147 lines
3.1 KiB
C++
147 lines
3.1 KiB
C++
#include "Yin.h"
|
|
#include <stdlib.h>
|
|
//#include "WProgram.h"
|
|
|
|
void Yin::initialize(float yinSampleRate,int yinBufferSize){
|
|
bufferSize = yinBufferSize;
|
|
sampleRate = yinSampleRate;
|
|
halfBufferSize = bufferSize / 2;
|
|
threshold = 0.15;
|
|
probability = 0.0;
|
|
//initialize array and set it to zero
|
|
yinBuffer = (float *) malloc(sizeof(float)* halfBufferSize);
|
|
for(int i = 0; i < halfBufferSize; i++){
|
|
yinBuffer[i] = 0;
|
|
}
|
|
}
|
|
|
|
Yin::Yin(){
|
|
}
|
|
|
|
Yin::Yin(float yinSampleRate,int yinBufferSize){
|
|
initialize(yinSampleRate,yinBufferSize);
|
|
}
|
|
|
|
float Yin::getProbability(){
|
|
return probability;
|
|
}
|
|
|
|
float Yin::getPitch(float* buffer){
|
|
int tauEstimate = -1;
|
|
float pitchInHertz = -1;
|
|
|
|
//step 2
|
|
difference(buffer);
|
|
|
|
// step 3
|
|
cumulativeMeanNormalizedDifference();
|
|
|
|
//step 4
|
|
tauEstimate = absoluteThreshold();
|
|
|
|
//step 5
|
|
if(tauEstimate != -1){
|
|
|
|
pitchInHertz = sampleRate / parabolicInterpolation(tauEstimate);
|
|
}
|
|
|
|
return pitchInHertz;
|
|
}
|
|
|
|
float Yin::parabolicInterpolation(int tauEstimate) {
|
|
float betterTau;
|
|
int x0;
|
|
int x2;
|
|
|
|
if (tauEstimate < 1) {
|
|
x0 = tauEstimate;
|
|
}
|
|
else {
|
|
x0 = tauEstimate - 1;
|
|
}
|
|
if (tauEstimate + 1 < halfBufferSize) {
|
|
x2 = tauEstimate + 1;
|
|
}
|
|
else {
|
|
x2 = tauEstimate;
|
|
}
|
|
if (x0 == tauEstimate) {
|
|
if (yinBuffer[tauEstimate] <= yinBuffer[x2]) {
|
|
betterTau = tauEstimate;
|
|
}
|
|
else {
|
|
betterTau = x2;
|
|
}
|
|
}
|
|
else if (x2 == tauEstimate) {
|
|
if (yinBuffer[tauEstimate] <= yinBuffer[x0]) {
|
|
betterTau = tauEstimate;
|
|
}
|
|
else {
|
|
betterTau = x0;
|
|
}
|
|
}
|
|
else {
|
|
float s0, s1, s2;
|
|
s0 = yinBuffer[x0];
|
|
s1 = yinBuffer[tauEstimate];
|
|
s2 = yinBuffer[x2];
|
|
// fixed AUBIO implementation, thanks to Karl Helgason:
|
|
// (2.0f * s1 - s2 - s0) was incorrectly multiplied with -1
|
|
betterTau = tauEstimate + (s2 - s0) / (2 * (2 * s1 - s2 - s0));
|
|
}
|
|
return betterTau;
|
|
}
|
|
|
|
void Yin::cumulativeMeanNormalizedDifference(){
|
|
int tau;
|
|
yinBuffer[0] = 1;
|
|
float runningSum = 0;
|
|
for (tau = 1; tau < halfBufferSize; tau++) {
|
|
runningSum += yinBuffer[tau];
|
|
yinBuffer[tau] *= tau / runningSum;
|
|
}
|
|
}
|
|
|
|
void Yin::difference(float* buffer){
|
|
int index;
|
|
int tau;
|
|
float delta;
|
|
for(tau = 0 ; tau < halfBufferSize; tau++){
|
|
for(index = 0; index < halfBufferSize; index++){
|
|
delta= buffer[index] - buffer[index + tau];
|
|
yinBuffer[tau] += delta * delta;
|
|
}
|
|
}
|
|
}
|
|
|
|
int Yin::absoluteThreshold(){
|
|
int tau;
|
|
// first two positions in yinBuffer are always 1
|
|
// So start at the third (index 2)
|
|
for (tau = 2; tau < halfBufferSize ; tau++) {
|
|
if (yinBuffer[tau] < threshold) {
|
|
while (tau + 1 < halfBufferSize && yinBuffer[tau + 1] < yinBuffer[tau]) {
|
|
tau++;
|
|
}
|
|
// found tau, exit loop and return
|
|
// store the probability
|
|
// From the YIN paper: The threshold determines the list of
|
|
// candidates admitted to the set, and can be interpreted as the
|
|
// proportion of aperiodic power tolerated
|
|
// within a ëëperiodicíí signal.
|
|
//
|
|
// Since we want the periodicity and and not aperiodicity:
|
|
// periodicity = 1 - aperiodicity
|
|
probability = 1 - yinBuffer[tau];
|
|
break;
|
|
}
|
|
}
|
|
// if no pitch found, tau => -1
|
|
if (tau == halfBufferSize || yinBuffer[tau] >= threshold) {
|
|
tau = -1;
|
|
probability = 0;
|
|
}
|
|
return tau;
|
|
}
|