// Information on the quantity of data available for each client
// Outputs a list with the selected files for a defined quantity of data
int ExtractTargetDataInfo(Config& config)
{
String inputClientListFileName = config.getParam("targetIdList");
bool fixedLabelSelectedFrame;
String labelSelectedFrames;
if (config.existsParam("useIdForSelectedFrame")) // the ID of each speaker is used as labelSelectedFrame
fixedLabelSelectedFrame=false;
else{ // the label is decided by the command line and is unique for the run
labelSelectedFrames=config.getParam("labelSelectedFrames");
if (verbose) cout << "Computing on" << labelSelectedFrames << " label" << endl;
fixedLabelSelectedFrame=true;
}
unsigned long maxFrame=config.getParam("maxFrame").toLong();
String outputFilename=config.getParam("outputFilename");
ofstream outputFile(outputFilename.c_str(),ios::out| ios::trunc);
try{
XList inputClientList(inputClientListFileName,config); // read the Id + filenames for each client
XLine * linep;
if (verbose) cout << "InfoTarget" << endl;
// *********** Target loop *****************
while ((linep=inputClientList.getLine()) != NULL){ // linep gives the XLine with the Id of a given client and the list of files
String *id=linep->getElement(); // Get the Client ID (id)
outputFile<<*id;
String currentFile="";
XLine featureFileListp=linep->getElements(); // Get the list of feature file for the client (end of the line)
if (verbose) cout << "Info model ["<<*id<<"]"<<endl;
if (!fixedLabelSelectedFrame){ // the ID is used as label for selecting the frame
labelSelectedFrames=*id;
if (debug) cout <<*id<<" is used for label selected frames"<<endl;
}
// label files reading - It creates, for each file and each label, a cluster of segments - will be integrated witth the featre s - asap
SegServer segmentsServer; // Reading the segmentation files for each feature input file
LabelServer labelServer;
initializeClusters(featureFileListp,segmentsServer,labelServer,config); // Reading the segmentation files for each feature input file
unsigned long codeSelectedFrame=labelServer.getLabelIndexByString(labelSelectedFrames); // Get the index of the cluster with in interest audio segments
SegCluster& selectedSegments=segmentsServer.getCluster(codeSelectedFrame); // Gives the cluster of the selected/used segments
Seg *seg; // Will give the current segment
unsigned long frameCount=0;
selectedSegments.rewind(); // at the begin of the selected segments list
while(((seg=selectedSegments.getSeg())!=NULL) && (frameCount<maxFrame)){ // For each of the selected segments until the amount of data is get
frameCount+=seg->length();
cout << seg->sourceName()<<" "<<seg->begin()<<" "<<seg->length()<<" Total time="<<frameCount<<endl;
if (seg->sourceName()!=currentFile){
outputFile<<" "<<seg->sourceName();
currentFile=seg->sourceName();
}
} // end of the initial Train Iteration loop
outputFile<<endl;
if (verbose) cout << "Save info client ["<<*id<<"]" << endl;
} // end of the the target loop
} // fin try
catch (Exception& e)
{
cout << e.toString().c_str() << endl;
}
return 0;
}
/*
void baggedSegments(SegCluster &selectedSegments,RefVector<SegCluster> &baggedA,double & baggedProbability,
unsigned long minimumLength,unsigned long maximumLength){
Seg* seg; // reset the reader at the begin of the input stream
selectedSegments.rewind();
seg=selectedSegments.getSeg();
bool end=(seg==NULL);
unsigned long beginSeg=0,lengthSeg=0;
if (!end){
beginSeg=seg->begin();
lengthSeg=seg->length();
}
while(!end){
if (debug) cout << "bagged, current input seg ["<<beginSeg<<","<<lengthSeg<<"]"<<endl;
