/* ** Copyright (C) 2002-2015 Erik de Castro Lopo ** ** This program is free software; you can redistribute it and/or modify ** it under the terms of the GNU General Public License as published by ** the Free Software Foundation; either version 2 of the License, or ** (at your option) any later version. ** ** This program is distributed in the hope that it will be useful, ** but WITHOUT ANY WARRANTY; without even the implied warranty of ** MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the ** GNU General Public License for more details. ** ** You should have received a copy of the GNU General Public License ** along with this program; if not, write to the Free Software ** Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ #include #include #include #include "dft_cmp.h" #include "utils.h" #ifndef M_PI #define M_PI 3.14159265358979323846264338 #endif #define DFT_SPEC_LENGTH (DFT_DATA_LENGTH / 2) static void dft_magnitude (const double *data, double *spectrum) ; static double calc_max_spectral_difference (const double *spec1, const double *spec2) ; /*-------------------------------------------------------------------------------- ** Public functions. */ double dft_cmp_float (int linenum, const float *in_data, const float *test_data, int len, double target_snr, int allow_exit) { static double orig [DFT_DATA_LENGTH] ; static double test [DFT_DATA_LENGTH] ; unsigned k ; if (len != DFT_DATA_LENGTH) { printf ("Error (line %d) : dft_cmp_float : Bad input array length.\n", linenum) ; return 1 ; } ; for (k = 0 ; k < ARRAY_LEN (orig) ; k++) { test [k] = test_data [k] ; orig [k] = in_data [k] ; } ; return dft_cmp_double (linenum, orig, test, len, target_snr, allow_exit) ; } /* dft_cmp_float */ double dft_cmp_double (int linenum, const double *orig, const double *test, int len, double target_snr, int allow_exit) { static double orig_spec [DFT_SPEC_LENGTH] ; static double test_spec [DFT_SPEC_LENGTH] ; double snr ; if (! orig || ! test) { printf ("Error (line %d) : dft_cmp_double : Bad input arrays.\n", linenum) ; return 1 ; } ; if (len != DFT_DATA_LENGTH) { printf ("Error (line %d) : dft_cmp_double : Bad input array length.\n", linenum) ; return 1 ; } ; dft_magnitude (orig, orig_spec) ; dft_magnitude (test, test_spec) ; snr = calc_max_spectral_difference (orig_spec, test_spec) ; if (snr > target_snr) { printf ("\n\nLine %d: Actual SNR (% 4.1f) > target SNR (% 4.1f).\n\n", linenum, snr, target_snr) ; oct_save_double (orig, test, len) ; if (allow_exit) exit (1) ; } ; if (snr < -500.0) snr = -500.0 ; return snr ; } /* dft_cmp_double */ /*-------------------------------------------------------------------------------- ** Quick dirty calculation of magnitude spectrum for real valued data using ** Discrete Fourier Transform. Since the data is real, the DFT is only ** calculated for positive frequencies. */ static void dft_magnitude (const double *data, double *spectrum) { static double cos_angle [DFT_DATA_LENGTH] = { 0.0 } ; static double sin_angle [DFT_DATA_LENGTH] ; double real_part, imag_part ; int k, n ; /* If sine and cosine tables haven't been initialised, do so. */ if (cos_angle [0] == 0.0) for (n = 0 ; n < DFT_DATA_LENGTH ; n++) { cos_angle [n] = cos (2.0 * M_PI * n / DFT_DATA_LENGTH) ; sin_angle [n] = -1.0 * sin (2.0 * M_PI * n / DFT_DATA_LENGTH) ; } ; /* DFT proper. Since the data is real, only generate a half spectrum. */ for (k = 1 ; k < DFT_SPEC_LENGTH ; k++) { real_part = 0.0 ; imag_part = 0.0 ; for (n = 0 ; n < DFT_DATA_LENGTH ; n++) { real_part += data [n] * cos_angle [(k * n) % DFT_DATA_LENGTH] ; imag_part += data [n] * sin_angle [(k * n) % DFT_DATA_LENGTH] ; } ; spectrum [k] = sqrt (real_part * real_part + imag_part * imag_part) ; } ; spectrum [DFT_SPEC_LENGTH - 1] = 0.0 ; spectrum [0] = spectrum [1] = spectrum [2] = 0.0 ; return ; } /* dft_magnitude */ static double calc_max_spectral_difference (const double *orig, const double *test) { double orig_max = 0.0, max_diff = 0.0 ; int k ; for (k = 0 ; k < DFT_SPEC_LENGTH ; k++) { if (orig_max < orig [k]) orig_max = orig [k] ; if (max_diff < fabs (orig [k] - test [k])) max_diff = fabs (orig [k] - test [k]) ; } ; if (max_diff < 1e-25) return -500.0 ; return 20.0 * log10 (max_diff / orig_max) ; } /* calc_max_spectral_difference */