Mercurial > hg > audio-time-warp

changeset 1:6842ff391568

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
getting rid of pointer use in processAudio - not many changes
author Andrew N Robertson <andrew.robertson@eecs.qmul.ac.uk>
date Wed, 18 May 2011 16:47:12 +0100
parents 572c856e38ac
children 9ce18f24b266
files chromagramm/Chromagram.cpp src/testApp.cpp src/testApp.h src/timeWarp.cpp src/timeWarp.h
diffstat 5 files changed, 50 insertions(+), 45 deletions(-) [+]
line wrap: on
line diff
--- a/chromagramm/Chromagram.cpp	Tue May 17 08:48:58 2011 +0100
+++ b/chromagramm/Chromagram.cpp	Wed May 18 16:47:12 2011 +0100
@@ -8,7 +8,7 @@
 Chromagram :: Chromagram()
 {
 	pi = 3.14159265;
-	bsize = 8192;
+	bsize = 8192;//buffersize - ie main framesize at 11025Hz
 		
 	// array init
 	buffer = new float[8192]; // make audio buffer
--- a/src/testApp.cpp	Tue May 17 08:48:58 2011 +0100
+++ b/src/testApp.cpp	Wed May 18 16:47:12 2011 +0100
@@ -74,9 +74,12 @@
 	backwardsAlignmentIndex = 0;//remember that this goes backwards!
 
 	tw.initialiseVariables();
+	
+	//here is the main timewarp similarity matrix calc, the minimum alignment matrix via dtw and then the backwards path estimate 
 	tw.calculateSimilarityMatrix();	
 	tw.calculateAlignmentMatrix();
 	tw.calculateMinimumAlignmentPath();
+	
 	printf("\n gettem hereafter!");	
 	//set not to play
 	audioPlaying = false;
@@ -92,6 +95,8 @@
 	
 	initialiseVariables();
 	
+	chromaG.initialise(FRAMESIZE, CHROMAGRAM_FRAMESIZE);
+	
 	
 }
 
@@ -150,6 +155,8 @@
 }
 
 void testApp::updateAlignmentPathIndex(int identifier){
+
+	//this is the alignment where we are currently playing - i.e. switching between files
 	
 	int chromaPosition = audioPosition/CHROMA_CONVERSION_FACTOR;
 	
@@ -537,15 +544,16 @@
 	
 }
 
-void testApp::processAudioToDoubleMatrix(Chromagram* chromaG, DoubleMatrix* myDoubleMatrix, DoubleVector* energyVector){
+void testApp::processAudioToDoubleMatrix(DoubleMatrix* myDoubleMatrix, DoubleVector* energyVector){
 	//wendy
 	myDoubleMatrix->clear();
 	energyVector->clear();
+
 	
 	frameIndex = 0;
-//	chromaIndex = 0;//	WHY NEED THIS?
-	chromaG->initialise(FRAMESIZE,2048);//framesize 512 and hopsize 2048 
-	chromaG->maximumChromaValue = 0;
+	
+	chromaG.initialise(FRAMESIZE, CHROMAGRAM_FRAMESIZE);//framesize 512 and hopsize 2048 - already done
+	chromaG.maximumChromaValue = 1;
 	
 	int readcount = 1; // counts number of samples read from sound file
 	printf("processing audio from doublematrix \n");
@@ -556,41 +564,34 @@
 		// read FRAMESIZE samples from 'infile' and save in 'data'
 		readcount = sf_read_float(infile, frame, FRAMESIZE);
 		
-		//printf("readcount %i", readcount);
+		//8192 samples per chroma frame  //processing frame - downsampled to 11025Hz
+		chromaG.processframe(frame);
 		
-		//processing frame - downsampled to 11025Hz
-		//8192 samples per chroma frame
-		chromaG->processframe(frame);
-		
-		if (chromaG->chromaready)
+		if (chromaG.chromaready)
 		{
 				DoubleVector d;
 				
 				for (int i = 0;i<12;i++){
-					d.push_back(chromaG->rawChroma[i]);// / chromaG->maximumChromaValue);	
+					d.push_back(chromaG.rawChroma[i]);// / chromaG->maximumChromaValue);	
 					
 				}	
 				//this would do chord detection
 				
 				myDoubleMatrix->push_back(d);
-	
+				//so now is storing at d[i][current_index]
 			
 		}//end if chromagRamm ready
 		
 		//printf("calling drawSndFile %i", frameIndex);
 		frameIndex++;
 		
