forked from Mirrors/freeswitch
377 lines
13 KiB
C
377 lines
13 KiB
C
/***********************************************************************
|
|
Copyright (c) 2006-2011, Skype Limited. All rights reserved.
|
|
Redistribution and use in source and binary forms, with or without
|
|
modification, (subject to the limitations in the disclaimer below)
|
|
are permitted provided that the following conditions are met:
|
|
- Redistributions of source code must retain the above copyright notice,
|
|
this list of conditions and the following disclaimer.
|
|
- Redistributions in binary form must reproduce the above copyright
|
|
notice, this list of conditions and the following disclaimer in the
|
|
documentation and/or other materials provided with the distribution.
|
|
- Neither the name of Skype Limited, nor the names of specific
|
|
contributors, may be used to endorse or promote products derived from
|
|
this software without specific prior written permission.
|
|
NO EXPRESS OR IMPLIED LICENSES TO ANY PARTY'S PATENT RIGHTS ARE GRANTED
|
|
BY THIS LICENSE. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND
|
|
CONTRIBUTORS ''AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING,
|
|
BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
|
|
FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
|
|
COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
|
|
INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
|
|
NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF
|
|
USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
|
|
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
|
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
|
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
|
***********************************************************************/
|
|
|
|
/*
|
|
* Compare two audio signals and compute weighted SNR difference
|
|
*/
|
|
|
|
#ifdef _WIN32
|
|
#define _CRT_SECURE_NO_DEPRECATE 1
|
|
#endif
|
|
|
|
#include <stdio.h>
|
|
#include <stdlib.h>
|
|
#include <string.h>
|
|
#include <math.h>
|
|
|
|
#include "SKP_Silk_SigProc_FIX.h"
|
|
|
|
#define FRAME_LENGTH_MS 10
|
|
#define WIN_LENGTH_MS 20
|
|
#define BW_EXPANSION 0.7f
|
|
|
|
#define MAX_FS_KHZ 48
|
|
#define LPC_ORDER 10
|
|
#define SNR_THRESHOLD 15.0
|
|
|
|
#ifdef __cplusplus
|
|
extern "C"
|
|
{
|
|
#endif
|
|
/* Internally used functions */
|
|
void Autocorrelation(
|
|
SKP_float *results, /* o result (length correlationCount) */
|
|
const SKP_float *inputData, /* i input data to correlate */
|
|
SKP_int inputDataSize, /* i length of input */
|
|
SKP_int correlationCount /* i number of correlation taps to compute */
|
|
);
|
|
|
|
/* inner product of two SKP_float arrays, with result as double */
|
|
double Inner_product(
|
|
const SKP_float *data1,
|
|
const SKP_float *data2,
|
|
SKP_int dataSize
|
|
);
|
|
/* Solve the normal equations using the Levinson-Durbin recursion */
|
|
SKP_float Levinsondurbin( /* O prediction error energy */