unsigned long verifyLength=correctedLength(lengthSeg,minimumLength,maximumLength);
bool moveSeg=true;
unsigned long length=0;
if (lengthSeg<=verifyLength){
moveSeg=true;
length=lengthSeg;
if (debug) cout <<"change seg"<<endl;
}
else{
moveSeg=false;
length=verifyLength;
}
// for all cluster in baggedA
if (length>0)
for (unsigned long idx=0;idx<baggedA.size();idx++) // For each component
if(baggedFrame(baggedProbability)){
SegServer &segServerOutput=baggedA[idx].getServer();
Seg &newSeg=segServerOutput.createSeg(beginSeg,length,0,seg->string(),seg->sourceName());
baggedA[idx].add(newSeg);
if (debug) cout << "bagged - Adding in bagged["<<idx<<"] the seg ["<<seg->sourceName()<<"]"<<newSeg.begin()<<" "<<newSeg.length()<<endl;
}
if (moveSeg){
seg=selectedSegments.getSeg();
end=(seg==NULL);
if (!end){
beginSeg=seg->begin();
lengthSeg=seg->length();
}
}
else{
lengthSeg-=length;
beginSeg+=length;
}
}
if ((debug) || (verboseLevel>3)){
cout <<"Bagged segments"<<endl;
for (unsigned long idx=0;idx<baggedA.size();idx++){
cout << "Bagged cluster["<<idx<<"]"<<endl;
showCluster(baggedA[idx]);
}
}
if (verbose){
unsigned long total=totalFrame(selectedSegments);
for (unsigned long idx=0;idx<baggedA.size();idx++){
unsigned long selected=totalFrame(baggedA[idx]);
double percent=(double)selected*100/(double) total;
cout <<"Bagged segments["<<idx<<"] Initial frames["<<total<<"] Selected frames["<<selected<<"] % selected["<<percent<<"]"<<endl;
}
}
}*/
void baggedSegments(SegCluster &selectedSegments,SegCluster &baggedFrameSegment,double baggedProbability,
unsigned long minimumLength,unsigned long maximumLength){
SegServer &segServerOutput=baggedFrameSegment.getServer();
Seg* seg; // reset the reader at the begin of the input stream
selectedSegments.rewind();
seg=selectedSegments.getSeg();
bool end=(seg==NULL);
unsigned long beginSeg=0,lengthSeg=0;
if (!end){
beginSeg=seg->begin();
lengthSeg=seg->length();
}
while(!end){
if (debug) cout << "bagged, current input seg ["<<beginSeg<<","<<lengthSeg<<"]"<<endl;
unsigned long verifyLength=correctedLength(lengthSeg,minimumLength,maximumLength);
double segBaggedProbability=baggedProbability;
bool moveSeg=true;
unsigned long length=0;
if (lengthSeg<=verifyLength){
moveSeg=true;
length=lengthSeg;
if (debug) cout <<"change seg"<<endl;
}
else{
moveSeg=false;
length=verifyLength;
}
if ((length>0) &&(baggedFrame(segBaggedProbability))){
Seg &newSeg=segServerOutput.createSeg(beginSeg,length,0,seg->string(),seg->sourceName());
baggedFrameSegment.add(newSeg);
if (debug) cout << "bagged - Adding the seg ["<<seg->sourceName()<<"]"<<newSeg.begin()<<" "<<newSeg.length()<<endl;
}
if (moveSeg){
seg=selectedSegments.getSeg();
end=(seg==NULL);
if (!end){
beginSeg=seg->begin();
lengthSeg=seg->length();
}
}
else{
lengthSeg-=length;
beginSeg+=length;
}
}
if ((debug) || (verboseLevel>3)){
cout <<"Bagged segments"<<endl;
showCluster(baggedFrameSegment);
}
if (verbose){
unsigned long total=totalFrame(selectedSegments);
unsigned long selected=totalFrame(baggedFrameSegment);
double percent=(double)selected*100/(double) total;
cout <<"Bagged segments, Initial frames["<<total<<"] Selected frames["<<selected<<"] % selected["<<percent<<"]"<<endl;
}
}
// Can use this function to get likelihood with a topgauss
double TopGauss::get(MixtureGD & UBM,FeatureServer &fs,String & featureFilename,Config & config){
StatServer ss(config);
String labelSelectedFrames =config.getParam("labelSelectedFrames");
unsigned long begin=fs.getFirstFeatureIndexOfASource(featureFilename);
fs.seekFeature(begin);
SegServer segmentsServer;
