Monday, July 8, 2013

8 Recognition Apps Work Almost Like Magic

In my Scientific American column this month, I noted that in consumer electronics the promises of magic sells. And one of the most important areas for magic simulation these days is recognition. That's when your phone or computer recognizes human speech, motion, visual cues and audio.
You've probably heard of some speech-recognition efforts, like Apple's Siri and the dictation program Dragon NaturallySpeaking. But the world is teeming with apps that recognize other sights, sounds and stimuli. Here's a taste.
Evernote (free; Android, iOS, Mac, Windows): Evernote is the popular notepad-of-all-trades app that keeps your notes synchronized across all of your gadgets. If you snap a photo of something that includes writing (or paste in such an image), even handwriting, its behind-the-scenes optical character-recognition algorithms decipher the writing as text. The accuracy isn't quite good enough to convert the writing into typed text, but it's good enough to let you perform searches on handwritten notes. That is, you can pull up the image of your scanned or photographed handwriting by typing a keyword into the search box.
Shazam (free; all smartphones): This app recognizes recorded songs—popular or not. You're sitting in some restaurant, bar, office or elevator; you let the app listen to the music for a few seconds and marvel as it tells you the song title, singer, album and so on. With another couple of taps, you're buying the song on iTunes or watching the music video on YouTube.
SoundHound (free; iOS and Android): This one is just like Shazam, but with one delicious added feature: you can also hum or sing into it. This is the app to use when something's running through your head and you wish you could remember the name of the song. Uncanny.
Bird Song ID (£3, about $4.60; iOS): It's Shazam for birdsong. Record a bird's warble—the cleaner the recording, the better the luck you'll have—and let the app tell you what species you're hearing. No Internet connection required for this task.
Leafsnap (free; iOS): Take a photo of a leaf (against a white background) and let the app's visual-recognition feature identify the plant it came from. An ingenious field guide, although it requires an Internet connection to perform its magic.
LookTel ($10; Mac, iOS): This company makes a pair of apps ($10 each) for people with vision impairments. Hold your phone over a bill—dollar, pound, euro, doesn't matter—and the app speaks: "20 dollars," for example. There's also LookTel Recognizer, which can identify and speak the name of precisely the kinds of things a blind person might have trouble differentiating: cans of food or soda, packages of food, money, videos and so on. The catch is that you first have to train it. You have to take a picture of each item and speak its name in your own voice: "Sprite Zero," for example, presumably with the help of a sighted person. Thereafter, the app briskly and confidently speaks the name of each item it's learned when the phone's camera sees it.
Color Identifier ($2; iOS), Color Reader (free; Android): Apps like this one (there are several) perform one simple, obvious function: they identify by name any color. Color Identifier lets you switch between simple color names ("pale green") and more exotic ones ("lavender rose"). If you're color-blind, this sort of app is of great help when you're trying to pick matching clothes in your closet. Or you can aim the phone's camera at, say, a shirt, and find out by name its color before you make an embarrassing fashion faux pas.

No comments:

Post a Comment