|
|
SKP_float A[], /* O prediction coefficients [order] */
|
|
const SKP_float corr[], /* I input auto-correlations [order + 1] */
|
|
const SKP_int order /* I prediction order */
|
|
);
|
|
|
|
/* Chirp (bw expand) LP AR filter */
|
|
void Bwexpander(
|
|
SKP_float *ar, /* io AR filter to be expanded (without leading 1) */
|
|
const SKP_int d, /* i length of ar */
|
|
const SKP_float chirp /* i chirp factor (typically in range (0..1) ) */
|
|
);
|
|
|
|
#ifdef __cplusplus
|
|
}
|
|
#endif
|
|
|
|
static void print_usage(char* argv[]) {
|
|
printf("\nusage: %s ref.pcm test.pcm [settings]\n", argv[ 0 ]);
|
|
printf("\nref.pcm : Reference file");
|
|
printf("\ntest.pcm : File to be tested, should be of same length as ref.pcm");
|
|
printf("\n settings:");
|
|
printf("\n-diff : Only determine bit-exactness");
|
|
printf("\n-fs <Hz> : Sampling rate in Hz, max: %d; default: 48000", MAX_FS_KHZ * 1000 );
|
|
printf("\n");
|
|
}
|
|
|
|
|
|
int main(int argc, char* argv[])
|
|
{
|
|
SKP_int args, n, i, counterRef, counterTest;
|
|
char testInFileName[150], refInFileName[150];
|
|
FILE *refInFile, *testInFile;
|
|
SKP_int nFrames = 0, isUnequal = 0;
|
|
SKP_int diff = 0, Fs_kHz;
|
|
SKP_int32 Fs_Hz = 24000;
|
|
SKP_float c, refWhtnd, testWhtnd, refNrg, diffNrg;
|
|
double SNR = 0.0;
|
|
SKP_int16 refIn[WIN_LENGTH_MS * MAX_FS_KHZ], testIn[WIN_LENGTH_MS * MAX_FS_KHZ];
|
|
SKP_float refWin[WIN_LENGTH_MS * MAX_FS_KHZ], testWin[WIN_LENGTH_MS * MAX_FS_KHZ];
|
|
SKP_float autoCorr[LPC_ORDER + 1], LPC_Coef[LPC_ORDER];
|
|
|
|
if (argc < 3) {
|
|
print_usage(argv);
|
|
exit(0);
|
|
}
|
|
|
|
/* get arguments */
|
|
args = 1;
|
|
strcpy(refInFileName, argv[args]);
|
|
args++;
|
|
strcpy(testInFileName, argv[args]);
|
|
args++;
|
|
while(args < argc ) {
|
|
if (SKP_STR_CASEINSENSITIVE_COMPARE(argv[args], "-diff") == 0) {
|
|
diff = 1;
|
|
args++;
|
|
}else if (SKP_STR_CASEINSENSITIVE_COMPARE(argv[args], "-fs") == 0) {
|
|
sscanf(argv[args+1], "%d", &Fs_Hz);
|
|
args += 2;
|
|
} else {
|
|
printf("Error: unrecognized setting: %s\n\n", argv[args]);
|
|
print_usage(argv);
|
|
exit(0);
|
|
}
|
|
}
|
|
|
|
Fs_kHz = SKP_DIV32_16( Fs_Hz, 1000 );
|
|
|
|
if( Fs_kHz > MAX_FS_KHZ ) {
|
|
printf("Error: sampling rate too high: %d\n\n", Fs_kHz);
|
|
print_usage(argv);
|
|
exit(0);
|
|
}
|
|
|
|
printf("Reference: %s\n", refInFileName);
|
|
//printf("Test: %s\n", testInFileName);
|
|
|
|
/* open files */
|
|
refInFile = fopen(refInFileName, "rb");
|
|
if (refInFile==NULL) {
|
|
printf("Error: could not open input file %s\n", refInFileName);
|
|
exit(0);
|
|
}
|
|
testInFile = fopen(testInFileName, "rb");
|
|
if (testInFile==NULL) {
|
|
printf("Error: could not open input file %s\n", testInFileName);
|
|
exit(0);
|
|
}
|
|
|
|
SKP_memset( refIn, 0, sizeof(refIn) );
|
|
SKP_memset( testIn, 0, sizeof(testIn) );
|
|
|
|
while(1) {
|
|
/* Read inputs */
|
|
counterRef = (SKP_int)fread(&refIn[(WIN_LENGTH_MS - FRAME_LENGTH_MS) * Fs_kHz],
