欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
为了提取到能够区分不同说话人个性特征的最优特征参数，采用在Mel频率倒谱系数（Mel—frequency cepstrum coefficients，MFCC）基础上进行改进的复合参数，即增加归一化短时能量参数和一阶差分所构成的特征矢量作为特征。针对高维特征参数，提出了一种基于相关距离Fisher准则的特征选取方法，利用该方法对提取出的参数进行加权降维。通过实验对比结果表明，该算法提高了识别率，具备可行性与优越性，是一种有效的特征提取算法。
英文摘要：
To extract speaker＇s personality characteristics that different speakers can be distinguished better. Firstly, a design method of Mel cepstrum composite coefficients based on the Mel Frequency cepstrum coefficient （MFCC） is constructed. The normalized short-time energy parameters and first-order difference are used to be the improved feature parameters. Then, in view of the high dimensional parameters, a algorithm for feature selection about Fisher criterion with correlation distance is introduced. The weighted algorithm designed to lower dimension for Mel cepstrum composite coefficients. Finally, an simulation example is presented to prove the recognition rate is improved as well as the feasibility and superiority of the proposed method, indicated that this study is an effective feature extraction algorithm.
同期刊论文项目
同项目期刊论文
Parameter optimization and application of support vector machine based on parallel artificial fish s
Study for classification of emotional speech by using optimized frame zero crossing with peak amplit
A robust feature extraction approach based on an auditory model for classification of speech and exp
期刊信息
位置：