智能家居语音控制系统的设计  

The Design of Intelligent Home voice Control System

作　　者：汪晟磊 宋星[1] 杨彦青[1] WANG Shenglei;SONG Xing;YANG Yanqing(Taizhou Vocational and Technical College,Taizhou Zhejiang 318000,China)

机构地区：[1]台州职业技术学院,浙江台州318000

出　　处：《自动化与仪器仪表》2023年第4期117-122,128,共7页Automation & Instrumentation

基　　金：2022年台州职业技术学院校级大学生科技创新项目《智能家居语音控制系统的设计》(2022DKC33)。

摘　　要：针对传统智能家居系统在智能终端控制中存在智能化和人性化水平低的问题,提出设计一个基于语音识别的智能家居控制系统。该系统主要由智能终端、主控中心和控制节点组成。对主控中心和控制节点的软硬件方案进行设计后,即可采用系统中的图像采集模块采集家居数据;然后通过改进信号子空间与维纳滤波的两级降噪方法进行语音信号增强;之后选用24维梅尔倒谱系数对语音特征进行提取;最后采用隐马尔可夫模型HMM算法进行模板训练和模式匹配,最终实现智能家居语音自动控制。实验结果表明,在800个测试样本中,共有789个样本被正确识别,平均识别率为98.6%。且在5种不同的信噪比下,语音识别率均保持在94%及以上,最高可达97.4%。由此说明本系统具备较好的抗噪能力,提出的语音识别算法对满足系统语音自动化和智能化需求,在实际产品应用中具有重要意义。For the traditional intelligent home system in the intelligent terminal control problem,a design of an intelligent home control system based on speech recognition is proposed.The system is mainly composed of intelligent terminal,master control center and control node.After designing the hardware and software scheme of the main control center and control node,the image acquisition module in the system can collect the home data;then enhance the voice signal through the twolevel noise reduction method of improving signal subspace and Wiener filtering;then,24 mel inverted spectrum coefficient is selected to extract speech features;and finally use hidden Markov model HMM algorithm for template training and mode matching,finally realize the automatic voice control of intelligent home.Experimental results show that a total of 789 samples were correctly identified in the 800 test samples,with an average recognition rate of 98.6%.And in the five different signal-tonoise ratio,the speech recognition rate remains at 94%and above,up to 97.4%.This shows that the system has good antinoise ability,and the proposed speech recognition algorithm is of great significance to meet the needs of the system for speech automation and intelligence.

关 键 词：智能家居系统 语音识别 梅尔倒谱系数 HMM 自动控制 

分 类 号：TP392[自动化与计算机技术—计算机应用技术]