|
|
sizeof(SKP_int16), FRAME_LENGTH_MS * Fs_kHz, refInFile);
|
|
counterTest = (SKP_int)fread(&testIn[(WIN_LENGTH_MS - FRAME_LENGTH_MS) * Fs_kHz],
|
|
sizeof(SKP_int16), FRAME_LENGTH_MS * Fs_kHz, testInFile);
|
|
if(counterRef != FRAME_LENGTH_MS * Fs_kHz || counterTest != FRAME_LENGTH_MS * Fs_kHz){
|
|
break;
|
|
}
|
|
|
|
/* test for bit-exactness */
|
|
for( n = 0; n < FRAME_LENGTH_MS * Fs_kHz; n++ ) {
|
|
if( refIn[(WIN_LENGTH_MS - FRAME_LENGTH_MS) * Fs_kHz + n] !=
|
|
testIn[(WIN_LENGTH_MS - FRAME_LENGTH_MS) * Fs_kHz + n] ) {
|
|
isUnequal = 1;
|
|
break;
|
|
}
|
|
}
|
|
|
|
/* apply sine window */
|
|
for( n = 0; n < WIN_LENGTH_MS * Fs_kHz; n++ ) {
|
|
c = (SKP_float)sin( 3.14159265 * (n + 1) / (WIN_LENGTH_MS * Fs_kHz + 1) );
|
|
refWin[n] = refIn[n] * c;
|
|
testWin[n] = testIn[n] * c;
|
|
}
|
|
|
|
/* LPC analysis on reference signal */
|
|
|
|
/* Calculate auto correlation */
|
|
Autocorrelation(autoCorr, refWin, WIN_LENGTH_MS * Fs_kHz, LPC_ORDER + 1);
|
|
|
|
/* Add white noise */
|
|
autoCorr[ 0 ] += autoCorr[ 0 ] * 1e-6f + 1.0f;
|
|
|
|
/* Convert correlations to prediction coefficients */
|
|
Levinsondurbin(LPC_Coef, autoCorr, LPC_ORDER);
|
|
|
|
/* Bandwdith expansion */
|
|
Bwexpander(LPC_Coef, LPC_ORDER, BW_EXPANSION);
|
|
|
|
/* Filter both signals */
|
|
refNrg = 1.0f;
|
|
diffNrg = 1e-10f;
|
|
for( n = (WIN_LENGTH_MS - FRAME_LENGTH_MS) / 2 * Fs_kHz;
|
|
n < (WIN_LENGTH_MS + FRAME_LENGTH_MS) / 2 * Fs_kHz; n++ ) {
|
|
refWhtnd = refIn[n];
|
|
testWhtnd = testIn[n];
|
|
for( i = 0; i < LPC_ORDER; i++ ) {
|
|
refWhtnd -= LPC_Coef[ i ] * refIn[n - i - 1];
|
|
testWhtnd -= LPC_Coef[ i ] * testIn[n - i - 1];
|
|
}
|
|
refNrg += refWhtnd * refWhtnd;
|
|
diffNrg += (refWhtnd - testWhtnd) * (refWhtnd - testWhtnd);
|
|
}
|
|
|
|
/* weighted SNR */
|
|
if( refNrg > FRAME_LENGTH_MS * Fs_kHz ) {
|
|
SNR += 10.0 * log10( refNrg / diffNrg );
|
|
nFrames++;
|
|
}
|
|
|
|
/* Update Buffer */
|
|
SKP_memmove( refIn, &refIn[FRAME_LENGTH_MS * Fs_kHz], (WIN_LENGTH_MS - FRAME_LENGTH_MS) * Fs_kHz * sizeof(SKP_int16));
|
|
SKP_memmove( testIn, &testIn[FRAME_LENGTH_MS * Fs_kHz], (WIN_LENGTH_MS - FRAME_LENGTH_MS) * Fs_kHz * sizeof(SKP_int16));
|
|
}
|
|
|
|
if( diff ) {
|
|
if( isUnequal ) {
|
|
printf("Signals differ\n");
|
|
} else {
|
|
if(counterRef != counterTest){
|
|
printf("Warning: signals differ in length\n");
|
|
}
|
|
printf("Signals are bit-exact PASS\n");
|
|
}
|
|
} else {
|
|
if( nFrames == 0 ) {
|
|
printf("At least one signal too short or not loud enough\n");
|
|
exit(0);
|
|
}
|
|
if(counterRef != counterTest){
|
|
printf("Warning: signals differ in length\n");
|
|
}
|
|
if( isUnequal == 0 ) {
|
|
printf("Signals are bit-exact PASS\n");
|
|
} else {
|
|
printf("Average weighted SNR: %4.1f dB ", SNR / nFrames);
|
|
if( SNR / nFrames < SNR_THRESHOLD ) {
|
|
printf("FAIL\n");
|
|
} else {
|
|
printf("PASS\n");
|
|
}
|
|
}
|
|
}
|
|
printf("\n");
|
|
|
|
/* Close Files */
|
