摘要 |
<p>A reduced-complexity real-time audio coder/decoder ("codec") using adaptive sparse vector quantization (ASVQ) classifies signal vectors into on e of a plurality of different types. The codec provides low bit-rate compression fo r music and speech, while being applicable to higher bit-rate audio compression. It includes an in-path implementation of psychoacoustic spectral masking, and frequency- domain quantization using ASVQ and audio-compression algorithms. Critically-sampled subband filter banks maintain time-domain continuity across frame boundaries . In-path spectral masking usually results in sparse vectors. ASVQ is particularly effective for quantizing sparse signal vectors. In the preferred embodiment, ASVQ classifi es signal vectors into six different types (504, 508, 512, 516, 520, and 522). The ASV Q technique applies to general-purpose data quantization as well as to quantization for audio compression. The invention also includes a "soft-clipping" algorithm i n the decoder as a post-processing stage, which preserves the waveform shapes of t he reconstructed time-domain audio signal in a frame- or block-oriented statele ss manner while maintaining continuity across frame or block boundaries.</p> |