摘要
该文设计了基于高斯混合模型的说话人识别系统实验,通过录制小型语音库、提取表征说话人个性的特征参数、训练说话人模型和似然度判决,实现文本无关的说话人识别。实验以小组形式协作完成,并鼓励学生展开进阶研究,帮助学生提升团队协作精神和创新精神,以及解决复杂工程问题的能力。所建立的全过程多维度立体化的实验考核体系,着眼于对学生的方案设计、实验操作、报告撰写、演示答辩等全过程评价,有助于激发学生的创造力和学习热情。
In this paper, a speaker recognition system experiment based on Gaussian mixture model is designed.Text independent speaker recognition is realized by recording a small speech database, extracting characteristic parameters representing the speaker’s personality, and training the speaker model and likelihood decision. The experiment is completed in groups, and students are encouraged to carry out advanced research to improve their teamwork spirit, innovation spirit and ability to solve complex engineering problems. The established multi-dimensional and stereoscopic experimental examination system focuses on the overall process evaluation of students’ scheme design, experimental operation, report writing, demonstration and defense, which helps to inspire students’ creativity and learning enthusiasm.
作者
芮贤义
陈小平
俞一彪
RUI Xianyi;CHEN Xiaoping;YU Yibiao(School of Electronics and Information Engineering,Soochow University,Suzhou 215006,China)
出处
《实验技术与管理》
CAS
北大核心
2022年第2期166-170,共5页
Experimental Technology and Management
基金
国家级一流本科专业建设点(教高厅函[2021]7号)
苏州大学一流本科教学团队建设项目(SDJ201911210)
苏州大学高等教育教改研究课题(苏大教[2021]108号)。
关键词
实验设计
说话人识别系统
实验考核
experimental design
speaker recognition system
experimental assessment
