/* * Musepack audio compression * Copyright (c) 2005-2009, The Musepack Development Team * Copyright (C) 1999-2004 Buschmann/Klemm/Piecha/Wolf * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public * License as published by the Free Software Foundation; either * version 2.1 of the License, or (at your option) any later version. * * This library 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 * Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public * License along with this library; if not, write to the Free Software * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ #include "libmpcpsy.h" #include #include /* * Klemm 1994 and 1997. Experimental data. Sorry, data looks a little bit * dodderly. Data below 30 Hz is extrapolated from other material, above 18 * kHz the ATH is limited due to the original purpose (too much noise at * ATH is not good even if it's theoretically inaudible). */ // FIXME : move this in a struct, to eliminate multiple declarations // w_low for long 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 const int wl [PART_LONG] = { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17, 19, 21, 23, 25, 27, 29, 31, 33, 35, 38, 41, 44, 47, 50, 54, 58, 62, 67, 72, 78, 84, 91, 98, 106, 115, 124, 134, 145, 157, 170, 184, 199, 216, 234, 254, 276, 301, 329, 360, 396, 437, 485 }; const int wh [PART_LONG] = { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30, 32, 34, 37, 40, 43, 46, 49, 53, 57, 61, 66, 71, 77, 83, 90, 97, 105, 114, 123, 133, 144, 156, 169, 183, 198, 215, 233, 253, 275, 300, 328, 359, 395, 436, 484, 511 }; // Width: 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2 3 3 3 3 3 4 4 4 5 5 6 6 7 7 8 9 9 10 11 12 13 14 15 17 18 20 22 25 28 31 36 41 48 27 // inverse partition-width for long const float iw [PART_LONG] = { 1.f, 1.f, 1.f, 1.f, 1.f, 1.f, 1.f, 1.f, 1.f, 1.f, 1.f, 1.f/2, 1.f/2, 1.f/2, 1.f/2, 1.f/2, 1.f/2, 1.f/2, 1.f/2, 1.f/2, 1.f/2, 1.f/2, 1.f/2, 1.f/3, 1.f/3, 1.f/3, 1.f/3, 1.f/3, 1.f/4, 1.f/4, 1.f/4, 1.f/5, 1.f/5, 1.f/6, 1.f/6, 1.f/7, 1.f/7, 1.f/8, 1.f/9, 1.f/9, 1.f/10, 1.f/11, 1.f/12, 1.f/13, 1.f/14, 1.f/15, 1.f/17, 1.f/18, 1.f/20, 1.f/22, 1.f/25, 1.f/28, 1.f/31, 1.f/36, 1.f/41, 1.f/48, 1.f/27 }; // w_low for short 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 const int wl_short [PART_SHORT] = { 0, 1, 2, 3, 4, 5, 6, 8, 10, 12, 15, 18, 23, 29, 36, 46, 59, 75, 99 }; const int wh_short [PART_SHORT] = { 0, 1, 2, 3, 5, 6, 7, 9, 12, 14, 18, 23, 29, 36, 46, 58, 75, 99, 127 }; // inverse partition-width for short const float iw_short [PART_SHORT] = { 1.f, 1.f, 1.f, 1.f, 1.f/2, 1.f/2, 1.f/2, 1.f/2, 1.f/3, 1.f/3, 1.f/4, 1.f/6, 1.f/7, 1.f/8, 1.f/11, 1.f/13, 1.f/17, 1.f/25, 1.f/29 }; float MinVal [PART_LONG]; // contains minimum tonality soffsets float Loudness [PART_LONG]; // weighting factors for loudness calculation float SPRD [PART_LONG] [PART_LONG]; // tabulated spreading function float O_MAX; float O_MIN; float FAC1; float FAC2; // constants for offset calculation float partLtq [PART_LONG]; // threshold in quiet (partitions) float invLtq [PART_LONG]; // inverse threshold in quiet (partitions, long) float fftLtq [512]; // threshold in quiet (FFT) static float ATHformula_Frank ( float freq ) { /* * one value per 100 cent = 1 * semitone = 1/4 * third = 1/12 * octave = 1/40 decade * rest is linear interpolated, values are currently in millibel rel. 20 Pa */ static short tab [] = { /* 10.0 */ 9669, 9669, 9626, 9512, /* 12.6 */ 9353, 9113, 8882, 8676, /* 15.8 */ 8469, 8243, 7997, 7748, /* 20.0 */ 7492, 7239, 7000, 6762, /* 25.1 */ 6529, 6302, 6084, 5900, /* 31.6 */ 5717, 5534, 5351, 5167, /* 39.8 */ 5004, 4812, 4638, 4466, /* 50.1 */ 4310, 4173, 4050, 3922, /* 63.1 */ 3723, 3577, 3451, 3281, /* 79.4 */ 3132, 3036, 2902, 2760, /* 100.0 */ 2658, 2591, 2441, 2301, /* 125.9 */ 2212, 2125, 2018, 1900, /* 158.5 */ 1770, 1682, 1594, 1512, /* 199.5 */ 1430, 1341, 1260, 1198, /* 251.2 */ 1136, 1057, 998, 943, /* 316.2 */ 887, 846, 744, 712, /* 398.1 */ 693, 668, 637, 606, /* 501.2 */ 580, 555, 529, 502, /* 631.0 */ 475, 448, 422, 398, /* 794.3 */ 375, 351, 327, 322, /* 1000.0 */ 312, 301, 291, 268, /* 1258.9 */ 246, 215, 182, 146, /* 1584.9 */ 107, 61, 13, -35, /* 1995.3 */ -96, -156, -179, -235, /* 2511.9 */ -295, -350, -401, -421, /* 3162.3 */ -446, -499, -532, -535, /* 3981.1 */ -513, -476, -431, -313, /* 5011.9 */ -179, 8, 203, 403, /* 6309.6 */ 580, 736, 881, 1022, /* 7943.3 */ 1154, 1251, 1348, 1421, /* 10000.0 */ 1479, 1399, 1285, 1193, /* 12589.3 */ 1287, 1519, 1914, 2369, #if 0 /* 15848.9 */ 3352, 4865, 5942, 6177, /* 19952.6 */ 6385, 6604, 6833, 7009, /* 25118.9 */ 7066, 7127, 7191, 7260, #else /* 15848.9 */ 3352, 4352, 5352, 6352, /* 19952.6 */ 7352, 8352, 9352, 9999, /* 25118.9 */ 9999, 9999, 9999, 9999, #endif }; double freq_log; unsigned index; if ( freq < 10. ) freq = 10.; if ( freq > 29853. ) freq = 29853.; freq_log = 40. * log10 (0.1 * freq); /* 4 steps per third, starting at 10 Hz */ index = (unsigned) freq_log; return 0.01 * (tab [index] * (1 + index - freq_log) + tab [index+1] * (freq_log - index)); } /* F U N C T I O N S */ // calculation of the threshold in quiet in FFT-resolution static void Ruhehoerschwelle ( PsyModel* m, unsigned int EarModelFlag, int Ltq_offset, int Ltq_max ) { int n; int k; float f; float erg; double tmp; float absLtq [512]; for ( n = 0; n < 512; n++ ) { f = (float) ( (n+1) * (float)(m->SampleFreq / 2000.) / 512 ); // Frequency in kHz switch ( EarModelFlag / 100 ) { case 0: // ISO-threshold in quiet tmp = 3.64*pow (f,-0.8) - 6.5*exp (-0.6*(f-3.3)*(f-3.3)) + 0.001*pow (f, 4.0); break; default: case 1: // measured threshold in quiet (Nick Berglmeir, Andree Buschmann, Kopfh�er) tmp = 3.00*pow (f,-0.8) - 5.0*exp (-0.1*(f-3.0)*(f-3.0)) + 0.0000015022693846297*pow (f, 6.0) + 10.