8.5 HYBRID WP AND ADAPTED WP/SINUSOIDAL ALGORITHMS

This section examines audio coding algorithms that make use of a hybrid wavelet packet/sinusoidal signal analysis. Hybrid coder architectures often improve coder robustness to diverse program material. In this case, the wavelet portion of a coder might be better suited to certain signal classes (e.g., transient), while the harmonic portion might be better suited to other classes of input signal (e.g., tonal or steady-state). In an effort to improve coder overall performance (e.g., better output quality for a given bit rate), several of the signal-adaptive wavelet and wavelet packet subband coding schemes presented in the previous section have been embedded in experimental hybrid coding schemes that seek to adapt the analysis properties of the coding algorithm to the signal content. Several examples are considered in this section.

Although the WP coder improvements reported in [Tewf93] addressed pre-echo control problems evident in [Sinh93b], they did not rectify the coder's inadequate performance for harmonic signals such as the piano test sequence. This is in part because the low-order FIR analysis filters typically employed in a WP decomposition are characterized by poor frequency selectivity, and therefore wavelet bases tend not to provide compact representations for strongly sinusoidal signals. On the other hand, wavelet decompositions provide some control over time resolution properties, leading to efficient representations ...

Get Audio Signal Processing and Coding now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.