Approaches to Improve Quantization Performance Over the Scalable Advanced Audio Coder
In this paper, we present quantization techniques to improve the low rate performance of a scalable audio coder. We show that a conditional enhancement-layer quantizer is effective in exploiting the statistical dependence of the enhancement-layer signal on the base-layer quantization parameters. It fundamentally extends our prior work on compander domain scalability, which was shown to be asymptotically optimal in the context of entropy coded uniform scalar quantization, to systems with non-uniform base-layer quantization. Moreover, in the important case that the source is well modeled by the Laplacian density, we show that the optimal conditional quantizer is implementable with only two distinct switchable quantizers. Hence, major savings in bit rate are recouped at virtually no additional computational cost. Further improvement in performance is achieved at the expense of computational complexity when the proposed quantization scheme is incorporated within an efficient "trellis-based" search for the quantization parameters. For example, we implemented the proposed scalable coder within the MPEG-AAC framework with four 16kbps layers build using the MPEG-AAC framework and achieved performance approximately that of a 56kbps non-scalable coder on the standard test database of 44.1kHz audio.
Click to purchase paper 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, $5 for AES members and is free for E-Library subscribers.