基于段级特征主成分分析的说话人识别算法

文章编号:10019081(2013)07193503

doi:10.11772/j.issn.10019081.2013.07.1935

摘 要:

为了提高说话人识别(SR)系统的运算速度，增强其鲁棒性，以现有的帧级语音特征为基础，提出了一种基于段级特征主成分分析的说话人识别算法。该算法在训练和识别阶段以段级特征代替帧级特征，然后用主成分分析方法对段级特征进行降维、去相关。实验结果表明，该算法的系统训练时间、测试时间分别为基线系统的47.8%、40.0%，同时识别率略有提高，抑制了噪声对说话人识别系统的影响。该结果验证了基于段级特征主成分分析的说话人识别算法在识别率有所提高的情况下取得了较快的识别速度，同时在不同噪声环境下的不同信噪比情况下均可以提高系统识别率。

关键词：说话人识别；非线性分段；主成分分析；说话人识别系统

中图分类号: TP18文献标志码:A

英文标题

Speaker recognition method based on utterance level principal component analysis

英文作者名

CHU Wen1,2*, LI Yinguo2, XU Yang2, MENG Xiangtao1,2

英文地址(

1. College of Computer Science and Technology, Chongqing University of Posts and Telecommunications, Chongqing 400065, China;

2. Research Center of Automotive Electronics and Embedded System Engineering, Chongqing University of Posts and Telecommunications, Chongqing 400065, China

英文摘要)

Abstract:

To improve the calculation speed and robustness of the Speaker Recognition (SR) system, the authors proposed a speaker recognition algorithm method based on utterance level Principal Component Analysis (PCA), which was derived from the frame level features. Instead of frame level features, this algorithm used the utterance level features in both training and recognition. Whats more, the PCA method was also used for dimension reduction and redundancy removing. The experimental results show that this algorithm not only gets a little higher recognition rate, but also suppresses the effect of the noise on speaker recognition system. It verifies that the algorithm based on utterance level features PCA can get faster recognition speed and higher system recognition rate, and it enhances system recognition rate in different noise environments under different SignaltoNoise Ratio (SNR) conditions.

To improve the calculation speed and robustness of the Speaker Recognition (SR) system, the authors proposed a method based on utterance level Principal Component Analysis (PCA) of speaker recognition algorithm, which is derived from the frame level features. Instead of frame level features, this algorithm used the utterance level features both in training and recognition. What’s more the PCA method is also used for dimension reduction and redundancy removing. The experimental result shows that this algorithm not only gets a little higher recognition rate, but also suppresses the effect of the noise to Speaker Recognition System. It verifies that the algorithm based on utterance level features PCA can get faster recognition speed, higher system recognition rate, and it enhances system recognition rate in different noise environment under different SNR conditions.

英文关键词Key words:

Speaker Recognition (SR); nonlinear partition; Principal Component Analysis (PCA); speaker recognition system



0 引言

语音识别是指计算机对人类语音进行正确响应的技术［1］。广义的语音识别技术具体包括：语音识别、说话人识别、语种识别、语音评分［2］。说话人识别（Speaker Recognition，SR）技术是一项根据语音中反映说话人生理和行为特征的语音参数自动识别说话人身份的技术，其关键问题之一是提取反映说话人个性的语音特征参数。说话人识别系统常用的语音特征参数主要有梅尔倒谱系数(MelFrequency Cepstrum Coefficient，MFCC）［3-6］、线性预测系数（Linear Prediction Coefficient，LPCC）［7-9］以及它们的变体。为了提高特征的可识别性，往往会对特征进行二次处理，包括差分、组合等，这导致特征参数变得庞大，增加了存储量和计算量。