REAL-TIME TRANSPUTER SIMULATION OF THE HUMAN PERIPHERAL HEARING SYSTEM

This paper presents new work on the implementation of a contemporary m odel of the human peripheral hearing system which is designed to opera te in real-time. This is achieved by means of transputer technology. P sychoacoustic research is providing new insights into the operation of the human peripheral hearing system. It is suggested that many of the se could be employed in the analysis of speech and singing to understa nd further the nature of the acoustic cues presented to the brain by e ach ear which are the basis for human communication. Contemporary psyc hoacoustic experiments into the nature of the peripheral hearing mecha nism suggest that the fixed filter bandwidths used in the speech spect rograph could be obscuring patterns in the acoustic speech signal whic h are essential cues in our perception of speech and singing naturalne ss. To achieve a more appropriate model of the hearing system more sop histicated filters are required. A real-time version will provide a ba sis for visual displays of acoustic cues to be implemented for those w ith speech disorders and those wishing to develop healthy and more eff ective voices. The development of a real-time version of the model is not possible by means of standard sequential processing techniques, si nce the human peripheral hearing system is itself an inherently parall el process. Increased processing speed has been achieved by implementa tion using a network of processors which are themselves designed to op erate in parallel.