Ever since 1953 when Dr. Colin Cherry coined the term “the cocktail party problem” to describe the mystery as to how humans are able to distinguish one voice from another in a noisy, reverberant room using only two ears, scientists have been trying to puzzle it out, without success. The mystery may now have been solved.
Researchers from Wave Sciences published a paper at the recent International Congress on Acoustics (Gyeongju, Korea) in which we hypothesized that the approach underlying our patented GLIMPSE speech separation method is a computational model for human spatial hearing.
The title of the paper was “A Proposed Signal Processing Model Of Human Spatial Hearing Using Interaural Cross Correlation and Auditory Glimpsing To Estimate Green’s Functions.”