Machines Learn to Mimic Speech
From: Wired News - 10/03/2003
By: Michelle Delio

Attendees at the SpeechTek tradeshow said speech technology companies have
started to take a more realistic view in realizing that voice technology has
not yet reached the point where computers can actually understand human
speech. "Now that the magic is gone, we don't believe in using speech
technology unless it serves a viable purpose--making it easier for people to
work with a computer system, making systems more secure or even making
computers more fun," remarked speech application programmer Frank Vertram.
Still, SpeechTek showcased some impressive products: One ATM product was
designed to aid visually handicapped or technology-evasive users by allowing
them to hear descriptions of onscreen options through headphones. Nuance
displayed a "say anything" natural language application that employs a
database to interpret users' intent from "freestyle conversations." Cepstral
unveiled two sets of computer voices, one geared for the American market and
the other for the Canadian market--the American voices are imbued with a
casual tone, while the Canadian voices speak with a French-Canadian accent.
IBM highlighted WebSphere speech offerings upgraded with VoiceXML 2.0
support, which allows speech technology to be embedded within Web sites.
SpeechTek's Speech Solutions challenge, which was set up to prove that
programming speech applications does not necessarily have to be a frustrating
experience, tasked seven teams with developing a workable application capable
of identifying car trouble and scheduling a session at a repair shop by the
end of the day; all seven teams met the challenge by 5:00 p.m. "Once we get
past the mistaken idea that computers should be able to really understand us
or that we can engage in meaningful conversations with machines, the new
voice and speech technology is absolutely amazing," declared SpeechTek
organizer James Larson. 

http://www.wired.com/news/technology/0,1282,60677,00.html

