广东工业大学学报 ›› 2013, Vol. 30 ›› Issue (2): 95-98.doi: 10.3969/j.issn.1007-7162.2013.02.018

• 综合研究 • 上一篇    下一篇

基于ARM的孤立语音识别系统的研究

张永刚1,2,余玉平3   

  1. 1. 住房和城乡建设部IC卡应用服务中心,2.中外建设信息有限责任公司,北京 100081;3. 嘉兴学院 机电工程学院, 浙江 嘉兴 314001
  • 收稿日期:2013-01-25 出版日期:2013-06-27 发布日期:2013-06-27
  • 作者简介:张永刚(1972-),男,高级工程师,主要研究方向为数字城市、智慧城市信息技术及标准.

The Design of the Isolated Speech Recognition System Based on ARM

Zhang Yong-gang1,2, Yu Yu-ping3   

  1. 1.Service Centre for IC Application of Ministry of Housing and UrbanRural Development, 2.Sinooverseas Construction Information CO.〖KG-*5〗, Ltd,Beijing 100081, China; 3.Mechanical and Electrical Engineering College, Jiaxing University, Jiaxing 314001 China
  • Received:2013-01-25 Online:2013-06-27 Published:2013-06-27

摘要: 提出了基于ARM(Advanced RISC Machines)的孤立语音识别系统的设计,并对实验系统进行了制作.本设计是基于广州友善之臂公司提供的ARM开发板进行开发,在硬件上核心板采用ARM微处理器并烧入Linux-2.6.32.2内核,声音的A/D采样和输出采用由UDA1341芯片及其外围电路组成的音频解码系统;在软件上采用MFCC算法进行特征提取,使用HMM算法进行模式匹配;最后利用科大讯飞语音识别、合成软件对目标语音进行识别并输出识别结果.本设计可广泛运用在机器人导游、智能家居以及汽车声控等系统中.

关键词: ARM微处理器;语音识别;HMM算法

Abstract: It proposes the design of the isolated speech recognition system, based on ARM, and then experiments with it. On the basis of ARM development board provided by Guangzhou Friendly Arm Corporation, in the design ARM Linux-2.6.32.2 has been adopted in the core board and Linux -2.6.32.2 in kernel to meet its hardware requirement. The A/D sampling and output of the voice was realized through the audio decoding system consisting of a UDA1341 chip and its peripheral circuit. MFCC algorithm was used to extract the feathers while HMM algorithm was used to match the patterns. Finally with the help of speech recognition software and speech synthesis software designed by IFLYTEK Corporation, the targeted speech was recognized and the result was put out. This design can find wide application in different systems, such as robot guide systems, intelligent home systems and the sound controlling system in cars.

Key words: ARM(Advanced RISC Machines); speech recognition; HMM algorithm

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!