对话电话语音的话者确认分析-电路与系统专业论文.docx

ABSTRACT ABSTRACT - - IV - these speech segments, and then, label the different segment areas with Viterbi algorithm. This thesis proposes two methods to deal the segment results, which can improve the system performance. First, after segment and clustering , the purify of the result is calculate, the segment with low score segments will be discard and the high score segments, which means clean speech ,will be saved ,then we use these speech for verification Second we can segment the speech with more then two method and then fusion the segment results. The same areas can be consider as cleaner speech ,we train the speaker models with these speech ,and then ,we label the different area with Viterbi algorithm In order to improve the systems performance, the conception of speech quality is introduced and then, a new score method is proposed based on GMM-UBM system. by integrating auxiliary side information, the Log-likelihood Ratio is calculated dynamically . the experiments indicate the good performance of proposed method . This thesis designed a two speaker verification system, and given the performance through some different conditions, the results show the robustness and validity of the system.. Key Words: Conversation Speech, Speaker Verification , Segmentation and Clustering, Speech Quality，Gaussian Mixture Model 图目录 图目录 - - PAGE IX - 图目录 图 1-1 语音识别分类 1 图 1-2 说话人辨识和说话人确认的框图 3 图 1-3 对话语音训练语音的组成结构 7 图 1-4 单话者与对话语音说话人确认系统框图 8 图 1-5 FA和FR示意图 9 图 1-6 DET曲线示例 10 图 2-1 人的发声器官解剖示意图 14 图 2-2 语音信号发生的离散时域模型 15 图 2-3 两个男性说话人的语音波形信号及其语谱图表示 16 图 2-4 Mel?刻度与实际频率对应关系曲线 21 图 2-5 MFCC参数的提取过程 21 图 2-6 三角形的 Mel滤波器组 22 图 2-7 两个不同的男性话者的基音频率统计分布 24 图 2-8 一帧信号及其经中心削波前后的自相关波形 26 图 2-9 多尺度小波分解过程 29 图 2-10 log (F0)~t及其一阶概貌系数、细节系数 30 图 2-11 F0 超音段韵律参数的提取与组成 30 图 3-1 GMM 通过若干个高斯概率密度函数的加权和拟合任意分布 35 图 3-2 高斯混合模型示意图 37 图 3-3 GMM-UBM话者确认框架 42 图 3-4 GMM-UBM-MAP框架 43 图 3-5 UBM自适应示意图 45 图 4-1 RASTA滤波器频率响应 52 图 4-2 参数高斯化概率分布规整流程图 54 图 4-3 几种参数规整技术的系统性能曲线 57 图 4-4 F0 系统性能曲线 5