欧几里德距离
Mel倒谱
相似性(几何)
语音识别
模式识别(心理学)
特征(语言学)
特征提取
说话人识别
相似性度量
人工智能
数学
计算机科学
语言学
图像(数学)
哲学
作者
Mahesh K. Singh,Narendra Singh,Ashima Singh
标识
DOI:10.1109/icsc45622.2019.8938366
摘要
The Euclidean distance and MFCC feature extraction technique are the composite way and useful for successfully measure the similarity between speakers. This type of approaches is used in the forensic statics for similarity of voice measurement. The speaker voice similarity is measured by using Euclidean distance (ED) through a different set of speakers as same-same speakers (SS) and difference speakers (DS). This approaches suggested that the feature coefficient of speech maximum dependent on speakers voice. The ED is calculated from the mean of Mel-frequency cepstral coefficient (MFCC) for speaker similarity measurement. There is ED difference method is proposed for finding the similarity index between SS and DS.
科研通智能强力驱动
Strongly Powered by AbleSci AI