Timbral and spatial fidelity improvement in ambisonics

Ambisonics renders a sound field through different kinds of loudspeaker layouts, which leads to different listening perceptions. While some loudspeaker arrays reinforce timbral fidelity, some improve localization accuracy. A split-band decoding is proposed that aims to select and then mix the better reconstructed frequency components from different loudspeaker arrays, thereby achieving the improved quality. The spectral reconstruction errors caused by truncation, comb filtering, and low-pass filtering are illustrated. The proposed solution is described, along with the experimental results from the listening tests. The split-band decoding method is especially suitable for binaural rendering and can also be applied to conventional loudspeaker arrays.