blob: 08536734fdc1fa387ab7fa8b16e41716c6192c64 [file] [log] [blame]
/*
* Copyright (C) 2021 The Android Open Source Project
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include <iostream>
#include <vector>
constexpr int kMinLoopLimitValue = 1;
constexpr int kNumPeaks = 3;
/*!
\brief Compute the length normalized correlation of two signals
\sigX Pointer to signal 1
\sigY Pointer to signal 2
\len Length of signals
\enableCrossCorr Flag to be set to 1 if cross-correlation is needed
\return First value is vector of correlation peak indices
Second value is vector of correlation peak values
*/
static std::pair<std::vector<int>, std::vector<float>> correlation(const int16_t* sigX,
const int16_t* sigY, int len,
int16_t enableCrossCorr) {
float maxCorrVal = 0.f, prevCorrVal = 0.f;
int peakIndex = 0, flag = 0;
int loopLim = (1 == enableCrossCorr) ? len : kMinLoopLimitValue;
std::vector<int> peakIndexVect(kNumPeaks, 0);
std::vector<float> peakValueVect(kNumPeaks, 0.f);
for (int i = 0; i < loopLim; i++) {
float corrVal = 0.f;
for (int j = i; j < len; j++) {
corrVal += (float)(sigX[j] * sigY[j - i]);
}
corrVal /= len - i;
if (corrVal > maxCorrVal) {
maxCorrVal = corrVal;
}
// Correlation peaks are expected to be observed at equal intervals. The interval length is
// expected to match with wave period.
// The following block of code saves the first kNumPeaks number of peaks and the index at
// which they occur.
if (peakIndex < kNumPeaks) {
if (corrVal > prevCorrVal) {
peakIndexVect[peakIndex] = i;
peakValueVect[peakIndex] = corrVal;
flag = 0;
} else if (0 == flag) {
peakIndex++;
flag = 1;
}
}
if (peakIndex == kNumPeaks) break;
prevCorrVal = corrVal;
}
return {peakIndexVect, peakValueVect};
}
void printUsage() {
printf("\nUsage: ");
printf("\n correlation <firstFile> <secondFile> [enableCrossCorr]\n");
printf("\nwhere, \n <firstFile> is the first file name");
printf("\n <secondFile> is the second file name");
printf("\n [enableCrossCorr] is flag to set for cross-correlation (Default 1)\n\n");
}
int main(int argc, const char* argv[]) {
if (argc < 3) {
printUsage();
return EXIT_FAILURE;
}
std::unique_ptr<FILE, decltype(&fclose)> fInput1(fopen(argv[1], "rb"), &fclose);
if (fInput1.get() == NULL) {
printf("\nError: missing file %s\n", argv[1]);
return EXIT_FAILURE;
}
std::unique_ptr<FILE, decltype(&fclose)> fInput2(fopen(argv[2], "rb"), &fclose);
if (fInput2.get() == NULL) {
printf("\nError: missing file %s\n", argv[2]);
return EXIT_FAILURE;
}
int16_t enableCrossCorr = (4 == argc) ? atoi(argv[3]) : 1;
fseek(fInput1.get(), 0L, SEEK_END);
unsigned int fileSize1 = ftell(fInput1.get());
rewind(fInput1.get());
fseek(fInput2.get(), 0L, SEEK_END);
unsigned int fileSize2 = ftell(fInput2.get());
rewind(fInput2.get());
if (fileSize1 != fileSize2) {
printf("\nError: File sizes different\n");
return EXIT_FAILURE;
}
size_t numFrames = fileSize1 / sizeof(int16_t);
std::unique_ptr<int16_t[]> inBuffer1(new int16_t[numFrames]());
std::unique_ptr<int16_t[]> inBuffer2(new int16_t[numFrames]());
if (numFrames != fread(inBuffer1.get(), sizeof(int16_t), numFrames, fInput1.get())) {
printf("\nError: Unable to read %zu samples from file %s\n", numFrames, argv[1]);
return EXIT_FAILURE;
}
if (numFrames != fread(inBuffer2.get(), sizeof(int16_t), numFrames, fInput2.get())) {
printf("\nError: Unable to read %zu samples from file %s\n", numFrames, argv[2]);
return EXIT_FAILURE;
}
auto pairAutoCorr1 = correlation(inBuffer1.get(), inBuffer1.get(), numFrames, enableCrossCorr);
auto pairAutoCorr2 = correlation(inBuffer2.get(), inBuffer2.get(), numFrames, enableCrossCorr);
// Following code block checks pitch period difference between two input signals. They must
// match as AGC applies only gain, no frequency related computation is done.
bool pitchMatch = false;
for (unsigned i = 0; i < pairAutoCorr1.first.size() - 1; i++) {
if (pairAutoCorr1.first[i + 1] - pairAutoCorr1.first[i] !=
pairAutoCorr2.first[i + 1] - pairAutoCorr2.first[i]) {
pitchMatch = false;
break;
}
pitchMatch = true;
}
if (pitchMatch) {
printf("Auto-correlation : Pitch matched\n");
} else {
printf("Auto-correlation : Pitch mismatch\n");
return EXIT_FAILURE;
}
if (enableCrossCorr) {
auto pairCrossCorr =
correlation(inBuffer1.get(), inBuffer2.get(), numFrames, enableCrossCorr);
// Since AGC applies only gain, the pitch information obtained from cross correlation data
// of input and output is expected to be same as the input signal's pitch information.
pitchMatch = false;
for (unsigned i = 0; i < pairCrossCorr.first.size() - 1; i++) {
if (pairAutoCorr1.first[i + 1] - pairAutoCorr1.first[i] !=
pairCrossCorr.first[i + 1] - pairCrossCorr.first[i]) {
pitchMatch = false;
break;
}
pitchMatch = true;
}
if (pitchMatch) {
printf("Cross-correlation : Pitch matched for AGC\n");
if (pairAutoCorr1.second[0]) {
printf("Expected gain : (maxCrossCorr / maxAutoCorr1) = %f\n",
pairCrossCorr.second[0] / pairAutoCorr1.second[0]);
}
} else {
printf("Cross-correlation : Pitch mismatch\n");
return EXIT_FAILURE;
}
}
return EXIT_SUCCESS;
}