LabelServer labelServer;
initializeClusters(featureFilename,segmentsServer,labelServer,config);
// __android_log_print(ANDROID_LOG_DEBUG, "TopGauss::get", " Feature file %s \n", featureFilename.c_str());
verifyClusterFile(segmentsServer,fs,config);
unsigned long codeSelectedFrame=labelServer.getLabelIndexByString(labelSelectedFrames);
SegCluster& selectedSegments=segmentsServer.getCluster(codeSelectedFrame);
MixtureGDStat &acc=ss.createAndStoreMixtureStat(UBM);
Seg *seg; // current selected segment
selectedSegments.rewind();
unsigned long t=0; //cnt frames
acc.resetLLK();
unsigned long idxBegin=0;
while((seg=selectedSegments.getSeg())!=NULL){
unsigned long begin=seg->begin()+fs.getFirstFeatureIndexOfASource(seg->sourceName());
fs.seekFeature(begin);
Feature f;
idxBegin=this->frameToIdx(t);
for (unsigned long idxFrame=0;idxFrame<seg->length();idxFrame++){
fs.readFeature(f);
//unsigned long idx=this->frameToIdx(t);
unsigned long nbg=_nbg[t];
ULongVector index;
double sumNonSelectedWeights=_snsw[t];
double sumNonSelectedLLK=_snsl[t];
for (unsigned long i=0;i<nbg;i++) {
index.addValue(_idx[idxBegin+i]);
}
char c[100];
sprintf(c,"%d",(int)index.size());
config.setParam("topDistribsCount",c); // this should be high enough
if (t==0) {acc.computeAndAccumulateLLK(f,1.0,DETERMINE_TOP_DISTRIBS);acc.resetLLK();} // to remove in ALIZE, this is to init the LKvector
ss.setTopDistribIndexVector(index, sumNonSelectedWeights, sumNonSelectedLLK);
acc.computeAndAccumulateLLK(f,1.0,USE_TOP_DISTRIBS);
idxBegin+=nbg;
t++;
}
}
//ss.deleteMixtureStat(acc);
if (t!=_nt || idxBegin !=_nbgcnt) cout << "W: t("<<t<<") != _nt(" <<_nt<<")"<<"W: idxBegin("<<idxBegin<<") != _nbgcnt(" <<_nbgcnt<<")"<<endl;
return acc.getMeanLLK();
}
void FactorAnalysisStat::computeAndAccumulateGeneralFAStats(SegCluster &selectedSegments,FeatureServer &fs,Config & config){
if (verbose) cout <<"(FactorAnalysisStat) Compute General FA Stats (Complete)" << endl;
double *N_h, *N, *S_X_h, *S_X,*ff;
_matN_h.setAllValues(0.0);
_matN.setAllValues(0.0);
_matS_X_h.setAllValues(0.0);
_matS_X.setAllValues(0.0);
N_h=_matN_h.getArray(); N=_matN.getArray(); S_X_h=_matS_X_h.getArray();S_X=_matS_X.getArray();
MixtureGD & UBM=_ms.getMixtureGD((unsigned long) 1);
MixtureGDStat &acc=_ss.createAndStoreMixtureStat(UBM);
// Compute Occupations and Statistics
acc.resetOcc();
Seg *seg;
selectedSegments.rewind();
String currentSource="";unsigned long loc=0;unsigned long sent=0;
while((seg=selectedSegments.getSeg())!=NULL){
unsigned long begin=seg->begin()+fs.getFirstFeatureIndexOfASource(seg->sourceName()); // Idx of the first frame of the current file in the feature server
if (currentSource!=seg->sourceName()) {
currentSource=seg->sourceName();
loc=_ndxTable.locNb(currentSource);
sent=_ndxTable.sessionNb(currentSource);
if (verbose)cout << "Processing speaker["<<currentSource<<"]"<< endl;
}
fs.seekFeature(begin);
Feature f;
if (!_topGauss) {
for (unsigned long idxFrame=0;idxFrame<seg->length();idxFrame++){
fs.readFeature(f);
acc.computeAndAccumulateOcc(f);
RealVector <double> aPost=acc.getOccVect();
ff=f.getDataVector();
for(unsigned long k=0;k<_mixsize;k++) {
N_h[sent*_mixsize+k]+=aPost[k];
N[loc*_mixsize+k] +=aPost[k];
for (unsigned long i=0;i<_vsize;i++) {
S_X_h[sent*_supervsize+(k*_vsize+i)]+=aPost[k]*ff[i];
S_X[loc*_supervsize+(k*_vsize+i)] +=aPost[k]*ff[i];
}
}
}
}
else throw Exception("ComputeGeneralStats TopGauss not done at this level",__FILE__,__LINE__);
}
};
/// Normalize features with a smooth mixture transformation o't=ot-sum(P(c|ot)Uc.x)
