语音是人类最自然便捷的交流方式,也是人机交互中最直接的交互模式之一.对语音识别的基本理论进行了介绍,包括语音信号的预处理、特征提取、模型匹配3个方面,给出了基于MFCC+△MFCC的语音特征参数提取方法.在实验室环境下提取了特定人的若干语音数据,利用MATLAB对语音特征参数MFCC+ △MFCC进行了仿真识别实验,实验结果表明,该方法识别率高达94.54％,具有较强的鲁棒性.在实验基础上,通过搭建设计4个自由度的Arduino双臂机器人和ASRM08-A语音识别模块等硬件电路,实现了语音控制机器人完成各种规划动作.%Speech is the most natural and convenient way of communication,it is also one of the most direct interaction pattern of human-computer interaction.Firstly,we introduce the basic theory of speech recognition,including speech signal preprocessing,feature extraction,and model matching.A speech feature parameter extraction method is then presented based on MFCC and delta MFCC.We get some speech data in laboratory environment,and use Matlab to make simulated experiment for speech feature parameters of MFCC and delta MFCC.The experiment result proves that this method can get 94.54％ recognition accuracy and it has high robustness.On the basis of the experiment,by constructing and designing four degrees of freedom of the Arduino robot arms and ASR M08-A speech recognition module and hardware circuits,the voice control of robot and planned actions are achieved.
展开▼
