/* * psymodel.c * * Copyright (c) 1999-2000 Mark Taylor * Copyright (c) 2001-2002 Naoki Shibata * Copyright (c) 2000-2003 Takehiro Tominaga * Copyright (c) 2000-2011 Robert Hegemann * Copyright (c) 2000-2005 Gabriel Bouvigne * Copyright (c) 2000-2005 Alexander Leidinger * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Library General Public * License as published by the Free Software Foundation; either * version 2 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 * Library General Public License for more details. * * You should have received a copy of the GNU Library 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. */ /* $Id: psymodel.c,v 1.209 2011/05/24 20:45:55 robert Exp $ */ /* PSYCHO ACOUSTICS This routine computes the psycho acoustics, delayed by one granule. Input: buffer of PCM data (1024 samples). This window should be centered over the 576 sample granule window. The routine will compute the psycho acoustics for this granule, but return the psycho acoustics computed for the *previous* granule. This is because the block type of the previous granule can only be determined after we have computed the psycho acoustics for the following granule. Output: maskings and energies for each scalefactor band. block type, PE, and some correlation measures. The PE is used by CBR modes to determine if extra bits from the bit reservoir should be used. The correlation measures are used to determine mid/side or regular stereo. */ /* Notation: barks: a non-linear frequency scale. Mapping from frequency to barks is given by freq2bark() scalefactor bands: The spectrum (frequencies) are broken into SBMAX "scalefactor bands". Thes bands are determined by the MPEG ISO spec. In the noise shaping/quantization code, we allocate bits among the partition bands to achieve the best possible quality partition bands: The spectrum is also broken into about 64 "partition bands". Each partition band is about .34 barks wide. There are about 2-5 partition bands for each scalefactor band. LAME computes all psycho acoustic information for each partition band. Then at the end of the computations, this information is mapped to scalefactor bands. The energy in each scalefactor band is taken as the sum of the energy in all partition bands which overlap the scalefactor band. The maskings can be computed in the same way (and thus represent the average masking in that band) or by taking the minmum value multiplied by the number of partition bands used (which represents a minimum masking in that band). */ /* The general outline is as follows: 1. compute the energy in each partition band 2. compute the tonality in each partition band 3. compute the strength of each partion band "masker" 4. compute the masking (via the spreading function applied to each masker) 5. Modifications for mid/side masking. Each partition band is considiered a "masker". The strength of the i'th masker in band j is given by: s3(bark(i)-bark(j))*strength(i) The strength of the masker is a function of the energy and tonality. The more tonal, the less masking. LAME uses a simple linear formula (controlled by NMT and TMN) which says the strength is given by the energy divided by a linear function of the tonality. */ /* s3() is the "spreading function". It is given by a formula determined via listening tests. The total masking in the j'th partition band is the sum over all maskings i. It is thus given by the convolution of the strength with s3(), the "spreading function." masking(j) = sum_over_i s3(i-j)*strength(i) = s3 o strength where "o" = convolution operator. s3 is given by a formula determined via listening tests. It is normalized so that s3 o 1 = 1. Note: instead of a simple convolution, LAME also has the option of using "additive masking" The most critical part is step 2, computing the tonality of each partition