期刊文献+

基于意图识别的智能语音交互机器人设计 被引量:2

Design of intelligent speech interactive robot based on intention recognition
原文传递
导出
摘要 为提高智能语音交互机器人语音交互的准确率,提出一种基于意图识别的机器人智能英语语音交互方法。通过引入Glove_BiGRU_Self-attention分类预测模型构建意图识别功能模块,并采用ROS分布式架构对系统功能模块进行整合,实现人机的智能语音交互。仿真结果表明,采用所提方法进行的语音意图识别,具有更高的准确率,相较于基于DCNN模型、基于CNN-LSTM模型与基于单向构建的GRU-Self-attention模型的意图识别方法,识别准确率分别高出8.03%、4.07%和2.14%,具有更好的识别效果;在特征提取上,训练时间较传统基于BiLSTM模型的提取方法,BiGRU的训练时间缩短了4倍,训练效率更高。实验结果表明,采用所提意图识别方法搭建的语音交互系统,对用户英语语音指令的识别准确率和识别效率依然拥有较好的结果,识别平均准确率达到了89.72%,识别时间均在0.35 s之内,证明所提方法可以应用于实际语音交互之中。应用实验表明,采用基于意图识别方法搭建的智能语音交互机器人,无论是在问答交互还是控制命令上,都可以准确对用户英语指令进行识别,根据用户要求进行相关回答或完成相应动作。由此得出,基于Glove_BiGRU_Self-attention的意图识别方法,可以应用于智能语音机器人的英语语音交互中。 In order to improve the accuracy of intelligent speech interaction robot,an intelligent speech interaction method based on intention recognition is proposed.By introducing Glove_BiGRU_Self-attention classification prediction model,intention recognition function module is constructed,and ROS distributed architecture is adopted to integrate the system function module,so as to realize intelligent speech interaction between human and machine.The simulation results show that the proposed method has a higher accuracy of speech intention recognition,which is 8.03%,4.07%and 2.14%higher than that of intention recognition methods based on DCNN model,CNN-LSTM model and unidirectional GRU-Self-attention model.It has better recognition effect;In terms of feature extraction,the training time of BiGRU is 4 times shorter than that of the traditional extraction method based on BiLSTM model,and the training efficiency is higher.The experimental results show that the speech interaction system built by the proposed intention recognition method still has good results on the recognition accuracy and efficiency of user voice commands,with the average recognition accuracy of 89.72%and the recognition time of 0.35 s,which proves that the proposed method can be applied to actual speech interaction.The application experiments show that the intelligent voice interactive robot built based on the intention recognition method can accurately identify user commands,and make relevant answers or complete corresponding actions according to user requirements,no matter in question and answer interaction or control commands.It is concluded that the intention recognition method based on Glove_BiGRU_Self-attention can be applied to the speech interaction of intelligent speech robots.
作者 苏岩 SU Yan(Xi’an Jiaotong University City College,Xi’an 710018,China)
出处 《自动化与仪器仪表》 2024年第1期131-136,共6页 Automation & Instrumentation
基金 《课程思政建设路径与创新方法研究——以大学英语为例》(KCSZ01039)。
关键词 人工智能 意图识别 预测分类 深度学习 英语语音交互 artificial intelligence intention recognition prediction classification deep learning voice interaction
  • 引文网络
  • 相关文献

参考文献15

二级参考文献107

共引文献56

同被引文献18

引证文献2

相关主题

;
使用帮助 返回顶部