Monday, December 7, 2009 | 11:30 AM
Editor's note: today Google held a launch event at the Computer History Museum in Mountain View, CA. Fresh off the stage, we've invited Vic to highlight the mobile team's announcements, and the unique set of technologies that make them possible. (All [video] links point to event footage that will be viewable later today.)
A New Era of Computing
Mobile devices straddle the intersection of three significant industry trends: computing (or Moore's Law), connectivity, and the cloud. Simply put:
- Phones get more powerful and less expensive all the time
- They're connected to the Internet more often, from more places; and
- They tap into computational power that's available in datacenters around the world
Just think: with a sensor-rich phone that's connected to the cloud, users can now search by voice (using the microphone), by location (using GPS and the compass), and by sight (using the camera). And we're excited to share Google's early contributions to this new era of computing.
Search by Voice
We first launched search by voice about a year ago, enabling millions of users to speak to Google. And we're constantly reminded that the combination of a powerful device, an Internet connection, and datacenters in the cloud is what makes it work. After all:
- We first stream sound files to Google's datacenters in real-time
- We then convert utterances into phonemes, into words, into phrases; and
- We then compare phrases against Google's billions of daily queries to assign probability scores to all possible transcriptions; and
- We do all of this in the time it takes to speak a few words
Looking ahead, we dream of combining voice recognition with our language translation infrastructure to provide in-conversation translation [video]-- a UN interpreter for everyone! And we're just getting started.
Search by Location
Your phone's location is usually your location: it's in your pocket, in your purse, or on your nightstand, and as a result it's more personal than any PC before it. This intimacy is what makes location-based services possible, and for its part, Google continues to invest in things like My Location, real-time traffic, and turn-by-turn navigation. Today we're tackling a question that's simple to ask, but surprisingly difficult to answer: "What's around here, anyway?"
Suppose you're early to pickup your child from school, or your drive to dinner was quicker than expected, or you've just checked into a new hotel. Chances are you've got time to kill, but you don't want to spend it entering addresses, sifting through POI categories, or even typing a search. Instead you just want stuff nearby, whatever that might be. Your location is your query, and we hear you loud and clear.
Of course our future plans include more than just nearby places. In the new year we'll begin showing local product inventory in search results [video]; and Google Suggest will even include location-specific search terms [video]. All thanks to powerful, Internet-enabled mobile devices.
Search by Sight
When you connect your phone's camera to datacenters in the cloud, it becomes an eye to see and search with. It sees the world like you do, but it simultaneously taps the world's info in ways that you can't. And this makes it a perfect answering machine for your visual questions.
Perhaps you're vacationing in a foreign country, and you want to learn more about the monument in your field of view. Maybe you're visiting a modern art museum, and you want to know who painted the work in front of you. Or maybe you want wine tasting notes for the Cabernet sitting on the dinner table. In every example, the query you care about isn't a text string, or a location -- it's whatever you're looking at. And today we're announcing a Labs product for Android 1.6+ devices that lets users search by sight: Google Goggles.
In a nutshell, Goggles lets users search for objects using images rather than words. Simply take a picture with your phone's camera, and if we recognize the item, Goggles returns relevant search results. Right now Goggles identifies landmarks, works of art, and products (among other things), and in all cases its ability to "see further" is rooted in powerful computing, pervasive connectivity, and the cloud:
- We first send the user's image to Google's datacenters
- We then create signatures of objects in the image using computer vision algorithms
- We then compare signatures against all other known items in our image recognition databases; and
- We then figure out how many matches exist; and
- We then return one or more search results, based on available meta data and ranking signals; and
- We do all of this in just a few seconds
Computer vision, like all of Google's extra-sensory efforts, is still in its infancy. Today Goggles recognizes certain images in certain categories, but our goal is to return high quality results for any image. Today you frame and snap a photo to get results, but one day visual search will be as natural as pointing a finger -- like a mouse for the real world. Either way we've got plenty of work to do, so please download Goggles from Android Market and help us get started.
The Beginning of the Beginning