Blind depth estimation based on primary-to-ambient energy ratio for 3-D acoustic depth rendering

Since the advent of 3-D video, the acoustic depth rendering for the proximity effect has been an issue of great interest. In this study, we propose an algorithm for estimating acoustic depth cues from stereo audio signal, without a priori knowledge about the source-to-listener geometry and room environments. We employ the principal component analysis (PCA) to estimate the acoustic depth based on the primary-to-ambient energy ratio (PAR) which is related with the front-back movement of the sound source. And for the acoustic depth rendering, the distance variation of the sound source is parameterized through tracking the estimated depth cue. The proposed estimation algorithm was evaluated using stereo audio clips extracted from a real 3-D movie, and the results confirmed effectiveness of the proposed acoustic depth estimation algorithm.