【作者】 胡多传；

【导师】 孔晓玲；

【作者基本信息】 安徽农业大学 ， 农业机械化工程， 2005， 硕士

【摘要】 MPEG-2 AAC 是目前最先进的数字音频压缩算法之一,在数字音频的存储和传输中有广泛的应用。与以往的 MPEG-1 伴音编码相比,该算法具有压缩质量更高,配置更为灵活等特点,但同时存在计算复杂度较高,实时性较差的特点,因此基于MPEG-2 AAC 的实时压缩技术的研究已经成为一个热点问题。 文章首先介绍了音频编码技术的发展情况和音频编解码技术的基本原理,对MPEG-2 AAC 标准的框架和各个模块的算法进行了研究和分析。在对 MPEG-2 AAC算法模型进行分析之后,对心理声学模型,滤波器组和量化编码模块等计算量较大的模块进行优化,同时也省略了部分运算量大但对编码效果增益不明显的几个模块,如时域预测和 TNS 等,并使用了一些快速计算技术对算法模型进行实时化改造。由于MPEG-2 AAC 编码的复杂度要远远大于解码,为了便于系统验证,本文对复杂度较低的解码部分,在 WINDOWS 平台下,用 VC++ 6.0 对 MPEG-2 AAC 的解码器进行软件实现;而对复杂度较高的编码部分,为了满足系统的实时性和性价比要求,本文采用数字信号处理器 TMS320C54x DSP 芯片作为编码主控芯片,成功在 DSP 上实现了,经过本文改造 MPEG-2 AAC(LC Profile)后的编码。 基于改造后的算法,基本上实现了 MPEG-2 AAC 的立体声软件编解码。经过反复调试和修改,试验结果表明,不仅运算速度大大提升,而且很好的保证了压缩后的音频质量,该算法是一种适合提高数字音频压缩算法实时性的有效算法。 论文对 MPEG-2 AAC 编解码系统实现过程中采用的一些重要的技术进行了阐述,包括 AAC 算法的定点化、查表法的应用以及所采用的 C 程序和汇编程序内嵌调用、使用 DMA 等程序优化技术。正是由于 MPEG-2 AAC 编码是一种复杂的运算,它需要使用大量的数据和程序存储空间。如何尽量压缩它的数据和程序占用的存储空间,以使编码能在 DSP 上实现,成为一个必须考虑的问题。一个良好的程序结构对节省数据存储空间和加快算法运算速度也有着极其重要的作用。因此,为了提高编解码的实时性,必须采用查表法、混合法和使用 DMA 等方法,以提高对指数和对数等非线性计算过程的计算速度。更多还原

【Abstract】 As one of the most advanced standards of digital audio compression, MPEG-2 AAC has been applied widely in the transmission and storage of digital audio signal. Compared with MPEG-1 audio coding, it has better audio compression quality and more flexible configuration beside it’s complicated computation and bad real time feature. Therefore, research on real time compression technology base on MPEG-2 AAC has become a hot research area. The thesis introduces the development and the fundamental principle of audio coding technology. It studies and analyzes the standard function blocks of MPEG-2 AAC and the algorithm of each function block. After the analysis of the algorithms model of MPEG-2 AAC, the thesis optimizes some blocks with great computation of Psychoacoustic Model, Filter banks and Quantization coding, etc. It omits some function blocks which have much computation and avail little to the gain of coding, for example, prediction in time domain, TNS, etc. It also adopts some fast algorithms to reconstruct some model for the reason of real time characteristic. Because the complexity of encoding is greater than decoding as for MPEG-2 AAC, the thesis uses VC++6.0 to implement a software decoder of MPEG-2 AAC LC profile to testify the system. As for the encoding part with greater complexity, taking the system’s real time feature and the demand of performance-price ratio into accounted, the thesis adopts digital signal processor TMS320C54x DSP as main control chip to implement the optimized coding algorithms of MPEG-2 AAC LC profile. The optimized algorithms implement the encoding and decoding of MPEG-2 AAC stereo audio basically. After repeatedly debugging and modification, experimentation result shows that not only the computing speed is improved but also the quality of compressed audio is well ensured. The algorithm is an effective one which is suitable for improve the real time feature of the audio compression algorithms. The thesis gives details to some important techniques used in the process of implementation of the MPEG-2 AAC encoding and decoding system, including point fixing of AAC, application of table consultation, compiling language embeds in C language programme and DMA using, etc. Because MPEG-2 AAC coding involves complicated computation, it needs a large storage space of data and programme. We have to pay attention to the problem how to reduce the storage space of the data and programme so as to implement it on a DSP. A good programme structure is significant in saving data storage space and speeding up the computation. Therefore, in order to improve the real time feature of encoding and decoding, we have to adopt these techniques to speed up the computation of some nonlinear process, such as exponent and logarithm.更多还原