摘要
针对传统的心理健康监测方法分析效果低下、监测效果不佳等问题。提出了一个知识图谱结合语音特征识别提取的新方法。研究结果表明,研究使用方错字率最低只有8.65%,相较于其他方法降低了10.67%。并且在不同关键词提取中错字率分别只有3.86%和4.01%。相较于感知线性预测系数方法,研究使用方的准确率提升了9.59%。同时研究使用方法的准确率和召回率也在不同相关关键词中表现更好,准确率和召回率分别高感知线性预测系数方法8.54%和11.34%。由此可见研究使用方法具有更好的关键词提取能力,能够更好地对心理咨询语音通话进行监测,这对心理监测分析效果地提升具有较好的指导作用。
Addressing the issues of low analysis and poor monitoring effectiveness of traditional mental health monitoring methods.A new method combining knowledge graph and speech feature recognition extraction has been proposed in the study.The research results indicate that the lowest spelling error rate in the study is only 8.65%,which is 10.67%lower than other methods.And the error rates in extracting different keywords were only 3.86%and 4.01%,respectively.Compared to the perceptual linear prediction coefficient method,the accuracy of the research user has increased by 9.59%.At the same time,the accuracy and recall of the research method also performed better in different related keywords,with high perceptual linear prediction coefficient methods of 8.54%and 11.34%for accuracy and recall,respectively.It can be seen that the research usage method has better keyword extraction ability and can better monitor psychological counseling voice calls,which has a good guiding role in improving the effectiveness of psychological monitoring analysis.
作者
秦波
任薇
QIN Bo;REN Wei(Xinjiang Institute of Engineering,Urumqi 830011,China)
出处
《自动化与仪器仪表》
2025年第7期174-178,共5页
Automation & Instrumentation
基金
新疆工程学院校内课题《高校心理健康教育中的价值研究-以新疆工程学院为例》(2019xgy922106)。
关键词
知识图谱
语音特征提取
语音特征识别
心理监测
knowledge graph
speech feature extraction
speech feature recognition
psychological monitoring
