The Optimal Algorithm Of Audio Decoding System And The Research Of Fixed-point On DSP

With the appearance of MCU, DSP, ARM and other embedded chips there, audio codec technology is developing rapidly. The current audio codec is developing with high compression ratio and high fidelity direction, the standards have emerged in a variety of formats one after another in the international, such as MPEG-1, AAC, AC-3 and so on.In numerous audio standard, Huffman decoding and IMDCT has always been an important part of the audio decoding, holds the high proportion in the decode procedure. This article has mainly researched the algorithm of Huffman fast table look-up and fast IMDCT processing in the audio decoding process. An optimal algorithm has been proposed based on the original algorithm, and the efficiency of look-up table and calculation has been raised.The main jobs of this thesis are follows:(1) The research status of audio decoding system and decoding algorithm has been analyzed. The performance of decoding systems and decoding algorithm has been compared separately in many literatures.(2)This paper has analyzed the psychological perception characteristics and masking effect of the encoding process. The performance of the sound masking has been verified by the simulation experiment. This paper makes a detailed analysis of the audio decoding system decoding process. It mainly studied the look-up table method of Huffman decoding algorithm and IMDCT process.(3)In this paper, Huffman direct querying method is proposed on the basis of rapid Huffman look-up table algorithm.According to the characteristics of Huffman code table, when the Huffman code table re-sorted in the code word size, according to the code to "0"at the beginning or " 1" at the beginning of the Huffman code table is divided into two categories. In decoding, the first step is to determine the start bit. If it is "0" to check the table at the beginning of "0", if it is "1" to check the table at the beginning of "1", then according to the length of code word the position in the table of Huffman code will be determined.Theoretical analysis shows that the proposed algorithm for fast lookup table look-up table of time than the original algorithm is reduced by half.(4)The algorithm of IMDCT processing with the calculation length of N has been improved.In the algorithms of IMDCT processing, based on the existing optimization algorithm N take a multiple of 4 is proposed on the basis of N take 8 multiples. N-point IMDCT is derived sequence of the formula. In this paper, the N point IMDCT processing into four N/8 point discrete cosine transform.According to the original algorithm for the corresponding formulas, improved optimization algorithm for calculating than the original algorithm reduces the cycle, the computation efficiency is improved.(5)This article is also studied the problem of the fixed-point to audio decoder based on TMS320C5416. In the decoding process, the steps of floating-point calculations which are dense will be fixed. Replacement parts will be replaced with fixed-point arithmetic, and carries on the simulation analysis after the fixed-point result. Simulation results show that the audio waveform is not very different after the fixed point, meets the acoustic fidelity requirement.