The MPEG-H 3D Audio standard applies singular value decomposition (SVD) to higher-order ambisonics data, and divides the outcome into prominent and ambient sound components, which are then separately encoded. We recently showed that significant compression gains are achievable by moving the SVD to the frequency domain, and ensuring smooth transition between frames. Frequency domain SVD also enables SVD adaptation to frequency, but the increase in side information, to specify additional basis vectors, compromises the gains. This paper overcomes this shortcoming by introducing backward adaptive estimation of SVD basis vectors, at no cost in side information, thereby approaching the full potential of frequency domain SVD. Objective and subjective tests show considerable gains that validate the effectiveness of the proposed approach.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.