band. LAME has two tonality estimators. The first is based on the ISO spec, and measures how predictiable the signal is over time. The more predictable, the more tonal. The second measure is based on looking at the spectrum of a single granule. The more peaky the spectrum, the more tonal. By most indications, the latter approach is better. Finally, in step 5, the maskings for the mid and side channel are possibly increased. Under certain circumstances, noise in the mid & side channels is assumed to also be masked by strong maskers in the L or R channels. Other data computed by the psy-model: ms_ratio side-channel / mid-channel masking ratio (for previous granule) ms_ratio_next side-channel / mid-channel masking ratio for this granule percep_entropy[2] L and R values (prev granule) of PE - A measure of how much pre-echo is in the previous granule percep_entropy_MS[2] mid and side channel values (prev granule) of percep_entropy energy[4] L,R,M,S energy in each channel, prev granule blocktype_d[2] block type to use for previous granule */ #ifdef HAVE_CONFIG_H # include #endif #include "lame.h" #include "machine.h" #include "encoder.h" #include "util.h" #include "psymodel.h" #include "lame_global_flags.h" #include "fft.h" #include "lame-analysis.h" #define NSFIRLEN 21 #ifdef M_LN10 #define LN_TO_LOG10 (M_LN10/10) #else #define LN_TO_LOG10 0.2302585093 #endif /* L3psycho_anal. Compute psycho acoustics. Data returned to the calling program must be delayed by one granule. This is done in two places. If we do not need to know the blocktype, the copying can be done here at the top of the program: we copy the data for the last granule (computed during the last call) before it is overwritten with the new data. It looks like this: 0. static psymodel_data 1. calling_program_data = psymodel_data 2. compute psymodel_data For data which needs to know the blocktype, the copying must be done at the end of this loop, and the old values must be saved: 0. static psymodel_data_old 1. compute psymodel_data 2. compute possible block type of this granule 3. compute final block type of previous granule based on #2. 4. calling_program_data = psymodel_data_old 5. psymodel_data_old = psymodel_data */ /* psycho_loudness_approx jd - 2001 mar 12 in: energy - BLKSIZE/2 elements of frequency magnitudes ^ 2 gfp - uses out_samplerate, ATHtype (also needed for ATHformula) returns: loudness^2 approximation, a positive value roughly tuned for a value of 1.0 for signals near clipping. notes: When calibrated, feeding this function binary white noise at sample values +32767 or -32768 should return values that approach 3. ATHformula is used to approximate an equal loudness curve. future: Data indicates that the shape of the equal loudness curve varies with intensity. This function might be improved by using an equal loudness curve shaped for typical playback levels (instead of the ATH, that is shaped for the threshold). A flexible realization might simply bend the existing ATH curve to achieve the desired shape. However, the potential gain may not be enough to justify an effort. */ static FLOAT psycho_loudness_approx(FLOAT const *energy, FLOAT const *eql_w) { int i; FLOAT loudness_power; loudness_power = 0.0; /* apply weights to power in freq. bands */ for (i = 0; i < BLKSIZE / 2; ++i) loudness_power += energy[i] * eql_w[i]; loudness_power *= VO_SCALE; return loudness_power; } /* mask_add optimization */ /* init the limit values used to avoid computing log in mask_add when it is not necessary */ /* For example, with i = 10*log10(m2/m1)/10*16 (= log10(m2/m1)*16) * * abs(i)>8 is equivalent (as i is an integer) to * abs(i)>=9 * i>=9 || i<=-9 * equivalent to (as i is the biggest integer smaller than log10(m2/m1)*16 * or the smallest integer bigger than log10(m2/m1)*16 depending on the sign of log10(m2/m1)*16) * log10(m2/m1)>=9/16 || log10(m2/m1)<=-9/16 * exp10 is strictly increasing thus this is equivalent to * m2/m1 >= 10^(9/16) || m2/m1<=10^(-9/16) which are comparisons to constants */ #define I1LIMIT 8 /* as in if(i>8) */ #define I2LIMIT 23 /* as in if(i>24) -> changed 23 */ #define MLIMIT 15 /* as in if(m<15) */ static FLOAT ma_max_i1; static FLOAT ma_max_i2; static FLOAT ma_max_m; /*This is the masking table: According to tonality, values are going from 0dB (TMN) to 9.3dB (NMT). After additive masking computation, 8dB are added, so final values are going from 8dB to 17.3dB */ static const FLOAT tab[] = { 1.0 /*pow(10, -0) */ , 0.79433 /*pow(10, -0.1) */ , 0.63096 /*pow(10, -0.2) */ , 0.63096 /*pow(10, -0.2) */ , 0.63096 /*pow(10, -0.2) */ , 0.63096 /*pow(10, -0.2) */ , 0.63096 /*pow(10, -0.2) */ , 0.25119 /*pow(10, -0.6) */ , 0.11749 /*pow(10, -0.93) */ }; static const int tab_mask_add_delta[] = { 2, 2, 2, 1, 1, 1, 0, 0, -1 }; #define STATIC_ASSERT_EQUAL_DIMENSION(A,B) {extern char static_assert_##A[dimension_of(A) == dimension_of(B) ? 1 : -1];(void) static_assert_##A;} inline static int mask_add_delta(int i) { STATIC_ASSERT_EQUAL_DIMENSION(tab_mask_add_delta,tab); assert(i < (int)dimension_of(tab)); return tab_mask_add_delta[i]; } static void init_mask_add_max_values(void) { ma_max_i1 = pow(10, (I1LIMIT + 1) / 16.0); ma_max_i2 = pow(10, (I2LIMIT + 1) / 16.0); ma_max_m = pow(10, (MLIMIT) / 10.0); } /* addition of simultaneous masking Naoki Shibata 2000/7 */ inline static FLOAT vbrpsy_mask_add(FLOAT m1, FLOAT m2, int b, int delta) { static const FLOAT table2[] = { 1.33352 * 1.33352, 1.35879 * 1.35879, 1.38454 * 1.38454, 1.39497 * 1.39497, 1.40548 * 1.40548, 1.3537 * 1.3537, 1.30382 * 1.30382, 1.22321 * 1.22321, 1.14758 * 1.14758, 1 }; FLOAT ratio; if (m1 < 0) { m1 = 0; } if (m2 < 0) { m2 = 0; } if (m1 <= 0) { return m2; } if (m2 <= 0) { return m1; } if (m2 > m1) { ratio = m2 / m1; } else { ratio = m1 / m2; } if (abs(b) <= delta) { /* approximately, 1 bark = 3 partitions */ /* originally 'if(i > 8)' */ if (ratio >= ma_max_i1) { return m1 + m2; } else { int i = (int) (FAST_LOG10_X(ratio, 16.0f)); return (m1 + m2) * table2[i]; } } if (ratio < ma_max_i2) { return m1 + m2; } if (m1 < m2) { m1 = m2; } return m1; } /* short block threshold calculation (part 2) partition band bo_s[sfb] is at the transition from scalefactor band sfb to the next one sfb+1; enn and thmm have to be split between them */ static void convert_partition2scalefac(PsyConst_CB2SB_t const *const gd, FLOAT const *eb, FLOAT const *thr, FLOAT enn_out[], FLOAT thm_out[]) { FLOAT enn, thmm; int sb, b, n = gd->n_sb; enn = thmm = 0.0f; for (sb = b = 0; sb < n; ++b, ++sb) { int const bo_sb = gd->bo[sb]; int const npart = gd->npart; int const b_lim = bo_sb < npart ? bo_sb : npart; while (b < b_lim) { assert(eb[b] >= 0); /* iff failed, it may indicate some index error elsewhere */ assert(thr[b] >= 0); enn += eb[b]; thmm += thr[b]; b++; } if (b >= npart) { enn_out[sb] = enn; thm_out[sb] = thmm; ++sb; break; } assert(eb[b] >= 0); /* iff failed, it may indicate some index error elsewhere */ assert(thr[b] >= 0); { /* at transition sfb -> sfb+1 */ FLOAT const w_curr = gd->bo_weight[sb]; FLOAT const w_next = 1.0f - w_curr; enn += w_curr * eb[b]; thmm += w_curr * thr[b]; enn_out[sb] = enn; thm_out[sb] = thmm; enn = w_next * eb[b]; thmm = w_next * thr[b]; } } /* zero initialize the rest */ for (; sb < n; ++sb) { enn_out[sb] = 0; thm_out[sb] = 0; } } static void convert_partition2scalefac_s(lame_internal_flags * gfc, FLOAT const *eb, FLOAT const *thr, int chn, int sblock) { PsyStateVar_t *const psv = &gfc->sv_psy; PsyConst_CB2SB_t const *const gds = &gfc->cd_psy->s; FLOAT enn[SBMAX_s], thm[SBMAX_s]; int sb; convert_partition2scalefac(gds, eb, thr, enn, thm); for (sb = 0; sb < SBMAX_s; ++sb) { psv->en[chn].s[sb][sblock] = enn[sb]; psv->thm[chn].s[sb][sblock] = thm[sb]; } } /* longblock