void FactorAnalysisStat::normalizeFeatures(SegCluster &selectedSegments,FeatureServer &fs,Config & config){
if (verbose) cout << "(FactorAnalysisStat) Normalize Features" << endl;
MixtureGD & clientMixture=_ms.getMixtureGD(1); // copy the UBM mixture
unsigned long nt=0;
RealVector <double> m_xh_1; m_xh_1.setSize(_supervsize);
double *_m_xh_1=m_xh_1.getArray();
Seg *seg; // current selectd segment
selectedSegments.rewind();
String currentSource="";
while((seg=selectedSegments.getSeg())!=NULL){
unsigned long begin=seg->begin()+fs.getFirstFeatureIndexOfASource(seg->sourceName());
if (currentSource!=seg->sourceName()) {
currentSource=seg->sourceName();
this->getUX(m_xh_1,currentSource);
this->getSpeakerModel(clientMixture,currentSource);
if (verbose)cout << "Processing speaker["<<currentSource<<"]"<< endl;
}
fs.seekFeature(begin);
Feature f;
if (!_topGauss) {
for (unsigned long idxFrame=0;idxFrame<seg->length();idxFrame++){
fs.readFeature(f,0);
double *ff=f.getDataVector();
double sum=0.0;
RealVector <double> P;
P.setSize(_mixsize);
double *Prob=P.getArray();
for(unsigned long k=0;k<_mixsize;k++) {
Prob[k]=clientMixture.weight(k)*clientMixture.getDistrib(k).computeLK(f);
sum+=Prob[k];
}
for(unsigned long k=0;k<_mixsize;k++)
Prob[k]/=sum;
for(unsigned long k=0;k<_mixsize;k++) {
for (unsigned long i=0;i<_vsize;i++)
ff[i]-= Prob[k]*_m_xh_1[k*_vsize+i];
}
fs.writeFeature(f);
nt++;
}
}
else {
throw Exception("no topgauss yet",__FILE__,__LINE__);
}
}
};
/// Compute Log Likelihood of the Factor Analysis model
double FactorAnalysisStat::getLLK(SegCluster &selectedSegments,MixtureGD &model,FeatureServer&fs,Config & config){
if (verbose) cout << "(FactorAnalysisStat) Compute Likelihood" << endl;
double llk=0.0;
MixtureGDStat &acc=_ss.createAndStoreMixtureStat(model);
Seg *seg;
selectedSegments.rewind();
while((seg=selectedSegments.getSeg())!=NULL){
unsigned long begin=seg->begin()+fs.getFirstFeatureIndexOfASource(seg->sourceName());
fs.seekFeature(begin);
Feature f;
for (unsigned long idxFrame=0;idxFrame<seg->length();idxFrame++){
fs.readFeature(f);
acc.computeAndAccumulateLLK(f,1.0,TOP_DISTRIBS_NO_ACTION);
}
}
llk= acc.getMeanLLK();
_ss.deleteMixtureStat(acc);
return llk;
};
// Build the segments with the energized frames
unsigned long selectFrames(FeatureServer &fs,SegServer & segServer,double threshold,SegCluster &selectedSeg,SegCluster &outputSeg,String labelOutput,String fileName)
{
unsigned long countFrames=0;
fs.reset(); // feature server reset
unsigned long ind=0;
unsigned long begin=0;
bool in=false;
Seg *seg; // current selectd segment
selectedSeg.rewind(); // reset the reader at the begin of the input stream
while((seg=selectedSeg.getSeg())!=NULL){ // For each input segments
for (unsigned long idx=seg->begin();idx<seg->begin()+seg->length();idx++){ // for each frame
Feature f;
fs.seekFeature(idx);
fs.readFeature(f);
if (f[0]>threshold){ // the frame is selected
countFrames++;
if (in==false){ // Begin of a new segment
in=true;
begin=ind;
}
}
else if (in){ // End of a segment
in=false;
Seg & segFake=segServer.createSeg(begin,ind-begin,0, // Create a segment - Take care : length=end-begin+1 but ind =end+1 !!
labelOutput,fileName);
outputSeg.add(segFake); // Add a segment
}
ind++; // Increment the frame index
} // end of one input segment
if (in){ // deal with the last energized segmeent inside the current input segment
in=false;
Seg & segFake=segServer.createSeg(begin,ind-begin+1,0, // Create a segment
labelOutput,fileName);
outputSeg.add(segFake); // Add a segment - Take care : length=end-begin+1 and ind=end in this case !!
