基于韵律特征的SVM说话人确认

SVM speaker verification based on prosodic feature

提出了一种基于韵律特征和SVM的文本无关说话人确认系统。采用小波分析方法,从语音信号的MFCC、F0和能量轨迹中提取出超音段韵律特征,通过实验研究三者的韵律特征在特征层的最佳互补融合,得到信号的韵律特征PMFCCFE,用韵律特征的GMM均值超矢量作为参数训练目标话者的SVM模型,以更有效地区分目标话者和冒认话者。在NIST068side-1side数据库的实验表明,以短时倒谱参数的GMM-UBM系统为基准,超音段韵律特征的GMM-SVM系统的EER相对下降了57.9,MinDCF相对下降了41.4。

A text-independent speaker verification method based on prosodic features and SVM model is proposed.With wavelet analysis, prosodic features are extracted from MFCC, F0 and energy contours respectively, these complementary features are fused at feature level to yield a most effective feature PMFCCFE, GMM mean supervectors of PMFCCFE are used to train SVM models to discriminates target speakers and imposters more effectively.The experiments conducted on the 2006 NIST 8side-lside subset show that the prosodic GMM-SVM system relatively improves the performance of the verification system by 57.9% in EER,41.4% in MinDCF,compared with the MFCC-based GMM-UBM system.