-		
-		//	putEnergyInFrame();
-		//get energy of the current frame and wait
-		
 		double energyValue = getEnergyOfFrame();
 		energyVector->push_back(energyValue);
 
 		
 	}//end while readcount
 	
-	printf("Max chroma value is %f \n", chromaG->maximumChromaValue);
+	printf("Max chroma value is %f \n", chromaG.maximumChromaValue);
 	
 	//normalise
 	int length = myDoubleMatrix->size();
@@ -600,7 +601,8 @@
 	
 	for (int i = 0; i < myDoubleMatrix->size();i++){
 		for (int j = 0; j < ((*myDoubleMatrix)[0]).size();j++){
-	(*myDoubleMatrix)[i][j] /= chromaG->maximumChromaValue;	
+			//non-causal normalisation
+			(*myDoubleMatrix)[i][j] /= chromaG.maximumChromaValue;	
 		}
 	}
 	
@@ -608,13 +610,13 @@
 	size = energyVector->size();
 	printf("size of energy vector is %d \n", size);
 	
-	totalNumberOfFrames = frameIndex;//used to use this - but switch to energy vector's size instead
-//	totalNumberOfFrames = size;
+//	totalNumberOfFrames = frameIndex;//used to use this - but switch to energy vector's size instead
+	totalNumberOfFrames = size;
 	
 //	int size = myDoubleMatrix->size() * CHROMA_CONVERSION_FACTOR;
 //	printf("size of double matrix is %d and frame index %d", size, frameIndex);
 	
-	printf("Total frames %i  and Chroma index %i \n", frameIndex, chromaIndex);
+	printf("Total frames %i  \n", frameIndex);
 	
 	
 }
@@ -867,15 +869,16 @@
 
 
 void testApp::loadFirstAudioFile(){
-	Chromagram* cptr;
+/*
+ Chromagram* cptr;
 	DoubleMatrix* dptr;
 	DoubleVector* eptr;
 	
 	cptr = &(tw.chromoGramm);
 	dptr = &(tw.chromaMatrix);
 	eptr = &(tw.firstEnergyVector);
-	
-	processAudioToDoubleMatrix(cptr, dptr, eptr);
+	*/
+	processAudioToDoubleMatrix(&tw.chromaMatrix, &tw.firstEnergyVector);
 	
 }
 
@@ -893,7 +896,7 @@
 	dptr = &(tw.secondMatrix);
 	eptr = &(tw.secondEnergyVector);
 	
-	processAudioToDoubleMatrix(cptr, dptr, eptr);//chromaMatrix);
+	processAudioToDoubleMatrix(dptr, eptr);//chromaMatrix);
 	
 }
 
--- a/src/testApp.h	Tue May 17 08:48:58 2011 +0100
+++ b/src/testApp.h	Wed May 18 16:47:12 2011 +0100
@@ -17,6 +17,9 @@
 #define ENERGY_LENGTH 80000
 #define CHROMA_LENGTH 12000
 #define CHROMA_CONVERSION_FACTOR 16 //16 times as many frames in energy as in chroma
+#define CHROMAGRAM_FRAMESIZE 2048
+
+
 //length in terms of frames (at 512 samples per frame - there are 90 per second) => 900: 10 seconds
 
 class testApp : public ofBaseApp{
@@ -109,7 +112,7 @@
 	
 	void drawEnergyVectorFromPointer(DoubleVector* energyVec);
 	
-		void processAudioToDoubleMatrix(Chromagram* chromaG, DoubleMatrix* myDoubleMatrix, DoubleVector* energyVector);
+		void processAudioToDoubleMatrix(DoubleMatrix* myDoubleMatrix, DoubleVector* energyVector);
 	
 		void loadNewAudio(string soundFileName);
 		void loadSecondAudio(string soundFileName);
@@ -188,7 +191,7 @@
 	
 	
 		timeWarp tw;
-
+		Chromagram chromaG;
 		
 };
 
--- a/src/timeWarp.cpp	Tue May 17 08:48:58 2011 +0100
+++ b/src/timeWarp.cpp	Wed May 18 16:47:12 2011 +0100
@@ -37,7 +37,7 @@
 void timeWarp::initialiseVariables(){
 	diagonalPenalty = 2;//penalises diagonal so all path gradients equal weighting
 
-	chromoGramm.initialise(FRAMESIZE,2048);//framesize 512 and hopsize 2048
+	//chromoGramm.initialise(FRAMESIZE,2048);//framesize 512 and hopsize 2048
 
 }
 
@@ -71,8 +71,6 @@
 		DoubleVector d;
 		for (int y = 0;y < secondMatrix.size();y++){
 			
