Per Rubak

Objective evaluation of the sweet spot size in spatial sound reproduction using elevated loudspeakers.

In a previous study, three crosstalk cancellation techniques were evaluated and compared under different conditions. Least-squares approximations in the frequency and time domain were evaluated along with a method based on minimum-phase decomposition and a frequency independent delay. In general, the least-squares methods outperformed the method based on the minimum-phase decomposition. However, the evaluation was only done for the best-case scenario, where the transfer functions used to design the filters correspond to the listeners transfer functions and his/her location and orientation relative to the loudspeakers. This paper presents a follow-up evaluation of the performance of the three inversion techniques when the above mentioned conditions are relaxed. A setup to measure the sweet spot of different loudspeaker arrangements is described. The sweet spot was measured for 21 different loudspeaker configurations, including two- and four-channel setups. Lateral and frontal displacement were measured along with head rotations. The setups were evaluated at different elevation angles. The results suggest that when the loudspeakers are placed at elevated positions, a wider effective area is obtained. Additionally, the two-channel configurations showed to be more robust to head misalignments than the four-channel configurations.

Analysis of ill-conditioning of multi-channel deconvolution problems

Deconvolution of single- and multichannel systems is often an ill-conditioned problem whose exact solution boosts certain frequency bands excessively. Frequency-dependent regularisation can used to prevent this by attenuating sharp peaks in the magnitude response of the optimal filters. A z-domain analysis demonstrates that frequency-dependent regularisation works by pushing the poles of an ideal optimal solution away from the unit circle.

Capturing blocked-entrance binaural signals from open-entrance recordings

Binaural recordings enable us to capture all sound attributes including spatial information, room effect, and source characteristics in a given environment. It has been shown that blocked‐entrance binaural recordings provide advantages over open‐entrance recordings, primarily because the blocked‐entrance recordings is not influenced by the ear canal acoustics of the individual for which it is recorded. However, blocking the ear canal for recoding imposes an obvious disruption to normal hearing conditions, which may be unacceptable for applications in which binaural audio capturing is desired but without interfering the individuals hearing and doing. In this work we propose a strategy for the recording of binaural audio with minimal hearing interference, and for transforming these recordings to blocked‐entrance versions that are more suitable for analysis and reproduction of binaural audio in a more general context. To this purpose, equalization filters are derived from the ratio between blocked and open ...