嵌入式语音识别系统的测试方法研究-《计算机技术与发展》

文章信息/Info

Author(s):: CHEN Chun-ling; LI Wen-wen; YU Han; School of Computer,Nanjing University of Posts and Telecommunications,Nanjing 210003,China

Keywords:: embedded software testing; speech recognition system; test methods; automated testing

摘要:: 为更全面地评估嵌入式语音识别系统的性能,定位识别系统的性能瓶颈,以及提高嵌入式语音识别系统的测试效率,提出一套对嵌入式语音识别系统通用的测试方案。采用交叉测试的方法,将语音识别系统移植到全实物仿真测试环境,通过主机向目标机发送调用命令,控制目标机的执行内容,并将测试结果传输到主机,在主机上通过自动化测试完成结果统计及分析。实验结果表明,全实物仿真测试环境下的测试结果更贴近真实的应用效果,通过交叉测试的方法完成主机对目标机的控制,实现非侵入式测试,借助自动化测试工具可以自动捕获并校验测试结果,辅助定位系统存在的问题。该测试方法适用于嵌入式语音识别系统的各个应用场景,全面评估系统性能,自动化测试可显著提高测试效率、提升测试结果的可信度。

Abstract:: In order to more comprehensively evaluate the performance of the embedded speech recognition system,locate the performance bottleneck of the recognition system,and improve the test efficiency of the embedded speech recognition system,we propose a general test plan for the embedded speech recognition system. The cross-test method is used to transplant the speech recognition system to the whole physical simulation test environment. The host sends a call command to the target machine to control its execution content. The test result will be transmitted to the host for statistical and analysis by automated test method. Experiment shows the test results in the whole physical simulation test environment are closer to the real application effect. Through the cross-test method the host control of the target machine is completed,achieving non - intrusive testing. Automated test tools automatically capture and verify test results to locate problems in the system. The test method is applicable to various application scenarios of the embedded speech recognition system,and the automatic test can significantly improve the test efficiency and improve the credibility of the test result.