Key Technology Research On Scalable Audio Coding

Scalable audio coding is a coding scheme which put forward to ensure the quality of audio transmission in variable bandwidth channel,in view of the current rapid development of digital audio communication.With the coming of the era of big data,and the increasing demand for audio-visual quality,obviously,it's unrealistic that only rely on the capacity of the memory and the bandwidth of the transmission network to ensure the quality of a transmission service,which in terms of cost control.In audio communications,we insist that audio quality can not only adapt to the changing bandwidth of network in the process of transmission,but also expect the audio signals can be arbitrary truncated according to variable rate in the unstable channel to better adapt to the channel conditions.Therefore,on the basis of perceptual audio coding,we put forward the idea of scalable audio coding.MPEG-4 audio standard supports a range of different techniques to promote scalable audio coding,the basic idea is: a scalable perception audio encoder produces a bit stream which contains many audio fidelity layers,then with specific encoding to reach a target that when adds a fidelity layer can have some obvious improved on auditory quality.The layer that guarantee the quality of the basic hearing is known as core layer,and each fidelity layer called enhancement layer.This paper analyzes the existing general audio coding models of MPEG-4,in view of their low coding quality in low bit rate(less than 16 Kbits/s/channel),therefore imports transform domain weight interleave vector quantization(TWIN-VQ)as a research starting point,which has a better performance in low bit rate.Then take advantage of the auditory characteristics of the psychoacoustics model to design the scalable audio coding perceptively,and in addition puts forward a scheme called Frame-Adaptive Quantization Bit-allocation(F-AQB)based on audio frame features.Finally,creates a new type of fine granularity scalable codecs.The simulation that performed for the designed codec is based on the MATLAB platform,then evaluates the reconstructed audio from subjective and objective aspects respectively,and compared with the TWIN-VQ codec without F-AQB scheme and the MPEG AAC-BSAC codec.Results show that the new scalable codec has a better performance of compression and reconstruction in low bit rate for the audio signals that have a variety of characteristics of different frames,also the waveform of decoding audio signals have smaller distortion and more clear audio quality.