语音识别架构下英语口语考试辅助评分系统  

English Speaking Test Auxiliary Scoring System Based on Speech Recognition Architecture

作　　者：田伟伟 TIAN Wei-wei(Harbin Medical University,Daqing 163319 China)

机构地区：[1]哈尔滨医科大学大庆校区,黑龙江大庆163319

出　　处：《自动化技术与应用》2022年第5期168-170,179,共4页Techniques of Automation and Applications

摘　　要：现有的英语口语考试评分系统的评分过程大量依靠人工辅助,导致评分结果客观性不足。针对该情况,提出一种语音识别架构下英语口语考试辅助评分系统。针对语音识别芯片型号进行研究,引入Sphinx-4语音识别架构,提取语音共振峰特征频率,设计评分流程,完成语音识别架构下英语口语考试辅助评分系统设计。系统性能测试结果显示:设计的辅助评分系统不同题型的不可接受差异,均能控制在10%以下,并且效度值均在最优区间0.5-0.7内,验证设计系统的可靠性。The existing scoring process of the existing spoken English examination scoring system relies on manual assistance,leading to the insufficient objectivity of scoring results.For this case,a spoken English test auxiliary scoring system under speech recognition architecture is proposed.For the speech recognition chip model,the Sphinx-4 speech recognition architecture is introduced,and the characteristic frequency of the speech resonance peak is extracted,the scoring process is designed,and the design of the spoken English examination auxiliary scoring system under the speech recognition architecture is completed.The system performance test results show that the unacceptable differences of different types of the designed auxiliary scoring system can be controlled below 10%,and the validity value is within the optimal range of 0.5-0.7 to verify the reliability of the design system.

关 键 词：语音识别 特征提取 辅助评分 

分 类 号：TP391.43[自动化与计算机技术—计算机应用技术]