A new approach for rendering Ambisonics recordings on loudspeakers in a reverberant environment
Abstract
It is desirable to reproduce a recorded sound using loudspeakers in a way that a listener has the impression of hearing the original sound. Theoretically, this is possible by wave-field synthesise (WFS) and Ambisonics. In both, the reproduction room is presumed to be an ideal anechoic room. But, in practice, the distortions related to the reverberation of the reproduction room and non- ideal responses of loudspeakers and position misalignments are inevitable. We propose a method that records, compensates and reproduces the direct and reverb parts of a signal separately. The recording is performed by a spherical microphone array and the spherical harmonic coefficients are extracted. After the beamforming which enhances the level of the direct sound, the reverb sound is extracted by subtraction of spatialized direct sound from the original spherical harmonic coefficients. The direct sound is reproduced by the vector- based amplitude panning (VBAP) method and the reverb sound is produced by 50 loudspeakers arranged on a Lebedev grid. Only two spherical harmonic coefficients are then used for the compensation and reproduction of the reverb signal in the playback room. The rendering of the direct and reverberant sound field is spectrally adjusted based on perceptually optimization of room-in- room impulse responses in each auditory frequency band. The direct and reverb parts are filtered in a way that the desired energy and Interaural Coherence (IC) of the recorded sound field is obtained for both the direct and the reverberant sound component. This perceptually motivated compensation provides a perceptually good rendering of the original recorded sound field and compensates for the acoustics of the reproduction environment.
Origin : Publisher files allowed on an open archive