Say What? Voice Recognition Apps, Maps And Traps

movedearΤεχνίτη Νοημοσύνη και Ρομποτική

17 Νοε 2013 (πριν από 5 χρόνια και 1 μήνα)

382 εμφανίσεις

Say What? Voice Recognition Apps, Maps
And Traps
November 19, 2010
3:40 PM
Listen to the Story
All Things Considered
5 min 38 sec
Instead of talking on your smart phone, think about talking
it. And
don't forget to tell your tablet what's on your holiday shopping list.
Courtesy of Nuance
Advances in voice recognition technology are
making all of this possible. The technology has
migrated to a number of free apps created by
Google and Nuance, the company that owns
and sells Dragon NaturallySpeaking software
for computers.
Nuance's apps — essentially stripped down
versions of Dragon — for the iPhone and iPad
Dragon Dictation
Dragon Search
) and
Blackberry (
Dragon for E-mail
) transform words
into action.
Dragon Dictation allows you to dictate things and then use your
fingers to paste words into e-mail, texts, or into Twitter or Facebook.
And Dragon Search lets you search the Web with your voice.
Don't Forget To Buy Bananas
I tested out both apps on an iPod Touch and an iPad. The search
app works the same on both. But the dictation app for the iPad
allows you to save your dictation as a note and then update it
whenever you want. This feature is especially handy for things like
shopping or to-do lists.
Nuance's app for the BlackBerry is directly integrated into e-mail so
that you can use your voice to address the e-mail, specify the
subject and dictate the text into the body of the message. But you
still have to use your fingers to send it on its way.
Google's New Voice Actions
Nuance does face some competition from Google, which has
developed its own voice recognition technology. It's used on
Android mobile devices such as the popular Droid family of phones
and Samsung's new tablet, the Galaxy Tab.
Courtesy of Google
I downloaded Google's
Voice Search App
a Droid X smart phone on Verizon's network.
Then, I tested out the new "
voice actions
commands (this requires Android 2.2) while
walking the streets of Washington, D.C., during
rush hour.
The voice actions allow you to do everything
from listening to music to sending a note to
yourself (akin to the shopping list feature
described above). You simply press and hold
the phone's search button (you can also click
on the microphone icon on the screen) to call
up the "speak now" command.
So, how were the results? It was spot on most
of the time. But accuracy largely depends on
how clearly you speak, whether there's a
foreign word or name involved and if you hesitate too much (think,
before speaking).
A Portable Phonebook, Navigation Tool
One of my favorite features is the ability to call businesses or
restaurants by simply saying the name of the establishment and the
city, such as "call the Metropolitan Opera in New York City." The
phone leaps into action as if you have a personal concierge at your
The same goes for navigation: The command "navigate to the
Washington Monument" provides turn-by-turn directions for driving
or walking using Google Maps.
Monet At A Moment's Notice
Google has had a basic voice Web search built into Android phones
for some time. But the latest version is much improved. By saying
"pictures of Monet," images of the artist's work pop up within
Bill Byrne, who heads Google's voice interface team, says his group
is especially interested in developing more voice search features for
the Web for people who don't speak English. The voice search
feature is available now for more than a dozen
. That's in
addition to seven versions of English based on someone's accent.
Basic voice Web search is built into Android devices by just clicking
the microphone to the right of the Google search window. It's also
available for other smart phones by downloading the
Google Mobile
for the iPhone and BlackBerry.
Dragon Software Advances
Nuance's apps give you a feel for what their software can do on a
computer. But it's just the tip of the iceberg. That's because the
software actually learns from you and you can personalize it.
I tested the software on a PC laptop. The price of the software
remains relatively accessible — it costs about $200 for the premium
edition for PCs (it's also available for Macs).
This summer, Nuance released version 11 of Dragon and there are
some noticeable improvements in accuracy over the
. It actually watches, and then learns from corrections
entered on the keyboard. That's in addition to corrections you can
make using your voice.
It also will learn your writing style by searching through documents
or e-mails you specify. A number of authors are using voice
recognition software to write books. Some, like David Henry Sterry,
have turned to the technology because of wrist, hand or finger
injuries from relentless typing.
Sterry says he prefers it to typing because it creates a "more direct
synaptic connection to my subconscious."
Custom Voice Commands
One of my favorite new features is a pop-up screen where you can
easily access tips and commands. There's also a nifty tool for
creating custom voice commands, which enable you to say a
keyword or a phrase and Dragon then inserts a passage of text
you've already dictated or written. This is a handy feature for
anything you might normally type over and over again, such as a
mailing address, a customized e-mail signature or any type of legal
In my conversation with NPR host Robert Siegel, we demonstrated
how voice recognition technology is still far from perfect —
especially if you're trying to communicate words in foreign
You also have to be connected to the Internet to be able to use
voice recognition apps because the heavy lifting of deciphering
what you said is done by servers in the cloud — not directly on the
smart phone or device.
Using Dragon NaturallySpeaking effectively also requires you to be
in a relatively quiet space. I sometimes run a fan at my desk. And if
the microphone is on and I'm not speaking, the program thinks I'm
saying the word "him" over and over again and will proceed to fill
my screen with that single word, akin to what happened to the
villain Carl in the movie
It's easy to turn the microphone on and off with your voice by saying
"go to sleep" and "wake up!" But saying this just results in a lot of
sideways glances from my colleagues, who can only assume that
I'm sleep deprived.
Related NPR Stories
©2013 NPR
Technology Podcast
Talking Back To Your Device Has
Never Been Easier
Aug. 11, 2009
I'm Listening: Products And Services
That Talk
Aug. 10, 2009
Online Dating: OK, Cupid, Show Me Your Seal Of Approval
'Epic Mickey' Takes Disney's Mouse Into A Video Game 'Wasteland'