*exp (-(f-0.1)*(f-0.1)); break; case 2: // measured threshold in quiet (Filburt, Kopfh�er) tmp = 9.00*pow (f,-0.5) - 15.0*exp (-0.1*(f-4.0)*(f-4.0)) + 0.0341796875*pow (f, 2.5) + 15.*exp (-(f-0.1)*(f-0.1)) - 18; tmp = mind ( tmp, Ltq_max - 18 ); break; case 3: tmp = ATHformula_Frank ( 1.e3 * f ); break; case 4: tmp = ATHformula_Frank ( 1.e3 * f ); if ( f > 4.8 ) { tmp += 3.00*pow (f,-0.8) - 5.0*exp (-0.1*(f-3.0)*(f-3.0)) + 0.0000015022693846297*pow (f, 6.0) + 10.*exp (-(f-0.1)*(f-0.1)); tmp *= 0.5 ; } break; case 5: tmp = ATHformula_Frank ( 1.e3 * f ); if ( f > 4.8 ) { tmp = 3.00*pow (f,-0.8) - 5.0*exp (-0.1*(f-3.0)*(f-3.0)) + 0.0000015022693846297*pow (f, 6.0) + 10.*exp (-(f-0.1)*(f-0.1)); } break; } tmp -= f * f * (int)(EarModelFlag % 100 - 50) * 0.0015; // 00: +30 dB, 100: -30 dB @20 kHz tmp = mind ( tmp, Ltq_max ); // Limit ATH tmp += Ltq_offset - 23; // Add chosen Offset fftLtq[n] = absLtq[n] = POW10 ( 0.1 * tmp); // conversion into power } // threshold in quiet in partitions (long) for ( n = 0; n < PART_LONG; n++ ) { erg = 1.e20f; for ( k = wl[n]; k <= wh[n]; k++ ) erg = minf (erg, absLtq[k]); partLtq[n] = erg; // threshold in quiet invLtq [n] = 1.f / partLtq[n]; // Inverse } } #ifdef _MSC_VER static double asinh ( double x ) { return x >= 0 ? log (sqrt (x*x+1) + x) : -log (sqrt (x*x+1) - x); } #endif static double Freq2Bark ( double Hz ) // Klemm 2002 { return 9.97074*asinh (1.1268e-3 * Hz) - 6.25817*asinh (0.197193e-3 * Hz) ; } // static double // Bark2Freq ( double Bark ) // Klemm 2002 // { // return 956.86 * sinh (0.101561*Bark) + 11.7296 * sinh (0.304992*Bark) + 6.33622e-3*sinh (0.538621*Bark); // } static double LongPart2Bark ( PsyModel* m, int Part ) { return Freq2Bark ((wl [Part] + wh [Part]) * m->SampleFreq / 2048.); } // calculating the table for loudness calculation based on absLtq = ank static void Loudness_Tabelle (PsyModel* m) { int n; float midfreq; float tmp; // ca. dB(A) for ( n = 0; n < PART_LONG; n++ ){ midfreq = (wh[n] + wl[n] + 3) * (0.25 * m->SampleFreq / 512); // center frequency in kHz, why +3 ??? tmp = LOG10 (midfreq) - 3.5f; // dB(A) tmp = -10 * tmp * tmp + 3 - midfreq/3000; Loudness [n] = POW10 ( 0.1 * tmp ); // conversion into power } } static double Bass ( float f, float TMN, float NMT, float bass ) { static unsigned char lfe [11] = { 120, 100, 80, 60, 50, 40, 30, 20, 15, 10, 5 }; int tmp = (int) ( 1024/44100. * f + 0.5 ); switch ( tmp ) { case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7: case 8: case 9: case 10: return TMN + bass * lfe [tmp]; case 11: case 12: case 13: case 14: case 15: case 16: case 17: case 18: return TMN; case 19: case 20: case 21: case 22: return TMN*0.75 + NMT*0.25; case 23: case 24: return TMN*0.50 + NMT*0.50; case 25: case 26: return TMN*0.25 + NMT*0.75; default: return NMT; } } // calculating the coefficient for utilization of the tonality offset, depending on TMN und NMT static void Tonalitaetskoeffizienten ( PsyModel* m ) { double tmp; int n; float bass; bass = 0.1/8 * m->NMT; if ( m->MinValChoice <= 2 && bass > 0.1 ) bass = 0.1f; if ( m->MinValChoice <= 1 ) bass = 0.0f; // alternative: calculation of the minval-values dependent on TMN and TMN for ( n = 0; n < PART_LONG; n++ ) { tmp = Bass ( (wl [n] + wh [n]) / 2048. * m->SampleFreq, m->TMN, m->NMT, bass ); MinVal [n] = POW10 ( -0.1 * tmp ); // conversion into power } // calculation of the constants for "tonality offset" O_MAX = POW10 ( -0.1 * m->TMN ); O_MIN = POW10 ( -0.1 * m->NMT ); FAC1 = POW10 ( -0.1 * (m->NMT - (m->TMN - m->NMT) * 0.229) ) ; FAC2 = (m->TMN - m->NMT) * (0.99011159 * 0.1); } // calculation of the spreading function static void Spread ( PsyModel* m ) { int i; int j; float tmpx; float tmpy; float tmpz; float x; // calculation of the spreading-function for all occuring values for ( i = 0; i < PART_LONG; i++ ) { // i is masking Partition, Source for ( j = 0; j < PART_LONG; j++ ) { // j is masking Partition, Target tmpx = LongPart2Bark (m, j) - LongPart2Bark (m, i);// Difference of the partitions in Bark tmpy = tmpz = 0.; // tmpz = 0: no dip if ( tmpx < 0 ) { // downwards (S1) tmpy = -32.f * tmpx; // 32 dB per Bark, e33 (10) } else if ( tmpx > 0 ) { // upwards (S2) #if 0 x = (wl[i]+wh[i])/2 * (float)(SampleFreq / 2000)/512; // center frequency in kHz ??????? if (i==0) x = 0.5f * (float)(SampleFreq / 2000)/512; // if first spectral line #else x = i ? wl[i]+wh[i] : 1; x *= m->SampleFreq / 1000. / 2048; // center frequency in kHz #endif // dB/Bark tmpy = (22.f + 0.23f / x) * tmpx; // e33 (10) // dip (up to 6 dB) tmpz = 8 * minf ( (tmpx-0.5f) * (tmpx-0.5f) - 2 * (tmpx-0.5f), 0.f ); } // calculate coefficient SPRD[i][j] = POW10 ( -0.1 * (tmpy+tmpz) ); // [Source] [Target] } } // Normierung e33 (10) for ( i = 0; i < PART_LONG; i++ ) { // i is masked Partition float norm = 0.f; for ( j = 0; j < PART_LONG; j++ ) // j is masking Partition norm += SPRD [j] [i]; for ( j = 0; j < PART_LONG; j++ ) // j is masking Partition SPRD [j] [i] /= norm; } } // call all initialisation procedures void Init_Psychoakustiktabellen ( PsyModel* m ) { m->Max_Band = (int) ( m->BandWidth * 64. / m->SampleFreq ); if ( m->Max_Band < 1 ) m->Max_Band = 1; if ( m->Max_Band > 31 ) m->Max_Band = 31; Tonalitaetskoeffizienten (m); Ruhehoerschwelle ( m, m->EarModelFlag, m->Ltq_offset, m->Ltq_max ); Loudness_Tabelle (m); Spread (m); } /* end of psy_tab.c */