|
|
George Fortier's new teaching assistant
doesn't have much of a life.
After all, he works for
free, never goes home, never gets tired, never takes a break and never
eats. In fact, Baldi, as this newest addition to Tucker-Maxon Oral
School in Portland is called, doesn't have much of a personality at all.
But in Fortier's
Classroon, Baldis an asset. The students love him. He's patient.
He never gets angry. He never complains. He's always amiable. And he's a
good listener.
Most Importantly, the
students listen to him.
Baldi is an animated ,
three dimensional talking head that actually speaks to the 8-to-11 year
old children, all of whom are hearing impaired, using easy-to-use spoken
language technology developed by the Oregon Graduate Institute for
Spoken Language Understanding, a division of the Department of Computer
Science and Engineering
The CSLU Toolkit, a
collection of software the combines voice recognition and text-to-speech
synthesis, has been meshed with the talking head developed at the
University of California, Santa Cruz.
Baldi can ask the
students questions as part of a word game, listen to the students'
answers, then tell them is their answers are correct. If incorrect
answers are given, Baldi works with the children if they get them right.
Best of all, Baldis
facial movements, including lips, teeth, and tongue, are extremely
articulate
|
|
- the can be exaggerated or toned down to
cater to different learning levels - allowing the students to read his
lips while he talks and further enhancing their language development
skills.
"It's
fantastic," says Fortier, who has been using the system in his
classroom for the last three months. "It's something we've always
dreamed about."
Fortier was among a
handful of Tucker-Maxon educator offered free training of the Toolkit by
CSLU last fall. CSLU received a $1.8 million grant from the National
Science foundation to create the Tucker-Maxon research program.
Intel Corp also donated
five top-of-the-line Pentium II computer platforms to the project.
For Fortier, the system
has given him an extra instructor - with all the time in the world.
"It's always an
issue, the time we can spend teaching something," Fortier says. The
students "can go over on their own and spend as much time as
necessary to sound that word out. Baldi's someone who's going to sit
there and do the repetitions with them."
In one lesson, the
children are shown a map of the land mass and are asked to name a
highlighted section.
"What land form is
this?" Baldi asks, his lips and his tongue moving in smooth tandem.
"Coastal
plain," one of the students responds.
At times, the response is
not articulate enough for Baldi, so he asks again and again until the
students get it right.
|
|
"If Baldi
doesn't' recognize a word, he takes them through a training loop,"
Fortier says. "He's very lip readable and very accurate."
Another computer in Fortier's
classroom is for speech development. It teaches students how to say
words by breaking them up into syllables, consonants and vowels. Long
words, such as "revolutionary," have some students stumped,
but Baldi's patience and ability to divide them into segments speeds the
learning process.
In the coming months,
Tucker-Maxon teachers expect CSLU and Santa Cruz researchers to improve
the timing of Baldi's speech, which is monotone and robotic, to allow
their own voices to be integrated with the technology.
Alice Davis, another
Tucker-Maxon teacher who uses Baldi and the Toolkit in her classroom,
says the technology has been a godsend.
"I've got stories on
there, I put math on there," Davis says, staring at the monitor.
" I put speech on there, I put vocabulary on there. I've got
constellations."
The Toolkit has allowed
Davis to integrate small comprehension quizzes into real children's
stories she input into the computer using the system's text-to-speech
components.
When the students answer
the questions correctly, Baldi moves onto the next "page" of
the story. Baldi is even helping Davis' students learn Haiku poetry.
"We're doing Japan
as an all-school theme," she says.
|