For using this code you have to add the netlab and the gmeans packages to your path.
Usage:
for calculating MFCC features : CC4M_CompFeat(folder,parameters,projectname)
for training VQ on these: 
1. CC4M_AllFileList(folder,parameters,projectname): write wav file list 
2. VQonMFCC(folder,parameters,projectname): train Vector Quantization

The folder has to contain subfolders with the wav files of the target speakers.
The parameters structure is explained in the CC4M_ParamInit.m file
projectname specifies the name of the cluster file to be stored.

The feature names have the form:
$WAVFILENAME_$WINDOWLENGTH_$OVERLAP_$NUMCEPCOEFFS_$ENERGYUSED_$DELTASUSED.mat and the frames are contained in the columns
The cluster file name has the form
Clusters_$CLUSTERINGMETHOD_$NUMCENTRES_$NUMCEPCOEFFS_$ENERGYUSED_$DELTASUSED and it contains:
1. mfcc_all: structure with the fields mfcc(data), index(frame index), name(file name), cluster(to which cluster the frame is assigned)
2. centres: centres of the clusters in the columns of the matrix