智能家具语音识别精准度优化仿真

Intelligent Furniture Speech Recognition Accuracy Optimization Simulation

为了使家具的控制变得智能化,需要对智能家具语音识别方法进行研究。采用当前智能家具语音识别方法对人发出的命令语音进行识别时,存在抗干扰性能差和识别精准度低的问题。提出一种智能家具语音识别精准度优化方法,采用动态时间规整方法将识别过程中的距离测试计算和时间规整结合到一起;通过谱减法去除语音信号中存在的噪声信号,得到纯净的语音信号;采用线性系数对语音信号的特征进行反映,根据线性系数在半监督的思想上采用直推式支持向量机建立目标函数,根据得到的最优解对语音信号进行分类,完成智能家具的语音识别。仿真结果表明,所提方法的去噪效果好、可将识别过程中的词错率控制在10%以下,验证所提方法的抗干扰性高、识别精准度高。

Due to poor anti-interference performance and low recognition accuracy of the existing intelligent furni- ture voice recognition methods, an accuracy optimization method for intelligent furniture voice recognition was put for- ward. At first, the dynamic time warping method was used to combine the distance test with the time warping in the recognition process. Then, the method of spectrum subtraction was used to remove noise signal existing in the voice signal, so as to obtain the pure voice signal. In addition, linear coefficient was used to reflect feature of voice signal. According to the linear coefficient, the transduetive support vector machine was used to establish objective function on the semi-supervised idea. According to the optimal solution, the speech signal was classified. Thus, the voice recog- nition of intelligent furniture was completed. Simulation results show that the proposed method has good denoising effect, which can control the word error rate in the recognition process to be less than 10%. Thus, the proposed method has high anti-interference performance and high recognition accuracy.