Publication or Conference Title:Ph.D. Thesis, McGill University
Acoustic models driven by real-time velocity signals can suffer unduly from quality issues due to sampling and differentiation, especially at high sampling rates. In audio-haptic friction interaction, as found in a bowed string simulation for example, this noise appears as a gritty or dry feel, and is audible in the sound.
In this thesis, two approaches to this problem are proposed: firstly, reduction of the sensitivity of the model to velocity noise by the application of a position-dependent friction model; secondly, the improvement of velocity estimation by means of filtering and enhanced sensing.
Several estimators are compared, by means of parameter optimisation, to direct velocity measurement in order to find a good trade-off between filter-imposed delay and noise rejection. Optimised estimators are then compared by subjects in an online scenario to test their respective effect on the impedance range and noise qualities of a bowed string friction display.