CAJ | 학술논문

提出一种采用超音段韵律特征和GMM-UBM模型结构的文本无关的说话人识别方法,用多尺度小波分析方法从短时倒谱参数MFCC和基频F_0随时间变化的韵律中分别提取可用于文本无关说话人识别的超音段韵律特征参数PMFCC和PF_0,并组成联合参数PMFCCF_0.在NIST06 8side-1side复杂背景电话手机语音数据库上的说话人确认实验则表明,采用一阶小波分析方法提取的超音段韵律参数PMFCC的识别性能与短时MFCC相当,采用超音段韵律特征PMFCCF_0的系统确认性能比采用短时MFCC系统有较大的提高.在微软数据库进行不同信噪比测试语音的说话人辨认实验表明,PMFCCF_0有比短时MFCC更好的噪声鲁棒性.
제출일충채용초음단운률특정화GMM-UBM모형결구적문본무관적설화인식별방법,용다척도소파분석방법종단시도보삼수MFCC화기빈F_0수시간변화적운률중분별제취가용우문본무관설화인식별적초음단운률특정삼수PMFCC화PF_0,병조성연합삼수PMFCCF_0.재NIST06 8side-1side복잡배경전화수궤어음수거고상적설화인학인실험칙표명,채용일계소파분석방법제취적초음단운률삼수PMFCC적식별성능여단시MFCC상당,채용초음단운률특정PMFCCF_0적계통학인성능비채용단시MFCC계통유교대적제고.재미연수거고진행불동신조비측시어음적설화인변인실험표명,PMFCCF_0유비단시MFCC경호적조성로봉성.
A text-independent speaker recognition method was proposed based on the super-segment prosodic feature and GMM-UBM. With wavelet multiresolution analysis, the super-segment prosodic feature PF_0 from F_0～t and PMFCC from MFCC～t were extracted, which were used for text-independent speaker recognition and could be combined as PMFCCF_0. Experiments of speaker identification in different SNRs on Microsoft database indicate that PMFCCF_0 is more robust than MFCC. Experiments on the 2006 NIST 8side-1side subset speaker recognition evaluation task show that PMFCC performs quite as well as MFCC in speaker recognition and the system verification performance based on PMFCCF_0 exhibits better noise robustness compared with MFCC.