College Media Network

Mute man looks to find voice from BU research

Borana Greku

Print this article

Published: Friday, December 5, 2008

Updated: Friday, December 5, 2008

Erik Ramsey has not been able to speak since he was involved in a car accident at 16 that resulted in a brain-stem stroke, leaving him with locked-in syndrome and only able to move his eyes.

Now, nine years later, Ramsey may have new hope for regaining his speech, thanks to Boston University cognitive and neural systems professor Frank Guenther, who is researching a way to create sound through thoughts.

As part of the research, which Guenther started in 1992, a volunteer produces specific sounds while Guenther records what is happening in the volunteer’s brain, or the volunteer’s neural signals. A computer decoder then translates these neural signals into speech sounds.

Jonathan Brumberg, a research associate who joined Guenther in 2006, explained that the model essentially “tries to understand what’s going on in the human brain while we’re trying to speak,” and then uses that information to create sound.

Brumberg said they started to work with Ramsey, the stroke victim, after Neural Signals, Inc. Director Philip Kennedy approached Guenther for help. Kennedy had invented an electrode that was implanted in Ramsey’s brain in December 2004 to help restore speech.

Guenther said their findings have shown that they may be able to decode auditory information about “the sounds that are supposed to come out.”

“If we decode this auditory info and send it through a speech synthesizer so the person can hear it himself right away as he thinks it, he’s able to practice with the system and improve his production,” he said.

Brumberg said they had been working on this system for a little more than a year. Over that time, they have been able to create a device that synthesizes speech quickly, making it potentially usable in real time.

“That meant that he could think of saying some sounds . . . and in under 50 milliseconds, the computer would produce a speech-like sound that was directly controlled by him.” he said.

The first prototype shows that it works for producing vowels, but Guenther and his colleagues are working on improvements that will allow him to generate consonants and, eventually, full words, he said. 

“Our five-year goal is that the patient we’re working with now will be able to produce words using the system,” he said.

Guenther said his team has presented its results at conferences over the past year and submitted papers to publications this past week.

“I’m definitely lucky to be working on exciting work like this,” Brumberg said. “It feels very good to actually take what I’ve been able to learn here and actually use it in an application and help someone who is paralyzed.”

Cognitive and neural systems department Director Stephen Grossberg said he has known Guenther for a long time and supported him when he was a young investigator.

“I assume that he’s continuing the work, because not only it’s medically valuable, but also because he hopes that it can really help these patients,” Grossberg said.

Comments

Be the first to comment on this article!





Verify you are human: