A progressive algorithm for perceptual coding of digital audio signals

Abstract

In this paper, a progressive audio coding algorithm is presented. Distinctive from existing transform-based perceptual audio coding schemes, which allow either constant bit rate or multiple fixed bit rates, our proposed algorithm achieves a fully embedded audio coding scheme whose bit rate can be controlled depending on the user's requirement. This feature makes it extremely useful in audio networking applications, such as audio on-demand and audio broadcasting over the Internet. In the proposed audio coding scheme, efficient successive quantization and bit layer coding are applied to the compact audio signal representation obtained from traditional subband decomposition. Our research focuses on the improved successive approximation quantisation (ISAQ), which performs bit allocation for quantization based on psychoacoustic analysis of the input audio signal. Thus the progressive audio coding is achieved, while perceptually transparent audio coding is also maintained to the allowance of bit rate constraint. The performance measurement of our presented algorithm is demonstrated.