Research Project

A Computational Model of Elevation Localisation

Tommy Ashby
Russell Mason
Tim Brookes

A lot of work has already been undertaken on developing computational auditory models to predict the perceived location of sound sources in terms of azimuth, yet little has been done to predict the perceived elevation of a sound source. We have already undertaken research into developing elevation metrics based on spectral cues, and on combining the input from multiple pairs of microphones to simulate head movement. However, further work is required to further develop and combine these approaches, and to evaluate the practicality of these in real-world situations. The research will involve analysing measured head-related transfer functions of our Cortex head and torso simulator, and extracting interaural time and level difference cues, and spectral cues. These will be used to create either look-up table or polynomial equations of the important cues in order to develop a practical binaural model. The success of the model will be evaluated either by comparing measured results with those available from the literature, or by conducting custom subjective experiments.