A hearing aid for subjects with extreme high-frequency losses

A speech processing system has been developed in which the unvoiced portion of speech is bandwidth compressed from an original bandwidth of 4000 Hz into a low-frequency band not exceeding 1000 Hz, in which hearing impaired subjects with severe high-frequency hearing losses still possess some residual speech perception. The basic compression operation is based upon a time-domain time expansion technique, and the resulting reduction in bandwidth is accomplished without relinquishing the essential information contained in unvoiced speech. Thus, subjects are able again to perceive unvoiced speech of fair intelligibility where conventional hearing aids normally fail to be of any assistance. The imposition of stringent operating requirements such as portability, real-time operation, and functionality in a real-listening environment composed of many competing speech and noise sources, eliminated numerous elegant speech processing approaches.