U of T professor Parham Aarabi is working on technology that would allow electronic devices to orient themselves in space using sound.

In his lab, Aarabi experiments with an array of microphones placed along one wall of the room. This array can orient itself in relation to a reference noise signal, and can identify the location of sound sources within a certain distance of the array. By analyzing the time it takes the signal to reach each of the microphones, the device can pinpoint the source’s location. This process is similar to that used by the human ear to tell the relative distance and direction towards sound sources, and is the main way blind people are able to orient themselves by constructing maps of rooms in their minds.

Aarabi, who works in U of T’s department of electrical and computer engineering, was originally interested in the problems associated with much of today’s computer-human interfaces. Keyboards and mice cannot easily be used by blind people and prove an inconvenience to portable devices like laptops and hand-helds. One possible solution is speech recognition, but this technology fails in noisy environments such as the subway or a loud restaurant. New research is focussed on filtering out background noise. The key is to localize each speaker so that their speech may be separated from most of the noise—something that can be done with as few as two microphones.

The main advantage of this localization technique over others is its passivity. While a system such as radar or active sonar needs to send out its own signals, microphones can be dropped in any location to listen passively to the sounds in the environment.

This convenience has implications in military settings where an array of microphones could be dropped into hostile territory to passively create a map of an unknown area for reconnaissance. Many researchers in sound navigation are funded by the military.

Aarabi believes the technology will become available commercially in cell phones and other electronic devices in five to 10 years. Used in phones, the technology could help reduce background noise, resulting in a cleaner and stronger signal.

The biggest obstacle to commercialization is the processing power such arrays demand. Aarabi’s super-fast, 2.4-gigahertz computer is only capable of processing data from five to six microphones in real time. This is because the microphones must continually analyze data in intervals as small as 10 milliseconds. The solution, according to Aarabi, is the creation of a chip that could be directly connected to the microphone array and do all the necessary calculations. This would accelerate the analytic process enough for such a technology to be built in to cellphones and other small electronic devices.

Photograph by Simon Turnbull