Cocktail party acoustics: How humans perceive sound in noisy and complex environments

The team is testing how humans track sound patterns over time, and under what circumstances the brain registers that the pattern has been broken. The researchers will present preliminary findings at the Acoustics 2012 meeting in Hong Kong, May 13-18, a joint meeting of the Acoustical Society of America (ASA), Acoustical Society of China, Western Pacific Acoustics Conference, and the Hong Kong Institute of Acoustics.

"When a person hears a sound, both what we call 'bottom-up' and 'top-down' processes occur in the brain," says Mounya Elhilali, an assistant professor at the Center for Language and Speech Processing at John Hopkins. Hearing a whole range of sounds in a room is a bottom-up process, but choosing to pay attention to a particular voice would be an example of a top-down process, Elhilali explains. "We try to understand the interaction of these two processes," she says.

Elhilali and her colleagues ask volunteers to listen to a series of sounds and press a button when they hear something unusual. For example, the researchers may start out playing violin music and then introduce the sound of a piano. The change to piano music represents a change in the timbre, or sound quality. The researchers also changed the pitch (going from low to high notes or vice versa), and the loudness. As expected, results indicate that humans will perceive these changes as salient sound events that grab their attention.

Forming an expectation about the timbre, pitch, and loudness of sounds, and then realizing that the expectation has been broken generally takes a few seconds, says Elhilali, although the scientists are still in the process of fully analyzing their data. With further analysis the researchers also hope to glean information about how the different expectations interact and what happens when multiple changes, for example loudness and pitch, are made at the same time.

Eventually the team would like to repeat the experiments while monitoring the volunteers' brain waves through sensors placed on the skin. This would offer the scientists a glimpse at the neural changes that take place in the brain as the sound scene changes.

The ultimate aim is to understand how the brain adapts to different acoustic environments, says Elhilali. Engineers might be able to use the knowledge to design better hearing aids, voice recognition software, and recording equipment. Elhilali's group, which is based out of the Electrical and Computer Engineering Department, also works on such technological applications of the research. "When a person walks into a room, they first gather information that will help them adjust to the acoustic scene," says Elhilali. One day hearing aids and other human-designed sound processing technologies may be just as adaptive as the human brain.