Michigan State University - Research Groups

Despite decades of bioengineering research to develop hearing aids and cochlear implant devices to help hearing-impaired patients restore their hearing ability and improve their lifestyle, measuring quality gains of cochlear implant and hearing aid patients is usually based on subjects' performance on structured speech recognition tasks in acoustically pristine environments. However, these patients persistently complain about the diminished hearing ability under severe adverse conditions. These conditions are mostly manifested in the presence of competing speakers, transient and persistent sources of environmental noise. The objective of this project is to develop novel speech processing strategies to optimize the information transfer from speech signals under severe adverse conditions without compromising issues of intelligibility and recognition accuracy. Our approach is based on adequate representation of the transients in the speech signals while simultaneously minimizing the masking effects of noise, as these are the most important factors affecting speech intelligibility.