语音识别开发平台的设计与实现

发布时间：2018-01-16 15:18

本文关键词：语音识别开发平台的设计与实现　出处：《哈尔滨工业大学》2014年硕士论文　论文类型：学位论文

【摘要】：随着计算机技术的发展,电子产品的功能日渐强大,语音识别技术已经开始慢慢的走进我们的生活,只是具体应用仍然不够成熟。语音识别系统多在PC机上实现,但是因为PC机体积大,价格高,携带不便等原因并不适合应用到普通生活中。嵌入式系统易于裁剪,便于内嵌到生活中常用设备中,如果将语音识别在其平台上实现,将带新的生活体验,并且在特定领域能实现特别的效果,比如在较为危险的工业控制中,语音控制能解放人的双手,通过没有实体接触的说话声去控制可以避免危险操作;对于手脚不便的残疾人或者年迈的老人,语音控制能给他们生活极大的方便。本文的目的就是在嵌入式系统上,以最精简的软硬件系统去实现语音识别功能,这样能够便于后期将语音识别系统应用到生活设备的控制上。设计的语音识别类型属于孤立词汇识别,在设定的时间内以正常语速说话,系统检测到语音并识别语音,下一段时间系统开始检测第二句话。另外,非特定人说话,只要是标准的普通话,系统也能正常的识别出来。论文主体设计了以语音识别专用So C为基础设计的验证平台。论文分析了语音识别研究的背景和意义,语音识别算法以及本文的课题所采用的语音识别算法系统设计。然后论文以专用So C为核心,完成了整体的硬件电路设计,包括电源模块设计,存储模块的设计,输入输出模块设计,音频编解码电路设计以及用于调试的JTAG和串口等等各个功能模块。论文最后是电路板设计和功能调试。电路板设计主要涉及到电路板层数选择和元器件布局布线,论文对此详述了相关设计的基本原则及其改进事宜。最终实现了电路预期的功能。这里采用的语音验证方式是系统识别到语音结果后,将词条在词汇表中的位置显示出来,表明识别正确。另外,So C设计和硬件电路设计时预留了通用接口,便于语音识别的应用扩展。
[Abstract]:With the development of computer technology, the function of electronic products is becoming more and more powerful. Speech recognition technology has begun to come into our lives slowly. But the concrete application is still not mature. Most of the speech recognition systems are realized on PC, but because of the large size of PC, the price is high. The embedded system is easy to cut, easy to embed into the common equipment in daily life. If the speech recognition is realized on its platform, it will bring a new life experience. And can achieve special effects in specific areas, such as in the more dangerous industrial control, voice control can liberate people's hands, through the non-physical contact of the speech to control can avoid dangerous operations; For the handicapped or the elderly, voice control can give them a great convenience in life. The purpose of this paper is to achieve speech recognition in embedded system with the simplest software and hardware system. This will facilitate the later application of the speech recognition system to the control of living equipment. The design of speech recognition type belongs to isolated vocabulary recognition, speaking at a normal speed within a given time. The system detects speech and recognizes speech, and the next time the system begins to detect the second sentence. In addition, independent people speak as long as they are standard Mandarin. The system can also be recognized normally. The main body of the paper designs a verification platform based on the special so C for speech recognition. The background and significance of the research on speech recognition are analyzed in this paper. The design of speech recognition algorithm and the speech recognition algorithm system used in this paper. Then, the paper completed the whole hardware circuit design, including the power module design, with the special so C as the core. Storage module design, input and output module design. The design of audio codec circuit and the function modules of JTAG and serial port for debugging. Finally, the design and debugging of circuit board. The design of circuit board mainly involves the selection of circuit board layer number and the layout of components. Wiring. In this paper, the basic principles of the related design and its improvement are described in detail. Finally, the expected function of the circuit is realized. The speech verification method adopted here is after the system recognizes the speech result. The position of the entries in the glossary shows that the recognition is correct. In addition, a general interface is reserved for the design of so C and hardware circuits, which is convenient for the application of speech recognition.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN912.34

【相似文献】