计算机工程 ›› 2020, Vol. 46 ›› Issue (3): 309-314.doi: 10.19678/j.issn.1000-3428.0053966

• 开发研究与工程应用 • 上一篇    下一篇

复杂噪声中基于MFCC距离的语音端点检测算法

韩云霄1, 邵清1, 符玉襄2, 郭庆2   

  1. 1. 上海理工大学 光电信息与计算机工程学院, 上海 200093;
    2. 中国电子科技集团公司第三十六研究所, 浙江 嘉兴 314000
  • 收稿日期:2019-02-21 修回日期:2019-04-20 发布日期:2019-03-27
  • 作者简介:韩云霄(1993-),女,硕士,主研方向为自然语言处理;邵清,副教授、博士;符玉襄,高级工程师、博士;郭庆,工程师、硕士。
  • 基金项目:
    国家自然科学基金(61703278);上海市科委科研计划(17511107203)。

Speech Endpoint Detection Algorithm Based on MFCC Distance in Complex Noise

HAN Yunxiao1, SHAO Qing1, FU Yuxiang2, GUO Qing2   

  1. 1. School of Optical-Electrical and Computer Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China;
    2. The 36 th Research Institute of China Electronics Technology Group Corporation, Jiaxing, Zhejiang 314000, China
  • Received:2019-02-21 Revised:2019-04-20 Published:2019-03-27

摘要: 为提高复杂噪声环境下语音信号端点检测的准确率,提出一种基于梅尔频谱倒谱系数(MFCC)距离的多维特征语音信号端点检测算法。通过计算语音信号的MFCC距离,结合短时能量和短时过零率对特征距离进行修正,并更新其阈值,建立自适应噪声模型,实现复杂噪声中语音信号端点的准确检测。实验结果表明,与基于双门限能量和基于倒谱距离的2种经典检测算法相比,在计算效率相同的条件下,该算法的检测准确率更高。

关键词: 语音信号, 端点检测, 多维特征, 梅尔频谱倒谱系数距离, 自适应噪声模型

Abstract: To improve the accuracy of speech signal endpoint detection under complex noise environment,this paper proposes a multidimensional feature speech signal endpoint detection algorithm based on MFCC distance.By calculating the MFCC distance of the speech signal and combining short time energy and short time over zero rate,this algorithm corrects the feature distance,updates the threshold value and establishes the adaptive noise model to achieve the speech signal endpoint detection in complex noise.Experimental results show that under the condition of same calculation efficiency,the proposed algorithm has higher detection accuracy compared with the two classic detection algorithms based on double threshold energy and cepstrum distance.

Key words: speech signal, endpoint detection, multidimensional features, MFCC distance, adaptive noise model

中图分类号: