基于语音情感分析系统的语音压力测谎

目的探究语音情感分析系统(Layered Voice Analysis,LVA)在不同情况下的表现,并通过分析数据找出影响系统识别率的因素。方法设置两种不同场景对20名研究对象进行提问,并录音。通过LVA语音系统对语音样本进行分析,使用SPSS统计软件对系统生成的参数进行统计分析。结果在未告知研究对象实验真相的情况下LVA语音系统平均识别率为91.75%,要远高于研究对象在知情情况下的识别率。在询问过程中,当未涉及敏感问题时,随着时间的推移研究对象的心理压力逐渐减小。结论LVA语音系统具有良好的情感分析和识别能力,可为司法工作提供参考。

Polygraphing from Speech Stress through Layered Voice Analysis

Objective To explore the performance of LVA(layered Voice Analysis)system in different situations so as to fi nd out the factors that affect the recognition rate of the system by analyzing the collected phonetic data.Methods Twenty subjects were asked with questions under two different scenarios,having their answering voices recorded.The recorded speech samples were detected by LVA system so that thus-generated parameters were analyzed by SPSS statistical software.Results The average recognition rate of LVA system was 91.75%when the subjects were not informed of the truth of experiment,contrasting against the comparatively lower recognition rate while the subjects were informed.When being questioned,the subjects were gradually decreasing their psychological pressure with time elapsing if no sensitive questions were involved.Conclusions LVA speech system is well able to use into emotion analysis and recognition,capable of providing reference for judicial work.