资源简介
MSR Identity Toolbox:
A Matlab Toolbox for Speaker Recognition Research
Version 1.0
Seyed Omid Sadjadi, Malcolm Slaney, and Larry Heck
Microsoft Research, Conversational Systems Research Center (CSRC)
s.omid.sadjadi@gmail.com, {mslaney,larry.heck}@microsoft.com
This report serves as a user manual for the tools available in the Microsoft Research (MSR) Identity Toolbox. This toolbox contains a collection of Matlab tools and routines that can be used for research and development in speaker recognition. It provides researchers with a test bed for developing new front-end and back-end techniques, allowing replicable evaluation of new advancements. It will also help newcomers in the field by lowering the “barrier to entry”, enabling them to quickly build baseline systems for their experiments. Although the focus of this toolbox is on speaker recognition, it can also be used for other speech related applications such as language, dialect and accent identification.
In recent years, the design of robust and effective speaker recognition algorithms has attracted significant research effort from academic and commercial institutions. Speaker recognition has evolved substantially over the past 40 years; from discrete vector quantization (VQ) based systems to adapted Gaussian mixture model (GMM) solutions, and more recently to factor analysis based Eigenvoice (i-vector) frameworks. The Identity Toolbox provides tools that implement both the conventional GMM-UBM and state-of-the-art i-vector based speaker recognition strategies.
A speaker recognition system includes two primary components: a front-end and a back-end. The front-end transforms acoustic waveforms into more compact and less redundant representations called acoustic features. Cepstral features are most often used for speaker recognition. It is practical to only retain the high signal-to-noise ratio (SNR) regions of the waveform, therefore there is also a need for a speech activity detector (SAD) in the fr
代码片段和文件信息
function Fea = cmvn(fea varnorm)
% performs cepstral mean and variance normalization
%
% Inputs:
% - fea : input ndim x nobs feature matrix where nobs is the
% number of frames and ndim is the feature dimension
% - varnorm : binary switch (false|true) if true variance is normalized
% as well
% Outputs:
% - Fea : output p x n normalized feature matrix.
%
% Omid Sadjadi
% Microsoft Research Conversational Systems Research Center
if ( nargin == 1 ) varnorm = false; end
mu = mean(fea 2);
if varnorm
stdev = std(fea [] 2);
else
stdev = 1;
end
Fea = bsxfun(@minus fea mu);
Fea = bsxfun(@rdivide Fea stdev);
属性 大小 日期 时间 名称
----------- --------- ---------- ----- ----
目录 0 2014-05-09 20:35 MSR Identity Toolkit v1.0\code\
文件 720 2013-10-15 15:39 MSR Identity Toolkit v1.0\code\cmvn.m
文件 2963 2013-10-15 15:39 MSR Identity Toolkit v1.0\code\compute_bw_stats.m
文件 3030 2013-10-15 15:39 MSR Identity Toolkit v1.0\code\compute_eer.m
文件 3260 2013-10-15 15:39 MSR Identity Toolkit v1.0\code\demo_gmm_ubm.m
文件 4339 2013-10-15 15:39 MSR Identity Toolkit v1.0\code\demo_gmm_ubm_artificial.m
文件 5370 2013-10-15 15:40 MSR Identity Toolkit v1.0\code\demo_ivector_plda.m
文件 5812 2013-10-15 15:40 MSR Identity Toolkit v1.0\code\demo_ivector_plda_artificial.m
文件 2421 2013-08-16 17:40 MSR Identity Toolkit v1.0\code\extract_ivector.m
文件 1557 2013-10-15 15:40 MSR Identity Toolkit v1.0\code\fea_warping.m
文件 6148 2013-10-15 15:40 MSR Identity Toolkit v1.0\code\gmm_em.m
文件 5108 2013-10-15 15:40 MSR Identity Toolkit v1.0\code\gplda_em.m
文件 223 2013-10-15 15:40 MSR Identity Toolkit v1.0\code\hamming.m
文件 603 2013-10-15 15:40 MSR Identity Toolkit v1.0\code\htkread.m
文件 534 2013-10-15 15:41 MSR Identity Toolkit v1.0\code\htkwrite.m
文件 2068 2013-10-15 15:41 MSR Identity Toolkit v1.0\code\lda.m
文件 428 2013-10-15 15:41 MSR Identity Toolkit v1.0\code\length_norm.m
文件 4930 2013-10-15 15:41 MSR Identity Toolkit v1.0\code\mapAdapt.m
文件 498 2013-10-15 15:41 MSR Identity Toolkit v1.0\code\rm_dc_n_dither.m
文件 2567 2013-10-15 15:41 MSR Identity Toolkit v1.0\code\score_gmm_trials.m
文件 2118 2013-10-15 15:41 MSR Identity Toolkit v1.0\code\score_gplda_trials.m
文件 5639 2013-10-15 15:41 MSR Identity Toolkit v1.0\code\train_tv_space.m
文件 1856 2013-10-15 15:41 MSR Identity Toolkit v1.0\code\wcmvn.m
目录 0 2014-05-09 20:35 MSR Identity Toolkit v1.0\doc\
文件 162 2013-08-28 03:13 MSR Identity Toolkit v1.0\doc\~$entity Toolbox-rev5.docx
文件 162 2013-10-07 09:14 MSR Identity Toolkit v1.0\doc\~$entity Toolbox-rev6.docx
文件 1082656 2013-10-11 10:39 MSR Identity Toolkit v1.0\doc\MSR Identity Toolbox.docx
文件 1276208 2013-10-15 15:45 MSR Identity Toolkit v1.0\doc\MSR Identity Toolbox.pdf
文件 24404 2013-10-15 15:35 MSR Identity Toolkit v1.0\MSR-LA No distrib-OK to modify.docx
文件 159491 2013-10-15 15:36 MSR Identity Toolkit v1.0\MSR-LA No distrib-OK to modify.pdf
文件 5808 2013-10-15 15:36 MSR Identity Toolkit v1.0\MSR-LA No distrib-OK to modify.txt
............此处省略0个文件信息
- 上一篇:MATLAB环境下基于神经网络的车牌识别
- 下一篇:功率放大器预失真matlab实现
评论
共有 条评论