}
} // end feature loop
return countFrames;
}
int saveApost(Config &config)
{
bool writeAllFeature=true; // Output a vector for all input vectors (selected and not selected vectors) - DEFAULT=on
if (config.existsParam("writeAllFeatures")) writeAllFeature=config.getParam("writeAllFeatures").toBool(); // Define if all the feature (selected or not) should be written
String modelname = config.getParam("inputModelFilename");
String inputFeatureFileName =config.getParam("inputFeatureFilename"); // input feature - could be a simple feature file or a list of filenames
XLine inputFeatureFileNameList; // The (feature) input filename list
if (inputFeatureFileName.endsWith(".lst")){ // If the file parameter is the name of a XList file
XList inputFileNameXList(inputFeatureFileName,config); // Read the filename list file
inputFeatureFileNameList=inputFileNameXList.getAllElements(); // And put the filename in a list if the file is a list of feature filenames
}
else { // It was a simple feature file and not a filename list
inputFeatureFileNameList.addElement(inputFeatureFileName); // add the filename in the list
}
try{
// read UBM
MixtureServer _ms(config);
StatServer _ss(config);
_ms.loadMixtureGD(config.getParam("inputWorldFilename"));
MixtureGD & UBM=_ms.getMixtureGD((unsigned long) 0);
MixtureGDStat &acc=_ss.createAndStoreMixtureStat(UBM);
unsigned long _vsize=UBM.getVectSize();
unsigned long _mixsize=UBM.getDistribCount();
// Loop over the list of feature files
String *file;
String labelSelectedFrames;
unsigned long codeSelectedFrame;
while ((file=inputFeatureFileNameList.getElement())!= NULL){
String & featureFilename=(*file);
FeatureServer fs(config,featureFilename);
FeatureServer fs_out(config,featureFilename);
SegServer segmentsServer;
LabelServer labelServer;
initializeClusters(featureFilename,segmentsServer,labelServer,config);
verifyClusterFile(segmentsServer,fs,config);
labelSelectedFrames=config.getParam("labelSelectedFrames");
codeSelectedFrame=labelServer.getLabelIndexByString(labelSelectedFrames);
SegCluster& selectedSegments=segmentsServer.getCluster(codeSelectedFrame);
// Compute Occupations and Statistics
acc.resetOcc();
Seg *seg;
selectedSegments.rewind();
String currentSource="";
while((seg=selectedSegments.getSeg())!=NULL){
unsigned long begin=seg->begin()+fs.getFirstFeatureIndexOfASource(seg->sourceName()); // Idx of the first frame of the current file in the feature server
if (currentSource!=seg->sourceName()) {
currentSource=seg->sourceName();
if (verbose)cout << "Processing speaker["<<currentSource<<"]"<< endl;
}
fs.seekFeature(begin);
Feature f;
for (unsigned long idxFrame=0;idxFrame<seg->length();idxFrame++){
fs.readFeature(f);
acc.computeAndAccumulateOcc(f);
RealVector <double> aPost=acc.getOccVect();
Feature tmpF;
for(unsigned long k=0;k<_mixsize;k++) {
tmpF[k]=aPost[k];
}
fs_out.addFeature(f);
}
}
// Writing apost probabilities to file
cout << "Writing to: " << featureFilename << endl;
FeatureFileWriter w(featureFilename, config); // build a featurefile writer to output the features (real features)
SegServer fakeSegServer;
if (writeAllFeature) { // Output all the features- feature count id the same SegServer fakeSegServer; // Create a new fake segment server
fakeSegServer.createCluster(0); // Create a new cluster
SegCluster& fakeSeg=fakeSegServer.getCluster(0); // Get the cluster
fakeSeg.add(fakeSegServer.createSeg(0,fs_out.getFeatureCount(),codeSelectedFrame, labelSelectedFrames,featureFilename)); // Add a segment with all the features
outputFeatureFile(config,fs_out,fakeSeg,w); // output all the features - giving the same file length
}
else
outputFeatureFile(config,fs_out,selectedSegments, w); // Output only the selected features - giving a shorter output
