|
When I sit down to browse the Internet, I always sit
in the comfortable desk chair. It's made of very soft
black leather and has padding in all the right places.
The armrests are high and in the perfect spot for my
elbows to rest as I type. When I'm at home in this
chair, I wind up online for at least an hour. I do my
personal Web browsing here (not only because it's
comfy, but also because I would never surf the Web for
personal reasons at the office).
I'll buy J. Crew sweaters, check out prices for a
new car, manage my stock portfolio, and maybe even do
a bit of gaming. In short, I take my time. I look at
things in detail and generally enjoy the process.
Unlike most people, I have a DSL line and never turn
off my computer so it's easy for me to get online and
begin my surfing. Even so, when I sit down, it's
usually for an extended stay.
But when I call for automated information I'm not
sitting in my comfy chair. The chair I'm in while on
the telephone may have very little padding or worse,
might even be a stool. Sometimes I call for
information while running for a plane, or while
driving at high speeds. These are not the moments when
I want to contemplate whether I want a turtle neck or
V-neck on my next sweater, or what my long term plans
are for retirement. I need information, I need it
easily, and I need it in a big hurry. While it may be
an oversimplification to categorize user interfaces
based on the relative comfort of my rear end, it is
oddly appropriate.
A FRIENDLY CONTEST
Even as an avid surfer, there are plenty of things I
don't like to do on the Internet. Movie times, stock
quotes, and airline tickets are good examples of
information tidbits that I find easier to access over
the phone. Data that is time-sensitive and doesn't
require elaborate descriptions like the weather,
flight status, or a quick stock quote are great
applications for the telephone. The problem is that
many vendors don't understand what belongs on the
phone and what belongs on the Web.
Clearly there is some overlap in these two
categories, but the point remains. Even with always-on
Internet access in the home, it's still easier to pick
up the phone for a fast bit of information. And while
the promise of always-on mobile data with 3G wireless
has been heavily touted in the media, it's still a
long way off (if it will be here at all) and is not
the best format for every application.
For fun, and to placate my nerdy competitive
nature, one of my techno-geek friends and I had a
little contest. He had a WAP (wireless access
protocol) phone and I had my plain old vanilla one. We
competed to see who could access a few tidbits of
information faster, my geeky friend with his mobile
data, or me using voice. For this test I made use of
the free services of Tellme (which you will find
reviewed here)
and BeVocal, both voice portal providers who have
800-numbers set up for demonstration purposes.
Test Number 1: Find Tomb Raider playing at a
close location.
- Speech recognition interface: 45 seconds (I was
familiar with the interface so I had a slight
advantage).
- WAP: 3 minutes 20 seconds (this time may have
been affected by variable download rates and the
expected lack of hand-eye coordination by my geeky
friend).
Test Number 2: Find the current value of Microsoft.
- Speech recognition interface: 12 seconds! Need I
say more?
- WAP: 1 minute 23 seconds (assuming the user
knows the stock symbol, which is not needed with
speech rec).
I agree that these are pretty basic applications
that don't have much business use, but this is only an
unscientific example. If we were to imagine a similar
situation with voice-enabled e-mail, customer shipping
information, or inventory, suddenly this becomes very
compelling. Couple that with the fact that both of the
examples above would be a practical impossibility for
WAP while driving, and suddenly it's clear why speech
recognition is so exciting.
BACK TO REALITY
The voice portal vendors need to understand that while
they have a great interface, it's not always
appropriate. No one is going to shop for any
non-commodity products via speech rec. Nor are they
going to administer their stock portfolio or manage
their supply chain. They may purchase a stock or two,
or check the inventory levels of a given product, but
nothing very involved.
Speech recognition should be thought of as a shot
of bourbon, not a pint of stout. It's very potent and
very quick, not involved or drawn out. For
complicated, time consuming applications, or those
simply for pure enjoyment, the pint (Internet) is the
far better choice. By focusing their marketing, speech
rec and voice portal vendors will be far more
successful by pairing the application with its best
use. For a quick quaff, speech recognition is the
obvious choice. For a leisurely and enjoyable
beverage, the Internet still reigns supreme. Cheers!
Brian Strachman is senior analyst, Voice and
Data Communications, Cahners
In-Stat Group. To correspond with the author,
please send your comments to brians@instat.com.
[ Return
To The August 2001 Table Of Contents ] |