The Research And Optimization On The Algorithm Of The Key Module Of Audio Processing

With the development of advanced electronic science and technology, consumerelectronics, communication network, broadcasting and computer technology havebeen being combined increasingly and closely, and the world has entered thehigh-tech information age of digital network. In the background of mutual integrationof computer technology communication network and entertainment, the digitalmultimedia technology has been on fast-growing. Multimedia technology influencespeople’s traditional way of life in many respects. Digital audio processing technologyis an important part of the technology. The algorithm design and optimization of thekey module of audio processing has become an important research field.The main research work includes an analysis filter bank module, the MDCTtransformation, quantization module, Huffman module and audio sampling rateconversion.Several algorithms on the analysis filter bank include the polyphase filteralgorithm, the basic fast algorithm, algorithms introduced by shine and Zhang Zhiyanand FCT algorithm, among which the FCT algorithm performs with the bestefficiency. So the FCT algorithm is introduced in the application of the analysis filterbank.MDCT transformation is the discrete cosine transform, which transforms theaudio signal from the time domain to the frequency domain. The adjacent audio signalin the time domain has correlation to some extent, which, through the MDCTtransform, can be eliminated. The MDCT transform is optimized in the paper, on onehand, MDCT algorithm is optimized to reduce the number of multiplication andaddition; on the other hand, the assembly language instruction is scheduled to reducethe instruction execution cycles.Quantization performs bit allocation for the output data of the MDCT transform to encode the signal with fewer bits, which belongs to lossy compression. Theoptimization of the quantization module is to convert the initial value of thequantization step and to change the amplitude of the quantization step.In the Huffman module, the audio decoding process is divided into the sectionsof software and hardware, and the co-design of hardware and software implementapproach fulfills the larger part of calculation task by hardware, and the smaller partby software.In the situation that the sampling rate of the audio signal is not fit for thestandard, the signal should be resampled. Audio sampling rate conversion is tointerpolate the original signal, which will be convoluted with the filter coefficients,and finally extracted in order to obtain the signal with sampling rate fit for thestandard.In this paper, the algorithm of the key module of audio processing is researchedand optimized, so that the efficiency of audio processing is greatly improved and thefunction of audio processing is more perfect.