基于多观测似然比的语音活动检测分析与系统实现-analysis and system implementation of speech activity detection based on multi-observation likelihood ratio.docxVIP

哈尔滨工业大学工学硕士学位论文 - II - Abstract Voice Activity Detection is the technology which can extract the starting and ending points from speech signal. Since voice activity detection plays an important role in many technologies such as speech analysis, speech synthesis, speech encoding and speech recognition, this paper researches on the voice activity detection technology, and realizes a voice ac tivity detection system. Because there have been many voice activity detection methods, and multiple observation likelihood ratio based voice activity detection is simple and can achieve good performance, this paper carrys out voice activity detection utilizinge multiple observation likelihood ratio feature . First we divide time frames from the speech signal, then we estimate the noise from the signal, and compute the likelihood ratio of each frame under the result of noise estimation. We constitute the features with the likelihood ratios of this frame and certain numbers of previous frames. After feature extraction, we train the decision rule through the training samples. This paper utilizes linear classifier as the decision rule model. S o the problem is how to train the proper weights of linear classifier to make the detection as accurate as possible. This paper presents Minimum Classification Error model, Maximum Area Under the ROC Curve model and Support Vector Machine model, and proposes an extended MaxA UC model. The experimental results indicate that the extended MaxAUC model and SVM model have better stability and detection performance. In order to understand the real requirement and module function, this paper analyzes the requirement and designs this system. In requirement analysis, this paper lists the demands of this system, and decompose this system as signal input , feature extraction, training and testing. So as to present the relationships between those modules, this paper uses data flow diagram to portray the transmition and processing of data in this system