基于改进C_0复杂度和MFCC相似度的端点检测  

Endpoint detection based improvement of C_0 complexity and MFCC similarity

作　　者：许昊[1] 张二华[1] 

机构地区：[1]南京理工大学计算机科学与工程学院,江苏南京210094

出　　处：《现代电子技术》2015年第10期7-9,15,共4页Modern Electronics Technique

摘　　要：为了提高语音端点检测的准确率，提出一种基于改进C0复杂度和MFCC相似度相结合的端点检测算法。首先，计算每一帧语音信号的C0复杂度以及MFCC相似度。然后，结合C0复杂度与MFCC相似度作为新的特征参数，设置阈值进行端点检测。对信噪比在-15~15 dB范围内的含噪语音进行端点检测，使用Matlab软件进行仿真实验。实验结果表明，该方法相对于单独的两种方法提高了检测率，且稳定性更强。In order to improve the accuracy of voice activity detection（VAD）,an algorithm based on improvement of C0 complexity and Mel frequency cepstrum coefficient（MFCC）similarity is proposed. First,the C0 complexity and MFCC similarity of speech signal in each frame is calculated first,and then the comprehensive judgment is made in combination with the C0 com-plexity and MFCC similarity. VAD for noisy speech whose SNR ranges from -15dB to 15dB is conducted. Matlab software is used for simulation experiment. The results show that the accuracy and stability of the proposed method is higher than only C0 complexity detecting method or MFCC similarity detecting method.

关 键 词：音信号处理 C0复杂度 MFCC相似度 端点检测 

分 类 号：TN912-34[电子电信—通信与信息系统]