This paper proposes a novel frame loss concealment technique for higher order ambisonics (HOA) audio signals. It is designed to overcome the challenge of interpolating lost frames of HOA data and recover a close approximation of the original data without significantly impacting it’s localization. The underlying idea uses two techniques. The first is cascaded long term prediction, a technique which uses a cascade of long-term prediction filters to capture periodic components of music signals, to predict the lost frame’s ambisonics channels, in the SVD domain, from the periodic components of the past and future frames. Additionally, spherical linear interpolation of the SVD basis vectors is used to accurately reconstruct the spatialization of the lost frame. Objective and subjective evaluations show this method to be superior in accurately reconstructing lost frames to cascaded long term prediction being applied directly to the ambisonics signal.
(See document for exact affiliation information.)
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 Join the AES. If you need to check your member status, login to the Member Portal.
