for $50.
AT&T's Watson wants you to talk your TV into changing the channel
AT&T's been cobbling together a voice recognition system for decades, affectionally known as Watson. The good news is that unlike most every other voice recognition system out there, AT&T's might be useful for, you know, stuff. AT&T's working on making your plain-language utterances into real-life gadget actions, and has built a voice-operated TV remote to make the idea a reality. Built to operate the U-verse service, AT&T's new mythical remote lets you ask for whatever comedies might be on, or -- in the inexplicable example cited by our source article -- search for movies starring Bruce Willis, all using your voice in regular person sentences. AT&T's also working with other developers to build applications like a yellow pages app for the iPhone. There's no word on when (or if) AT&T will actually offer up the remote for consumer use, or when other applications are due to hit the mainstream, but we must say we're ready for the tide to shift in human-computer interaction, where the computer does the heavy lifting. Because we have some Bruce Willis movies to watch.
First Look: Vocalia English speech recognition tool for iPhone
I want my iPhone to have voice dialing. Belgian developers Creaceed are moving in the right direction by introducing the Vocalia English speech recognition tool for iPhone. This app just does one thing right now -- displaying contacts from your Address Book when you say names. You still need to tap on the name that appears to open the contact, then tap on a number to dial it. Creaceed is working on adding new functionality to future versions.How well does Vocalia recognize names? I tested it by reading the first 100 names in Contacts. Without training Vocalia, it was able to correctly recognize 98 of the names -- that is, it put them on a screen with three of the closest sounding names. For 83 of the names, the correct name was at the top of the list. The two names that weren't recognized properly were Guilherme and Guiseppe. You can also add nicknames or other pronunciations to any name to further improve accuracy.Between now and October 18th, Vocalia is on sale for US$3.99 in the App Store (click opens iTunes). The regular price of Vocalia is $7.99.
FineDigital showcases voice-recognizing FineDrive X700 GPS
It's been a hot minute since we've seen anything fresh from FineDigital, but the firm has just introduced the slim and seductive FineDrive X700 over in South Korea. Humorously, the company is asserting that this particular unit is the first of its kind to offer voice recognition, but even a light amount of research would help it understand such a statement is far from being true. Nevertheless, the unit does look to offer a spacious screen and a bundled remote, but details beyond that are unfortunately absent.
Audeo "Think N' Spell" neckband allows voiceless phone calls
You know what's wrong with phone conversations? All that dang talking you have to do. Luckily for us, some scientists at Ambient Corporation are bent on fixing that with their "Audeo" wireless neckband which taps into nerve signals being sent to the vocal chords and vocalizes those "thoughts" for you -- without all that pesky implant nonsense. Users have to specifically think about voicing words for them to be picked up by the band, but it saves them from saying potentially sensitive things while on a cellphone in public. The device is also being aimed at those who have lost the ability to speak due to diseases such as Lou Gehrig's Disease. Currently the device can only recognize roughly 150 words and phrases, and has a decent delay between "speaking" with your mind and the computer vocalizing the words for you, but the upcoming version will be phenome-based and therefore allow you to speak anything -- though it will come at the cost of even slower recognition. Video is after the break.