}
}
catch (Exception& e){cout << e.toString().c_str() << endl;}
return 0;
}
// Main init function
double TopGauss::compute(MixtureGD & UBM,FeatureServer &fs,String & featureFilename,Config & config){
StatServer ss(config);
MixtureGDStat &acc=ss.createAndStoreMixtureStat(UBM);
unsigned long _mixsize=UBM.getDistribCount();
String labelSelectedFrames =config.getParam("labelSelectedFrames");
unsigned long begin=fs.getFirstFeatureIndexOfASource(featureFilename);
fs.seekFeature(begin);
SegServer segmentsServer;
LabelServer labelServer;
initializeClusters(featureFilename,segmentsServer,labelServer,config);
// __android_log_print(ANDROID_LOG_DEBUG, "TopGauss::compute", " Feature file %s \n", featureFilename.c_str());
verifyClusterFile(segmentsServer,fs,config);
unsigned long codeSelectedFrame=labelServer.getLabelIndexByString(labelSelectedFrames);
SegCluster& selectedSegments=segmentsServer.getCluster(codeSelectedFrame);
acc.resetLLK();
double topD=config.getParam("topGauss").toDouble();
if (verbose) {if(topD<1.0) cout << "LLK %="<< topD << "% ";else cout << "Top-"<<topD<<" ";}
// Class values
_nt=totalFrame(selectedSegments);
_nbg.setSize(_nt); _idx.setSize(0);_snsw.setSize(0); _snsl.setSize(0);
_nbg.setAllValues(0); _idx.setAllValues(0);_snsw.setAllValues(0.0);_snsl.setAllValues(0.0);
_nbgcnt=0;
Seg *seg; // current selected segment
selectedSegments.rewind();
unsigned long t=0; //cnt frames
while((seg=selectedSegments.getSeg())!=NULL){
unsigned long begin=seg->begin()+fs.getFirstFeatureIndexOfASource(seg->sourceName());
fs.seekFeature(begin);
Feature f;
for (unsigned long idxFrame=0;idxFrame<seg->length();idxFrame++){
fs.readFeature(f);
double llk=acc.computeAndAccumulateLLK(f,1.0,DETERMINE_TOP_DISTRIBS);
const LKVector &topV=ss.getTopDistribIndexVector();
double lk_tot=exp(llk);
double val=0.0;
if (topD<1.0) {
for(unsigned long j=0;j<_mixsize;j++){
if (val > topD*lk_tot) break;
val+=(topV[j].lk);
_nbg[t]++;
}
} else _nbg[t]=(unsigned long)topD;
_nbgcnt+=_nbg[t];
double snsw=1.0;
double snsl=lk_tot;
for(unsigned long j=0;j<_nbg[t];j++) {
_idx.addValue(topV[j].idx);
snsw -=UBM.weight(topV[j].idx);
snsl -=topV[j].lk;
}
_snsw.addValue(snsw);
if (snsl < EPS_LK)
_snsl.addValue(EPS_LK);
else _snsl.addValue(snsl);
t++;
}
}
if (t!=_nt) cout << "W: t("<<t<<") != _nt(" <<_nt<<")"<<endl;
return acc.getMeanLLK();
}
//-------------------------------------------------------------------------
int trainWorld(Config& config){
if (verbose) cout << "Begin world model training"<<endl;
try{
// Reading the data, one or multiple separate streams
unsigned long nbStream=0; // Number of Streams
FeatureServer **fsTab=NULL; // Array of FeatureServer (address) - one by input stream
SegServer **segServTab=NULL; // Array of segment server (address)- one by input stream
SegCluster **segTab=NULL; // Array of selected segments cluster(address) - one by stream
double *weightTab=NULL;
// Array of weight of each stream. i.e influence of a stream on the final model
String outputWorldFilename = config.getParam("outputWorldFilename"); // output worldmodel file filename
bool fileInit=config.existsParam("inputWorldFilename"); // if a inputWorlFilename is given, init by file, else from scratch
bool saveInitModel=true;
if (config.existsParam("saveInitModel")) saveInitModel=config.getParam("saveInitModel").toBool();
String inputWorldFilename="";
if (fileInit) inputWorldFilename=config.getParam("inputWorldFilename"); // if file init, the initial model filename
String labelSelectedFrames =config.getParam("labelSelectedFrames"); // label for selected frames
TrainCfg trainCfg(config); // Get the training algo params
// Reading the data
