9402246 Duda This is the first year award of a three-year project to investigate the creation and evaluation of a process model of how humans locate sounds in three dimensions. Emphasis is placed in understanding well- established but yet incompletely understood psycho-acoustical phenomena. An example of such phenomena is the human perception capability of determining whether sounds come from above or below without binaural differentiation of the sounds reaching the ears; another example is the ability to locate sounds in reverberant environments with multiple sound sources where echoes and reflections become virtual sound sources. Another one yet is the ability to judge distance to a sound source since loudness alone is not an adequate cue. This model is based not only on physical sound propagation effects but also on neurophysiological studies tracing the auditory pathways from the cochlea to the auditory cortex. The model extends current computational cochlea models by incorporating mono- and binaural, temporally-based, correlation methods for source-locality information extraction. The ability to understand these mechanisms allows the recognition of sounds and speech by computers in everyday environments, thereby improving the effectiveness of speech recognition in human-computer interaction.
