基于DSP的数字音频处理系统设计及实现

发布时间：2018-06-21 13:59

本文选题：TLV320AIC23 + 语音算法　；参考：《成都理工大学》2013年硕士论文

【摘要】：音频在现实生活中广泛存在，对于音频的处理很是重要。目前，在很多语音处理系统中都用到了语音分析模块，采集现场的声音并进行频谱分析。语音处理系统的实时性、功耗、体积、以及对语音信号的保真度都是很影响系统性能的关键因素。因此，音频信号的分析器的设计是非常必要的。随着技术的进步，数字化技术已经深入人们的生活，数字化技术具有诸多模拟技术无可比拟的优势。传统的音频技术已经不能满足人们的要求，数字化的音频处理方式成为了音频技术的发展趋势。数字技术的快速发展，，在很多场合使得数字音频技术逐步取代模拟音频技术，数字化的音频处理是利用数字滤波算法对采集到的信号进行变换处理来实现的。采用数字化的技术，就必定会涉及到很多复杂的数字运算，而数字运算处理器（DSP）的出现和发展正是迎合了这一要求。DSP芯片的很多特殊结构可以快速实现各种数字信号处理算法。本文首先介绍了基于TMS320C5402DSP芯片的语音分析系统的工作原理，给出了整体设计方案和工作框图，然后给出了系统的硬件设计方案；接着介绍了基于TMS320C5402DSP芯片的语音录放系统的软件设计。本文以小波变换及多分辨分析为理论基础，对语音端点检测中小波系数方差算法和子带平均能量算法进行了分析和研究，利用语音和噪声的频域差别，对这两种算法进行了优化，并应用于端点检测系统中，有效地改善了小波系数方差算法耗时长、实时性差的缺点，并克服了子带平均能量算法只对高斯白噪声检测效果好的局限性，提高了语音端点检测系统的实用性。在整个设计过程中，我们采用了TLV320AIC23DSP芯片为核心音频录放接口器件，结合TMS320C5402DSP芯片，语音数据存储FLASH存储器等完成了数字音频处理系统的设计。软件部分则采用模块化的设计方法，用C语言来实现。该数字音频处理系统的设计具有以下特点： 1．音频数据占用资源少。 2．音质通信级。结合了音频编码与双向通信所需的低延迟两大优势，增强了话音编码与优质音频编码的效果与质量。 3．开发难度低,仅通过TMS320C5402DSP芯片和几个相关软件（MATLAB和仿真软件）即完成了相关设计。 4．语音芯片与DSP接口电路简单，利于生产和体积控制。可以降低成本，并且方便携带。
[Abstract]:Audio is widely used in real life and is very important to audio processing. At present, voice analysis module is used in many speech processing systems. The real-time, power consumption, volume and fidelity of speech signal are the key factors that affect the performance of speech processing system. Therefore, the design of audio signal analyzer is very necessary. With the progress of technology, digital technology has been deep into people's lives, digital technology has many unparalleled advantages of analog technology. Traditional audio technology can not meet the needs of people, digital audio processing has become the development trend of audio technology. With the rapid development of digital technology, digital audio technology is gradually replacing analog audio technology in many situations. Digital audio processing is realized by using digital filtering algorithm to transform the collected signals. Using digital technology will inevitably involve a lot of complex digital operations, and the emergence and development of digital computing processor (DSP) meets this requirement. Many special structures of DSP chip can quickly realize various digital signal processing algorithms. This paper first introduces the working principle of speech analysis system based on TMS320C5402 DSP chip, gives the overall design scheme and working block diagram, and then gives the hardware design scheme of the system. Then the software design of speech recording and playback system based on TMS320C 5402 DSP is introduced. On the basis of wavelet transform and multi-resolution analysis, this paper analyzes and studies the variance algorithm of wavelet coefficients and subband average energy algorithm for speech endpoint detection, and makes use of the difference between speech and noise in frequency domain. The two algorithms are optimized and applied to the endpoint detection system, which effectively improves the shortcomings of the wavelet coefficient variance algorithm, such as long time consumption and poor real-time performance. It overcomes the limitation that the sub-band average energy algorithm is only good for Gao Si white noise detection and improves the practicability of speech endpoint detection system. In the whole design process, we used TLV320AIC23 DSP chip as the core audio recording and playback interface device, combined with TMS320C5402 DSP chip, voice data storage flash memory to complete the design of the digital audio processing system. In the software part, the modular design method is adopted, and C language is used to realize it. The design of the digital audio processing system has the following characteristics: 1. Audio data occupy less resources. 2. Audio quality communication level. Combining the advantages of audio coding and bidirectional communication with low delay, the effect and quality of voice coding and high quality audio coding are enhanced. 3. Only the TMS320C5402 DSP chip and several related software MATLAB and simulation software) are used to complete the related design. 4. The interface circuit between speech chip and DSP is simple, which is conducive to production and volume control. Can reduce the cost, and easy to carry.
【学位授予单位】：成都理工大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TN912.3;TP368.1

【参考文献】