基于Scratch的高能效人工智能语言教学系统构建  

Construction of an energy-efficient AI language teaching system based on Scratch

作　　者：祝带君 张伟明[1] ZHU Daijun;ZHANG Weiming(Xi’an Siyuan University,Xi’an 710038,China)

机构地区：[1]西安思源学院,西安710038

出　　处：《自动化与仪器仪表》2025年第2期266-269,274,共5页Automation & Instrumentation

基　　金：西安思源学院校级科研基金项目《基于地缘政治的澜湄区域中文教育发展与市场需求调查研究》(XASYZX-2210)。

摘　　要：针对传统中文语言教学方法存在智能化水平低,导致教学效果不佳的问题,提出构建一个基于Scratch的高能效人工智能语言教学系统。首先,采用树莓派3b+控制模块作为系统硬件主控芯片;然后利用Scratch编程平台和对应插件设计一个语音智能点灯;最后在语音识别模块中构建一个基于GMM-HMM的声学模型,通过其实现各种语音命令的准确识别。实验结果表明,本系统对两段中文语音文段识别的词错误率WER仅为15.06%,传统的Bi LSTM-CTC的词错误率WER取值为36.71%,本系统高出了21.65%。由此说明,本系统对中文语音文段的识别精度更高,满足中文语言教学智能化需求,中文语言教学效果和趣味性显著提升。In view of the problem that the intelligence level of traditional Chinese language teaching methods is low,which leads to poor teaching effect,it is proposed to build a high-efficiency artificial intelligence language teaching system based on Scratch.First,the Raspberry Pi 3b+control module is used as the system hardware master chip;then use the Scratch programming platform and the corresponding plug-in to design a voice intelligent light;finally build a GMM-HMM based acoustic model in the speech rec-ognition module to achieve accurate recognition of various voice commands.The experimental results show that the word error rate WER of the two Chinese speech paragraphs of the system is only 15.06%,and the word error rate WER of the traditional Bi LSTM-CTC is 36.71%,and this system is 21.65%higher.This shows that the system has a higher recognition accuracy of Chinese pronun-ciation and text paragraphs,which meets the needs of intelligent Chinese language teaching,and the effect and interest of Chinese language teaching are significantly improved.

关 键 词：Scratch编程 STM32控制 GMM-HMM 教学系统 语音识别 

分 类 号：TP392[自动化与计算机技术—计算机应用技术]