Scalar Quantization for Audio Data Coding

TL;DR

This paper introduces an adaptive Extended Zero Zone scalar quantizer for audio transform coefficients, demonstrating near-optimal coding efficiency with low side information bitrate, based on modeling coefficients with a generalized Gaussian distribution.

Contribution

It proposes a new adaptive EZZ scalar quantizer for audio data that effectively models transform coefficients using GGD, improving coding efficiency.

Findings

EZZ quantizer achieves near-optimal rate-distortion performance.

Adaptive EZZ works efficiently at low bitrate for side information.

Simulation confirms negligible loss compared to optimal scalar quantizers.

Abstract

This paper is concerned with scalar quantization of transform coefficients in an audio codec. The generalized Gaussian distribution (GGD) is used as an approximation of one-dimensional probability density function for transform coefficients obtained by modulated lapped transform (MLT) or modified cosine transform (MDCT) filterbank. The rationale of the model is provided in comparison with theoretically achievable rate-distortion function. The rate-distortion function computed for the random sequence obtained from a real sequence of samples from a large database is compared with that computed for random sequence obtained by a GGD random generator. A simple algorithm of constructing the Extended Zero Zone (EZZ) quantizer is proposed. Simulation results show that the EZZ quantizer yields a negligible loss in terms of coding efficiency compared to optimal scalar quantizers. Furthermore, we describe an adaptive version of the EZZ quantizer which works efficiently with low bitrate requirements for transmitting side information