基于AI人工智能的学前教育机器人对话系统研究

Research on preschool education robot dialogue system based on AI artificial intelligence

为了实现学前教育机器人对话系统的优化,以语音增强为切入点,提出一种基于双阶段注意力机制的儿童语音增强方法,以优化学前教育机器人对话系统,使机器人能够更准确地理解和执行学前儿童语音中的文本指令。首先分析了主流语音增强方法的基本原理以及单通道时域语音增强存在的问题;然后搭建双阶段注意力增强网络的语音增强方法,并在学前教育机器人对话系统中加入双阶段注意力语音增强模块进行测试。结果表明:双阶段注意力语音增强方法的SSNR得分相比于得分较高的通道注意力增强网络提高了3.3 dB,loss值是四种模型中最低的即0.98,且语音增强时域波形图清晰;综上研究得出,提出的语音增强方法具有可行性和有效性。

In order to realize the optimization of preschool education robot dialogue system,taking speech enhancement as the starting point,a children’s speech enhancement method based on two-stage attention mechanism is proposed to optimize the preschool education robot dialogue system,so that the robot can more accurately understand and execute the text instructions in preschool children’s speech.Firstly,the basic principles of mainstream speech enhancement methods and the problems of single-channel time-domain speech enhancement are analyzed.Then,a speech enhancement method of two-stage attention enhancement network is constructed,and a two-stage attention speech enhancement module is added to the preschool education robot dialogue system for testing.The results show that the SSNR score of the two-stage attention speech enhancement method is improved by 3.3dB compared with the channel attention enhancement network with high score,and the loss value is the lowest of the four models,that is,0.98,and the speech enhancement time domain waveform is clear.In summary,it is concluded that the speech enhancement method proposed in this paper is feasible and effective.