if(config.existsParam("inputStreamList")){// We want to work on separated list
XList tmp(config.getParam("inputStreamList"),config); // Each data set influence will be balanced during training
XLine & listInputFilename=tmp.getAllElements(); // Read the list of (list) filenames in tmp -> listInputFilename
nbStream=listInputFilename.getElementCount();
if (nbStream==0) throw Exception("TrainWorld error:no input stream" , __FILE__, __LINE__);
reserveMem(fsTab,segServTab,segTab,weightTab,nbStream);
for (unsigned i=0;i<nbStream;i++)
featureStream(config,listInputFilename.getElement(i),fsTab[i],segServTab[i],segTab[i],labelSelectedFrames);
if (config.existsParam("weightStreamList")){ // Read the weight of each stream, text file
XList tmpW(config.getParam("weightStreamList"),config);
XLine & listW=tmpW.getAllElements(); // Read the list of (list) filenames in tmp -> listInputFilename
if (listW.getElementCount()!=nbStream) throw Exception("TrainWorld error: number of weigths differs than number of input streams" , __FILE__, __LINE__);
for (unsigned i=0;i<nbStream;i++) weightTab[i]=listW.getElement(i).toDouble();
}
}
else{ // Only one input stream, no stream list
nbStream=1;
reserveMem(fsTab,segServTab,segTab,weightTab,nbStream);
featureStream(config,config.getParam("inputFeatureFilename"),fsTab[0],segServTab[0],segTab[0],labelSelectedFrames);
}
unsigned long vectSize=fsTab[0]->getVectSize(); // size of the input vectors
// Create stat server and mixture server
MixtureServer ms(config);
StatServer ss(config, ms);
if (debug || verbose) cout << "Stream mode, nb Stream="<<nbStream<<endl;
if (debug|| (verboseLevel>2)){
for (unsigned long i=0;i<nbStream;i++){
cout <<"Stream["<<i<<"]"<<endl;
segTab[i]->rewind();
Seg *seg; // Reset to the first segment
while((seg=segTab[i]->getSeg())!=NULL) // For each of the selected segments
cout << "File["<<seg->sourceName()<<"] Segment begin["<<
seg->begin()<<"] length["<<seg->length()<<"] index in the feature server["<<fsTab[i]->getFirstFeatureIndexOfASource(seg->sourceName())<<"]"<<endl;
}
}
// Global mean and variance matrices initialisation (computed from dataa or set to 0,1)
bool use01=false;
if (config.existsParam("use01")) use01=config.getParam("use01").toBool();
if (verbose){ if (use01) cout<<"Use 0 mean, 1 cov "<<endl; else cout << "Compute global mean and cov"<<endl;}
DoubleVector globalMean;
DoubleVector globalCov;
if (!use01){
FrameAccGD globalFrameAcc;
unsigned long nbFrame=computeMeanCov(config,fsTab,segTab,nbStream,globalMean,globalCov); // Compute the global mean and covariance
if (verboseLevel>1){
cout <<"global mean and cov of training data, number of frame= ["<<nbFrame<<"]"<<endl;
for (unsigned i=0; i < vectSize; i++)cout << "mean[" << i << "=" << globalMean[i] << "]\tcov[" << globalCov[i] << "]" << endl;
}
}
else initialize01(vectSize,globalMean,globalCov);
MixtureGD &world=ms.createMixtureGD();
if (fileInit){ // Load or initialize the initial model
if (verbose) cout << "Load initial world model ["<<inputWorldFilename<<"]" << endl;
world=ms.loadMixtureGD(inputWorldFilename); // Load
}
else{
if (verbose) cout <<"World model init from scratch"<<endl;
mixtureInit(ms,fsTab,segTab,weightTab,nbStream,world,globalCov,config,trainCfg); // Initialize
if (saveInitModel) world.save(outputWorldFilename+"init", config);
}
MixtureGD *newWorld=&world; // TODO Verify and suppress...
trainModelStream(config,ms,ss,fsTab,segTab,weightTab,nbStream,globalMean,globalCov,newWorld,trainCfg);
if (verbose) cout << "Save world model ["<<outputWorldFilename<<"]" << endl;
newWorld->save(outputWorldFilename, config);
// Cleaning the memory
freeMem(fsTab,segServTab,segTab,weightTab,nbStream);
}
catch (Exception& e){
cout << e.toString() << endl;
}
return 0;
}