Nonlinear Distortion in Sound Zones & Beamforming-Based Listener-Position Adaptive Crosstalk Cancelation

Publikation: Bog/antologi/afhandling/rapportPh.d.-afhandlingForskning

This thesis consists of two separate but related topics, sound zone control (SZC) and crosstalk cancelation (CTC). SZC systems create multiple listening zones within a shared space with minimum interference between zones. Early research showed that the actual performance was often inferior compared to the simulation predict and audible nonlinear distortion existed. One of the main tasks of this PhD project is to assess the impact of loudspeaker driver nonlinear distortion on sound zones through both numerical simulations and experimental measurements. The results showed that though harmonic distortion was audible in the dark zone, it was not the main cause of the contrast loss, but the nonlinear effects on the fundamental component. To reduce the impact of nonlinear distortion, a constraint on the total input power can be imposed, however nonlinear distortion can still be distinct due to one or several loudspeaker drivers having high input power. To further eliminate the impact of nonlinearities, the individually control effort constrained acoustic contrast control was proposed for finer power control over each single loud- speaker driver. Experiments using both two-tone and wideband multi-tone signals showed a balance could be reached between the achievable acoustic contrast and nonlinear distortion reduction.
The other topic covered in this thesis is the loudspeaker array based CTC, which eliminates the acoustic crosstalk between the ears of a listener, allowing for separate control of signals arising at listener’s ears, so that binaural cues can be imposed. The proposed CTC is based on the superdirective near field beamforming covering the vast of the considered frequency range. When constraining the movement trajectory as a straight line parallel to the array, the beamformer weights at listening positions along the trajectory exhibit a smooth evolution, and can be parameterized by a few parameters for each frequency. Accordingly, a two-phase adaptive CTC is proposed, including an off-line parameterization phase, and an on-line updating phase. Simulations and experimental measurements on the proposed CTC showed higher channel separation obtained compared to non-superdirective beamforming approaches, and was robust with respect to head movements. A user study involving 20 subjects was conducted, showing that the users could locate all virtual sources in the horizontal plane with small localization errors.
OriginalsprogEngelsk
ForlagAarhus Universitet
Antal sider119
StatusUdgivet - sep. 2019

Se relationer på Aarhus Universitet Citationsformater

ID: 161426932