Our research explores, and attempts to exploit, the human being's remarkable ability to hear sound in 3-D using just two channels of sound; one to each ear. Activities focus on acoustical modelling of the head and ears, binaural audio signal processing and perceptual testing.
Efficient individualisation of binaural audio signals
Members: S Shoji and A I Tew
A variety of binaural signal processing algorithms are being developed which will facilitate the processes of lossy binaural audio compression and binaural signal individualisation. By means of perceptual tests we are seeking to identify characteristics of the hearing system that may be exploited in developing particularly efficient binaural audio systems. External funding.
Estimation of acoustic spatial cues (1)
Members: J B A Thorpe, C T Hetherington, C Jin*, A van Schaik* and A I Tew
HRTFs are largely determined by the shape of a person's head. They are traditionally measured acoustically, a process which is both time-consuming and requires specialist facilities. In principle, numerical acoustic modelling of the head using the boundary element method (BEM) is an alternative to direct acoustic measurement for producing good estimates of HRTFs. In many applications, however, the BEM imposes excessive computational demands. Using acceleration techniques developed by this group, this work is investigating the complex way in which acoustic spatial cues depend on the shape of a listener's head. Funded by EPSRC. *Uni. of Sydney
Estimation of acoustic spatial cues (2)
Members: C T Hetherington, J B A Thorpe, C Jin*, A van Schaik*, C D Hunter and A I Tew
Once we have established the links between physical human features and the acoustic cues we use to localise sounds, we will know which measurements of a subjects’ head and ear shape are important in creating individualised HRTFs from shape. These measurements will be converted into compact sets of numbers, unique to each individual. Techniques pioneered by our partners in the CARLab, University of Sydney, will then be employed to convert these data sets into the required personalised HRTFs, whose effectiveness will be evaluated by means of listening tests. Funded by EPSRC and ARC. *Uni. of Sydney
Sound source analysis and manipulation in complex binaural audio signals
Members: G A Spittle, J A Carmichael and A I Tew
inaural audio signals comprising more than one sound source or in a reverberant environment induce complex behaviours in the spatial cues. By determining the psychoacoustic significance of these phenomena and the perceptual effects of altering them we seek to develop binaural signal processing methods for improving the intelligibility of speech in acoustic environments where conventional hearing aids are known to perform less well. University and external funding.