|
fclose(refInFile);
|
|
fclose(testInFile);
|
|
|
|
return 0;
|
|
}
|
|
|
|
/* compute autocorrelation */
|
|
void Autocorrelation(
|
|
SKP_float *results, /* o result (length correlationCount) */
|
|
const SKP_float *inputData, /* i input data to correlate */
|
|
SKP_int inputDataSize, /* i length of input */
|
|
SKP_int correlationCount /* i number of correlation taps to compute */
|
|
)
|
|
{
|
|
SKP_int i;
|
|
|
|
if (correlationCount > inputDataSize) {
|
|
correlationCount = inputDataSize;
|
|
}
|
|
|
|
for( i = 0; i < correlationCount; i++ ) {
|
|
results[ i ] = (SKP_float)Inner_product( inputData, inputData + i, inputDataSize - i );
|
|
}
|
|
}
|
|
|
|
/* inner product of two SKP_float arrays, with result as double */
|
|
double Inner_product(
|
|
const SKP_float *data1,
|
|
const SKP_float *data2,
|
|
SKP_int dataSize
|
|
)
|
|
{
|
|
SKP_int i, dataSize4;
|
|
double result;
|
|
|
|
/* 4x unrolled loop */
|
|
result = 0.0f;
|
|
dataSize4 = dataSize & 0xFFFC;
|
|
for( i = 0; i < dataSize4; i += 4 ) {
|
|
result += data1[ i + 0 ] * data2[ i + 0 ] +
|
|
data1[ i + 1 ] * data2[ i + 1 ] +
|
|
data1[ i + 2 ] * data2[ i + 2 ] +
|
|
data1[ i + 3 ] * data2[ i + 3 ];
|
|
}
|
|
|
|
/* add any remaining products */
|
|
for( ; i < dataSize; i++ ) {
|
|
result += data1[ i ] * data2[ i ];
|
|
}
|
|
|
|
return result;
|
|
}
|
|
|
|
/* Solve the normal equations using the Levinson-Durbin recursion */
|
|
SKP_float Levinsondurbin( /* O prediction error energy */
|
|
SKP_float A[], /* O prediction coefficients [order] */
|
|
const SKP_float corr[], /* I input auto-correlations [order + 1] */
|
|
const SKP_int order /* I prediction order */
|
|
)
|
|
{
|
|
SKP_int i, mHalf, m;
|
|
SKP_float min_nrg, nrg, t, km, Atmp1, Atmp2;
|
|
|
|
min_nrg = 1e-12f * corr[ 0 ] + 1e-9f;
|
|
nrg = corr[ 0 ];
|
|
nrg = SKP_max(min_nrg, nrg);
|
|
A[ 0 ] = corr[ 1 ] / nrg;
|
|
nrg -= A[ 0 ] * corr[ 1 ];
|
|
nrg = SKP_max(min_nrg, nrg);
|
|
|
|
for( m = 1; m < order; m++ )
|
|
{
|
|
t = corr[ m + 1 ];
|
|
for( i = 0; i < m; i++ ) {
|
|
t -= A[ i ] * corr[ m - i ];
|
|
}
|
|
|
|
/* reflection coefficient */
|
|
km = t / nrg;
|
|
|
|
/* residual energy */
|
|
nrg -= km * t;
|
|
nrg = SKP_max(min_nrg, nrg);
|
|
|
|
mHalf = m >> 1;
|
|
for( i = 0; i < mHalf; i++ ) {
|
|
Atmp1 = A[ i ];
|
|
Atmp2 = A[ m - i - 1 ];
|
|
A[ m - i - 1 ] -= km * Atmp1;
|
|
A[ i ] -= km * Atmp2;
|
|
}
|
|
if( m & 1 ) {
|
|
A[ mHalf ] -= km * A[ mHalf ];
|
|
}
|
|
A[ m ] = km;
|
|
}
|
|
|
|
/* return the residual energy */
|
|
return nrg;
|
|
}
|
|
|
|
/* Chirp (bw expand) LP AR filter */
|
|
void Bwexpander(
|
|
SKP_float *ar, /* io AR filter to be expanded (without leading 1) */
|
|
const SKP_int d, /* i length of ar */
|
|
const SKP_float chirp /* i chirp factor (typically in range (0..1) ) */
|
|
)
|
|
{
|
|
SKP_int i;
|
|
SKP_float cfac = chirp;
|
|
|
|
for( i = 0; i < d - 1; i++ ) {
|
|
ar[ i ] *= cfac;
|
|
cfac *= chirp;
|
|
}
|
|
ar[ d - 1 ] *= cfac;
|
|
}
|