Speech Is Helping to Identify Concussions

Indianapolis-based Waveform Communications is working on a speech recognition system that will be able to detect concussions based on how a person who suffers a brain injury produces vowel sounds.

The technology is based on Waveform's Elbow speech recognition engine and the Waveform Model of Vowel Perception and Production, which recognizes patterns in how people produce seven key vowel sounds. It would compare current speech with previous recordings.

Michael Stokes, president and CEO of Waveform Communication, says concussions affect motor control of muscles, including the lips and tongue, which directly relates to how people form vowel sounds.

"We can definitely identify concussions from speech patterns," Stokes says.

Codenamed Cobweb, the product is being developed with a recent award from the Methodist Sports Medicine Education and Research Foundation. Stokes says Cobweb is currently in the pilot phase and should be commercially available within six to 12 months. "The initial analysis shows that is worthy of pursuing further," he says. "The engine is already there, it's just a matter of adding the comparative analysis engine."

"Years of diligence are paying off right now," Stokes says.

Though the product could have many applications, Stokes says its best use case might be on the sidelines during sporting events. The application and player voiceprints could be stored on a laptop and then used when a player suffers an injury that could result in a concussion.

"A coach could store all his players' speech at the start of the season, record the speech when there's an injury, and compare the two," he says. "You'd be able to do this in real-time on the sidelines during a game to determine whether to send a player back out [on the field]."

The system, Stokes says, would require stored voiceprints for comparison because there is not a standard voice form that could be used across the board. And especially in youth sports, such as Little League, it would be a good idea to record new voiceprints every year to account for changes in the voice due to growth or puberty, for example.

The system, Stokes adds, has a number of benefits. "It's a very accurate, conclusive test that is non-intrusive," he states, "and it's relatively inexpensive to do."

And, "with speech technology, you can't fake it just to get back out on the field," Stokes adds.

The Cobweb system currently runs on a laptop with a microphone and headset, but Stokes envisions it as a smartphone app as well.

He also sees the patent-pending Waveform Model being used not only in concussion research but also other areas of healthcare, including medical transcription and hearing aid technology. It's could also be used to improve the overall accuracy of speech recognition technology.

"We're at a critical juncture right now where the technology available is ready to manifest the long-desired possibilities in ASR and the healthcare marketplace," Stokes boasts. "It's a very exciting time for our company."

SpeechTek Covers
Free
for qualified subscribers
Subscribe Now Current Issue Past Issues
Related Articles

Waveform Compiles Speech Database for Concussion Diagnoses

Waveform is working with Methodist Sports Medicine on speech technologies to help diagnose concussions.