Clearer chit chat.

Navigation: Main page

Author: Kren, Lawrence

with cell-phone conversations could be a thing of the past, thanks to a dual-microphone system from researchers at the University of Toronto.

"The system employs two microphones that act as human ears to focus on a speaker's voice and filter out other noises," explains U of T researcher Parham Aarabi. Time-frequency filters determine a speaker of interest's location based on the length of time it takes for the most intense sound to arrive at the microphones. A computer chip continuously decides which frequencies belong to the speaker and which to the extraneous noise. The interference is then "damaged" and its volume scaled back.

Other speech-recognition systems only reduce the background noise. But the U of T technology also deconstructs extraneous conversations into a slight hum. "The approach boosts recognition accuracy 30% over alternative state-of-the-art, multimicrophone speech recognition systems," says Aarabi.

A miniaturized version of the system suitable for cell phones is about two years away, Aarabi predicts. A customized chip that enhances voice-recognition software in PCs is only months away. The miniaturized version will be pen sized and contain two or four microphones, batteries, and electronics.

~~~~~~~~

By Lawrence Kren