-			//d.push_back( drand48() );	
-			
 			distance = 0;
 			firstSum = 0;
 			secondSum = 0;
@@ -97,7 +95,7 @@
 	}//end for x
 	//	userInfoString += "; size =";
 	//	userInfoString += ofToString(similarityMatrix.size() , 0);
-	printf("..sim size: %i, height: %i \n", similarityMatrix.size(), (chromaMatrix[0]).size());
+	printf("..sim size: %i, height: %i \n", (int) similarityMatrix.size(), (int) (chromaMatrix[0]).size());
 	
 }//end self sim
 
@@ -111,11 +109,9 @@
 	d.push_back(getDistance(0,0));
 	alignmentMeasureMatrix.push_back(d);
 	
-	
 	bool chromaCalculated = false;
 	bool secondCalculated = false;
 	
-	
 	while (!chromaCalculated || !secondCalculated) {
 		
 		if (!chromaCalculated)
@@ -126,20 +122,18 @@
 		
 	}
 	
-	
-	
 }
 
 bool timeWarp::extendAlignmentUp(){
 	DoubleVector d;
-	d = alignmentMeasureMatrix[0];
+	d = alignmentMeasureMatrix[0];//alignmentMatrix[0];//
 	int heightSize = d.size();
 	if (heightSize < secondMatrix.size()){
 		//then we haven't finished yet
 		for (int i = 0;i < alignmentMeasureMatrix.size();i++){
 			double value = getDistance(i, heightSize);
 			value += getMinimum(i, heightSize, value);	
-			alignmentMeasureMatrix[i].push_back(value);
+			alignmentMeasureMatrix[i].push_back(value);//
 		}
 	}
 	if (alignmentMeasureMatrix[0].size() == secondMatrix.size())
@@ -167,7 +161,6 @@
 			alignmentMeasureMatrix[widthSize].push_back(value);
 		}
 		
-		//alignmentMeasureMatrix.push_back(d);
 	}
 	
 	if (alignmentMeasureMatrix.size() == chromaMatrix.size())
@@ -181,7 +174,7 @@
 void timeWarp::calculateMinimumAlignmentPath(){
 	//this requires one pass of the DTW algorithm and then works backwards from (N,M)
 	//to find the optimal path to (0,0), where N and M are the lengths of the two chromoVectors respectively
-	//minimumAlignmentPath.clear();
+	
 	backwardsAlignmentPath.clear();
 	
 	printf("Finding minimum Path \n");
@@ -201,7 +194,7 @@
 		printf("backwards path %i : %i \n", backwardsAlignmentPath[0][indexOfBackwardsPath], backwardsAlignmentPath[1][indexOfBackwardsPath]);
 		
 	}
-	printf("final index of backwards path is %i and i is %i \n", backwardsAlignmentPath[0].size()-1, indexOfBackwardsPath);
+	printf("final index of backwards path is %i and i is %i \n", (int) backwardsAlignmentPath[0].size()-1, indexOfBackwardsPath);
 	
 	//	backwardsAlignmentIndex = backwardsAlignmentPath[0].size()-1;//remember that this goes backwards!
 	
--- a/src/timeWarp.h	Tue May 17 08:48:58 2011 +0100
+++ b/src/timeWarp.h	Wed May 18 16:47:12 2011 +0100
@@ -74,9 +74,17 @@
 //	void drawSimilarityMatrix();
 //	void printSimilarityMatrix(int sizeToPrint);
 	
-	DoubleMatrix alignmentMeasureMatrix;	
+	Chromagram chromoGramm;
+	Chromagram secondChromoGramm;
+	
+	
+	DoubleMatrix alignmentMeasureMatrix;//just the choma alignment	
 	DoubleVector minimumAlignmentPath;
 	
+	//new addition
+	DoubleMatrix superAlignmentMeasureMatrix;	//for the onset + chromagram alignment
+	DoubleVector superMinimumAlignmentPath;	
+	
 //	void drawAlignmentPath(int startingChromaXFrame, int startingChromaYFrame);
 	int findStartWidthFrame();	  
 	
@@ -99,7 +107,7 @@
 	
 	void calculateAlignmentMatrix();
 	double getDistance(int i, int j);
-	void printAlignmentMatrix();
+//	void printAlignmentMatrix();
 	double getMinimum(int i, int j, float newValue);
 	bool extendAlignmentUp();
 	bool extendAlignmentAlong();
@@ -115,8 +123,6 @@
 	
 	
 	
-	Chromagram chromoGramm;
-	Chromagram secondChromoGramm;