threshold calculation (part 2) */ static void convert_partition2scalefac_l(lame_internal_flags * gfc, FLOAT const *eb, FLOAT const *thr, int chn) { PsyStateVar_t *const psv = &gfc->sv_psy; PsyConst_CB2SB_t const *const gdl = &gfc->cd_psy->l; FLOAT *enn = &psv->en[chn].l[0]; FLOAT *thm = &psv->thm[chn].l[0]; convert_partition2scalefac(gdl, eb, thr, enn, thm); } static void convert_partition2scalefac_l_to_s(lame_internal_flags * gfc, FLOAT const *eb, FLOAT const *thr, int chn) { PsyStateVar_t *const psv = &gfc->sv_psy; PsyConst_CB2SB_t const *const gds = &gfc->cd_psy->l_to_s; FLOAT enn[SBMAX_s], thm[SBMAX_s]; int sb, sblock; convert_partition2scalefac(gds, eb, thr, enn, thm); for (sb = 0; sb < SBMAX_s; ++sb) { FLOAT const scale = 1. / 64.f; FLOAT const tmp_enn = enn[sb]; FLOAT const tmp_thm = thm[sb] * scale; for (sblock = 0; sblock < 3; ++sblock) { psv->en[chn].s[sb][sblock] = tmp_enn; psv->thm[chn].s[sb][sblock] = tmp_thm; } } } static inline FLOAT NS_INTERP(FLOAT x, FLOAT y, FLOAT r) { /* was pow((x),(r))*pow((y),1-(r)) */ if (r >= 1.0f) return x; /* 99.7% of the time */ if (r <= 0.0f) return y; if (y > 0.0f) return powf(x / y, r) * y; /* rest of the time */ return 0.0f; /* never happens */ } static FLOAT pecalc_s(III_psy_ratio const *mr, FLOAT masking_lower) { FLOAT pe_s; static const FLOAT regcoef_s[] = { 11.8, /* these values are tuned only for 44.1kHz... */ 13.6, 17.2, 32, 46.5, 51.3, 57.5, 67.1, 71.5, 84.6, 97.6, 130, /* 255.8 */ }; unsigned int sb, sblock; pe_s = 1236.28f / 4; for (sb = 0; sb < SBMAX_s - 1; sb++) { for (sblock = 0; sblock < 3; sblock++) { FLOAT const thm = mr->thm.s[sb][sblock]; assert(sb < dimension_of(regcoef_s)); if (thm > 0.0f) { FLOAT const x = thm * masking_lower; FLOAT const en = mr->en.s[sb][sblock]; if (en > x) { if (en > x * 1e10f) { pe_s += regcoef_s[sb] * (10.0f * LOG10); } else { assert(x > 0); pe_s += regcoef_s[sb] * FAST_LOG10(en / x); } } } } } return pe_s; } static FLOAT pecalc_l(III_psy_ratio const *mr, FLOAT masking_lower) { FLOAT pe_l; static const FLOAT regcoef_l[] = { 6.8, /* these values are tuned only for 44.1kHz... */ 5.8, 5.8, 6.4, 6.5, 9.9, 12.1, 14.4, 15, 18.9, 21.6, 26.9, 34.2, 40.2, 46.8, 56.5, 60.7, 73.9, 85.7, 93.4, 126.1, /* 241.3 */ }; unsigned int sb; pe_l = 1124.23f / 4; for (sb = 0; sb < SBMAX_l - 1; sb++) { FLOAT const thm = mr->thm.l[sb]; assert(sb < dimension_of(regcoef_l)); if (thm > 0.0f) { FLOAT const x = thm * masking_lower; FLOAT const en = mr->en.l[sb]; if (en > x) { if (en > x * 1e10f) { pe_l += regcoef_l[sb] * (10.0f * LOG10); } else { assert(x > 0); pe_l += regcoef_l[sb] * FAST_LOG10(en / x); } } } } return pe_l; } static void calc_energy(PsyConst_CB2SB_t const *l, FLOAT const *fftenergy, FLOAT * eb, FLOAT * max, FLOAT * avg) { int b, j; for (b = j = 0; b < l->npart; ++b) { FLOAT ebb = 0, m = 0; int i; for (i = 0; i < l->numlines[b]; ++i, ++j) { FLOAT const el = fftenergy[j]; assert(el >= 0); ebb += el; if (m < el) m = el; } eb[b] = ebb; max[b] = m; avg[b] = ebb * l->rnumlines[b]; assert(l->rnumlines[b] >= 0); assert(ebb >= 0); assert(eb[b] >= 0); assert(max[b] >= 0); assert(avg[b] >= 0); } } static void calc_mask_index_l(lame_internal_flags const *gfc, FLOAT const *max, FLOAT const *avg, unsigned char *mask_idx) { PsyConst_CB2SB_t const *const gdl = &gfc->cd_psy->l; FLOAT m, a; int b, k; int const last_tab_entry = sizeof(tab) / sizeof(tab[0]) - 1; b = 0; a = avg[b] + avg[b + 1]; assert(a >= 0); if (a > 0.0f) { m = max[b]; if (m < max[b + 1]) m = max[b + 1]; assert((gdl->numlines[b] + gdl->numlines[b + 1] - 1) > 0); a = 20.0f * (m * 2.0f - a) / (a * (gdl->numlines[b] + gdl->numlines[b + 1] - 1)); k = (int) a; if (k > last_tab_entry) k = last_tab_entry; mask_idx[b] = k; } else { mask_idx[b] = 0; } for (b = 1; b < gdl->npart - 1; b++) { a = avg[b - 1] + avg[b] + avg[b + 1]; assert(a >= 0); if (a > 0.0f) { m = max[b - 1]; if (m < max[b]) m = max[b]; if (m < max[b + 1]) m = max[b + 1]; assert((gdl->numlines[b - 1] + gdl->numlines[b] + gdl->numlines[b + 1] - 1) > 0); a = 20.0f * (m * 3.0f - a) / (a * (gdl->numlines[b - 1] + gdl->numlines[b] + gdl->numlines[b + 1] - 1)); k = (int) a; if (k > last_tab_entry) k = last_tab_entry; mask_idx[b] = k; } else { mask_idx[b] = 0; } } assert(b > 0); assert(b == gdl->npart - 1); a = avg[b - 1] + avg[b]; assert(a >= 0); if (a > 0.0f) { m = max[b - 1]; if (m < max[b]) m = max[b]; assert((gdl->numlines[b - 1] + gdl->numlines[b] - 1) > 0); a = 20.0f * (m * 2.0f - a) / (a * (gdl->numlines[b - 1] + gdl->numlines[b] - 1)); k = (int) a; if (k > last_tab_entry) k = last_tab_entry; mask_idx[b] = k; } else { mask_idx[b] = 0; } assert(b == (gdl->npart - 1)); } static void vbrpsy_compute_fft_l(lame_internal_flags * gfc, const sample_t * const buffer[2], int chn, int gr_out, FLOAT fftenergy[HBLKSIZE], FLOAT(*wsamp_l)[BLKSIZE]) { SessionConfig_t const *const cfg = &gfc->cfg; PsyStateVar_t *psv = &gfc->sv_psy; plotting_data *plt = cfg->analysis ? gfc->pinfo : 0; int j; if (chn < 2) { fft_long(gfc, *wsamp_l, chn, buffer); } else if (chn == 2) { FLOAT const sqrt2_half = SQRT2 * 0.5f; /* FFT data for mid and side channel is derived from L & R */ for (j = BLKSIZE - 1; j >= 0; --j) { FLOAT const l = wsamp_l[0][j]; FLOAT const r = wsamp_l[1][j]; wsamp_l[0][j] = (l + r) * sqrt2_half; wsamp_l[1][j] = (l - r) * sqrt2_half; } } /********************************************************************* * compute energies *********************************************************************/ fftenergy[0] = wsamp_l[0][0]; fftenergy[0] *= fftenergy[0]; for (j = BLKSIZE / 2 - 1; j >= 0; --j) { FLOAT const re = (*wsamp_l)[BLKSIZE / 2 - j]; FLOAT const im = (*wsamp_l)[BLKSIZE / 2 + j]; fftenergy[BLKSIZE / 2 - j] = (re * re + im * im) * 0.5f; } /* total energy */ { FLOAT totalenergy = 0.0f; for (j = 11; j < HBLKSIZE; j++) totalenergy += fftenergy[j]; psv->tot_ener[chn] = totalenergy; } if (plt) { for (j = 0; j < HBLKSIZE; j++) { plt->energy[gr_out][chn][j] = plt->energy_save[chn][j]; plt->energy_save[chn][j] = fftenergy[j]; } } } static void vbrpsy_compute_fft_s(lame_internal_flags const *gfc, const sample_t * const buffer[2], int chn, int sblock, FLOAT(*fftenergy_s)[HBLKSIZE_s], FLOAT(*wsamp_s)[3][BLKSIZE_s]) { int j; if (sblock == 0 && chn < 2) { fft_short(gfc, *wsamp_s, chn, buffer); } if (chn == 2) { FLOAT const sqrt2_half = SQRT2 * 0.5f; /* FFT data for mid and side channel is derived from L & R */ for (j = BLKSIZE_s - 1; j >= 0; --j) { FLOAT const l = wsamp_s[0][sblock][j]; FLOAT const r = wsamp_s[1][sblock][j]; wsamp_s[0][sblock][j] = (l + r) * sqrt2_half; wsamp_s[1][sblock][j] = (l - r) * sqrt2_half; } } /********************************************************************* * compute energies *********************************************************************/ fftenergy_s[sblock][0] = (*wsamp_s)[sblock][0]; fftenergy_s[sblock][0] *= fftenergy_s[sblock][0]; for (j = BLKSIZE_s / 2 - 1; j >= 0; --j) { FLOAT const re = (*wsamp_s)[sblock][BLKSIZE_s / 2 - j]; FLOAT const im = (*wsamp_s)[sblock][BLKSIZE_s / 2 + j]; fftenergy_s[sblock][BLKSIZE_s / 2 - j] = (re * re + im * im) * 0.5f; } } /********************************************************************* * compute loudness approximation (used for ATH auto-level adjustment) *********************************************************************/ static void vbrpsy_compute_loudness_approximation_l(lame_internal_flags * gfc, int gr_out, int chn, const FLOAT fftenergy[HBLKSIZE]) { PsyStateVar_t *psv = &gfc->sv_psy; if (chn < 2) { /*no loudness for mid/side ch */ gfc->ov_psy.loudness_sq[gr_out][chn] = psv->loudness_sq_save[chn]; psv->loudness_sq_save[chn] = psycho_loudness_approx(fftenergy, gfc->ATH->eql_w); } } /********************************************************************** * Apply HPF of fs/4 to the input signal. * This is used for attack detection / handling. **********************************************************************/ static void vbrpsy_attack_detection(lame_internal_flags * gfc, const sample_t * const buffer[2], int gr_out, III_psy_ratio masking_ratio[2][2], III_psy_ratio masking_MS_ratio[2][2], FLOAT energy[4], FLOAT sub_short_factor[4][3], int ns_attacks[4][4], int uselongblock[2]) { FLOAT ns_hpfsmpl[2][576]; SessionConfig_t const *const cfg = &gfc->cfg; PsyStateVar_t *const psv = &gfc->sv_psy; plotting_data *plt = cfg->analysis ? gfc->pinfo : 0; int const n_chn_out = cfg->channels_out; /* chn=2 and 3 = Mid and Side channels */ int const n_chn_psy = (cfg->mode == JOINT_STEREO) ? 4 : n_chn_out; int chn, i, j; memset(&ns_hpfsmpl[0][0], 0, sizeof(ns_hpfsmpl)); /* Don't copy the input buffer into a temporary buffer */ /* unroll the loop 2 times */ for (chn = 0; chn < n_chn_out; chn++) { static const FLOAT fircoef[] = { -8.65163e-18 * 2, -0.00851586 * 2, -6.74764e-18 * 2, 0.0209036 * 2, -3.36639e-17 * 2, -0.0438162 * 2, -1.54175e-17 * 2, 0.0931738 * 2, -5.52212e-17 * 2, -0.313819 * 2 }; /* apply high pass filter of fs/4 */ const sample_t *const firbuf = &buffer[chn][576 - 350 - NSFIRLEN + 192]; assert(dimension_of(fircoef) == ((NSFIRLEN - 1) / 2)); for (i = 0; i < 576; i++) { FLOAT sum1, sum2; sum1 = firbuf[i + 10]; sum2 = 0.0; for (j = 0; j < ((NSFIRLEN - 1) / 2) - 1; j += 2) { sum1 += fircoef[j] * (firbuf[i + j] + firbuf[i + NSFIRLEN - j]); sum2 += fircoef[j + 1] * (firbuf[i + j + 1] + firbuf[i + NSFIRLEN - j - 1]); } ns_hpfsmpl[chn][i] = sum1 + sum2; } masking_ratio[gr_out][chn].en = psv->en[chn]; masking_ratio[gr_out][chn].thm = psv->thm[chn]; if (n_chn_psy > 2) { /* MS maskings */ /*percep_MS_entropy [chn-2] = gfc -> pe [chn]; */ masking_MS_ratio[gr_out][chn].en = psv->en[chn + 2]; masking_MS_ratio[gr_out][chn].thm = psv->thm[chn + 2]; } } for (chn = 0; chn < n_chn_psy; chn++) { FLOAT attack_intensity[12]; FLOAT en_subshort[12]; FLOAT en_short[4] = { 0, 0, 0, 0 }; FLOAT const *pf = ns_hpfsmpl[chn & 1]; int ns_uselongblock = 1; if (chn == 2) { for (i = 0, j = 576; j > 0; ++i, --j) { FLOAT const l = ns_hpfsmpl[0][i]; FLOAT const r = ns_hpfsmpl[1][i]; ns_hpfsmpl[0][i] = l + r; ns_hpfsmpl[1][i] = l - r; } } /*************************************************************** * determine the block type (window type) ***************************************************************/ /* calculate energies of each sub-shortblocks */ for (i = 0; i < 3; i++) { en_subshort[i] = psv->last_en_subshort[chn][i + 6]; assert(psv->last_en_subshort[chn][i + 4] > 0); attack_intensity[i] = en_subshort[i] / psv->last_en_subshort[chn][i + 4]; en_short[0] += en_subshort[i]; } for (i = 0; i < 9; i++) { FLOAT const *const pfe = pf + 576 / 9; FLOAT p = 1.; for (; pf < pfe; pf++) if (p < fabs(*pf)) p = fabs(*pf); psv->last_en_subshort[chn][i] = en_subshort[i + 3] = p; en_short[1 + i / 3] += p; if (p > en_subshort[i + 3 - 2]) { assert(en_subshort[i + 3 - 2] > 0); p = p / en_subshort[i + 3 - 2]; } else if (en_subshort[i + 3 - 2] > p * 10.0f) { assert(p > 0); p = en_subshort[i + 3 - 2] / (p * 10.0f); } else { p = 0.0; } attack_intensity[i + 3] = p; } /* pulse like signal detection for fatboy.wav and so on */ for (i = 0; i < 3; ++i) { FLOAT const enn = en_subshort[i * 3 + 3] + en_subshort[i * 3 + 4] + en_subshort[i * 3 + 5]; FLOAT factor = 1.f; if (en_subshort[i * 3 + 5] * 6 < enn) { factor *= 0.5f; if (en_subshort[i * 3 + 4] * 6 < enn) { factor *= 0.5f; } } sub_short_factor[chn][i] = factor; } if (plt) { FLOAT x = attack_intensity[0]; for (i = 1; i < 12; i++) { if (x < attack_intensity[i]) { x = attack_intensity[i]; } } plt->ers[gr_out][chn] = plt->ers_save[chn]; plt->ers_save[chn] = x; } /* compare energies between sub-shortblocks */ { FLOAT x = gfc->cd_psy->attack_threshold[chn]; for (i = 0; i < 12; i++) { if (ns_attacks[chn][i / 3] == 0) { if (attack_intensity[i] > x) { ns_attacks[chn][i / 3] = (i % 3) + 1; } } } } /* should have energy change between short blocks, in order to avoid periodic signals */ /* Good samples to show the effect are Trumpet test songs */ /* GB: tuned (1) to avoid too many short blocks for test sample TRUMPET */ /* RH: tuned (2) to let enough short blocks through for test sample FSOL and SNAPS */ for (i = 1; i < 4; i++) { FLOAT const u = en_short[i - 1]; FLOAT const v = en_short[i]; FLOAT const m = Max(u, v); if (m < 40000) { /* (2) */ if (u < 1.7f * v && v < 1.7f * u) { /* (1) */ if (i == 1 && ns_attacks[chn][0] <= ns_attacks[chn][i]) { ns_attacks[chn][0] = 0; } ns_attacks[chn][i] = 0; } } } if (ns_attacks[chn][0] <= psv->last_attacks[chn]) { ns_attacks[chn][0] = 0; } if (psv->last_attacks[chn] == 3 || ns_attacks[chn][0] + ns_attacks[chn][1] + ns_attacks[chn][2] + ns_attacks[chn][3]) { ns_uselongblock = 0; if (ns_attacks[chn][1] && ns_attacks[chn][0]) { ns_attacks[chn][1] = 0; } if (ns_attacks[chn][2] && ns_attacks[chn][1]) { ns_attacks[chn][2] = 0; } if (ns_attacks[chn][3] && ns_attacks[chn][2]) { ns_attacks[chn][3] = 0; } } if (chn < 2) { uselongblock[chn] = ns_uselongblock; } else { if (ns_uselongblock == 0) { uselongblock[0] = uselongblock[1] = 0; } } /* there is a one granule delay. Copy maskings computed last call * into masking_ratio to return to calling program. */ energy[chn] = psv->tot_ener[chn]; } } static void vbrpsy_skip_masking_s(lame_internal_flags * gfc, int chn, int sblock) { if (sblock == 0) { FLOAT *nbs2 = &gfc->sv_psy.nb_s2[chn][0]; FLOAT *nbs1 = &gfc->sv_psy.nb_s1[chn][0]; int const n = gfc->cd_psy->s.npart; int b; for (b = 0; b < n; b++) { nbs2[b] = nbs1[b]; } } } static void vbrpsy_calc_mask_index_s(lame_internal_flags const *gfc, FLOAT const *max, FLOAT const *avg, unsigned char *mask_idx) { PsyConst_CB2SB_t const *const gds = &gfc->cd_psy->s; FLOAT m, a; int b, k; int const last_tab_entry = dimension_of(tab) - 1; b = 0; a = avg[b] + avg[b + 1]; assert(a >= 0); if (a > 0.0f) { m = max[b]; if (m < max[b + 1]) m = max[b + 1]; assert((gds->numlines[b] + gds->numlines[b + 1] - 1) > 0); a = 20.0f * (m * 2.0f - a) / (a * (gds->numlines[b] + gds->numlines[b + 1] - 1)); k = (int) a; if (k > last_tab_entry) k = last_tab_entry; mask_idx[b] = k; } else { mask_idx[b] = 0; } for (b = 1; b < gds->npart - 1; b++) { a = avg[b - 1] + avg[b] + avg[b + 1]; assert(b + 1 < gds->npart); assert(a >= 0); if (a > 0.0) { m = max[b - 1]; if (m < max[b]) m = max[b]; if (m < max[b + 1]) m = max[b + 1]; assert((gds->numlines[b - 1] + gds->numlines[b] + gds->numlines[b + 1] - 1) > 0); a = 20.0f * (m * 3.0f - a) / (a * (gds->numlines[b - 1] + gds->numlines[b] + gds->numlines[b + 1] - 1)); k = (int) a; if (k > last_tab_entry) k = last_tab_entry; mask_idx[b] = k; } else { mask_idx[b] = 0; } } assert(b > 0); assert(b == gds->npart - 1); a = avg[b - 1] + avg[b]; assert(a >= 0); if (a > 0.0f) { m = max[b - 1]; if (m < max[b]) m = max[b]; assert((gds->numlines[b - 1] + gds->numlines[b] - 1) > 0); a = 20.0f * (m * 2.0f - a) / (a * (gds->numlines[b - 1] + gds->numlines[b] - 1)); k = (int) a; if (k > last_tab_entry) k = last_tab_entry; mask_idx[b] = k; } else { mask_idx[b] = 0; } assert(b == (gds->npart - 1)); } static void vbrpsy_compute_masking_s(lame_internal_flags * gfc, const FLOAT(*fftenergy_s)[HBLKSIZE_s], FLOAT * eb, FLOAT * thr, int chn, int sblock) { PsyStateVar_t *const psv = &gfc->sv_psy; PsyConst_CB2SB_t const *const gds = &gfc->cd_psy->s; FLOAT max[CBANDS], avg[CBANDS]; int i, j, b; unsigned char mask_idx_s[CBANDS]; memset(max, 0, sizeof(max)); memset(avg, 0, sizeof(avg)); for (b = j = 0; b < gds->npart; ++b) { FLOAT ebb = 0, m = 0; int const n = gds->numlines[b]; for (i = 0; i < n; ++i, ++j) { FLOAT const el = fftenergy_s[sblock][j]; ebb += el; if (m < el) m = el; } eb[b] = ebb; assert(ebb >= 0); max[b] = m; assert(n > 0); avg[b] = ebb * gds->rnumlines[b]; assert(avg[b] >= 0); } assert(b == gds->npart); assert(j == 129); vbrpsy_calc_mask_index_s(gfc, max, avg, mask_idx_s); for (j = b = 0; b < gds->npart; b++) { int kk = gds->s3ind[b][0]; int const last = gds->s3ind[b][1]; int const delta = mask_add_delta(mask_idx_s[b]); int dd, dd_n; FLOAT x, ecb, avg_mask; FLOAT const masking_lower = gds->masking_lower[b] * gfc->sv_qnt.masking_lower; dd = mask_idx_s[kk]; dd_n = 1; ecb = gds->s3[j] * eb[kk] * tab[mask_idx_s[kk]]; ++j, ++kk; while (kk <= last) { dd += mask_idx_s[kk]; dd_n += 1; x = gds->s3[j] * eb[kk] * tab[mask_idx_s[kk]]; ecb = vbrpsy_mask_add(ecb, x, kk - b, delta); ++j, ++kk; } dd = (1 + 2 * dd) / (2 * dd_n); avg_mask = tab[dd] * 0.5f; ecb *= avg_mask; #if 0 /* we can do PRE ECHO control now here, or do it later */ if (psv->blocktype_old[chn & 0x01] == SHORT_TYPE) { /* limit calculated threshold by even older granule */ FLOAT const t1 = rpelev_s * psv->nb_s1[chn][b]; FLOAT const t2 = rpelev2_s * psv->nb_s2[chn][b]; FLOAT const tm = (t2 > 0) ? Min(ecb, t2) : ecb; thr[b] = (t1 > 0) ? NS_INTERP(Min(tm, t1), ecb, 0.6) : ecb; } else { /* limit calculated threshold by older granule */ FLOAT const t1 = rpelev_s * psv->nb_s1[chn][b]; thr[b] = (t1 > 0) ? NS_INTERP(Min(ecb, t1), ecb, 0.6) : ecb; } #else /* we do it later */ thr[b] = ecb; #endif psv->nb_s2[chn][b] = psv->nb_s1[chn][b]; psv->nb_s1[chn][b] = ecb; { /* if THR exceeds EB, the quantization routines will take the difference * from other bands. in case of strong tonal samples (tonaltest.wav) * this leads to heavy distortions. that's why we limit THR here. */ x = max[b]; x *= gds->minval[b]; x *= avg_mask; if (thr[b] > x) { thr[b] = x; } } if (masking_lower > 1) { thr[b] *= masking_lower; } if (thr[b] > eb[b]) { thr[b] = eb[b]; } if (masking_lower < 1) { thr[b] *= masking_lower; } assert(thr[b] >= 0); } for (; b < CBANDS; ++b) { eb[b] = 0; thr[b] = 0; } } static void vbrpsy_compute_masking_l(lame_internal_flags * gfc, const FLOAT fftenergy[HBLKSIZE], FLOAT eb_l[CBANDS], FLOAT thr[CBANDS], int chn) { PsyStateVar_t *const psv = &gfc->sv_psy; PsyConst_CB2SB_t const *const gdl = &gfc->cd_psy->l; FLOAT max[CBANDS], avg[CBANDS]; unsigned char mask_idx_l[CBANDS + 2]; int k, b; /********************************************************************* * Calculate the energy and the tonality of each partition. *********************************************************************/ calc_energy(gdl, fftenergy, eb_l, max, avg); calc_mask_index_l(gfc, max, avg, mask_idx_l); /********************************************************************* * convolve the partitioned energy and unpredictability * with the spreading function, s3_l[b][k] ********************************************************************/ k = 0; for (b = 0; b < gdl->npart; b++) { FLOAT x, ecb, avg_mask, t; FLOAT const masking_lower = gdl->masking_lower[b] * gfc->sv_qnt.masking_lower; /* convolve the partitioned energy with the spreading function */ int kk = gdl->s3ind[b][0]; int const last = gdl->s3ind[b][1]; int const delta = mask_add_delta(mask_idx_l[b]); int dd = 0, dd_n = 0; dd = mask_idx_l[kk]; dd_n += 1; ecb = gdl->s3[k] * eb_l[kk] * tab[mask_idx_l[kk]]; ++k, ++kk; while (kk <= last) { dd += mask_idx_l[kk]; dd_n += 1; x = gdl->s3[k] * eb_l[kk] * tab[mask_idx_l[kk]]; t = vbrpsy_mask_add(ecb, x, kk - b, delta); #if 0 ecb += eb_l[kk]; if (ecb > t) { ecb = t; } #else ecb = t; #endif ++k, ++kk; } dd = (1 + 2 * dd) / (2 * dd_n); avg_mask = tab[dd] * 0.5f; ecb *= avg_mask; /**** long block pre-echo control ****/ /* dont use long block pre-echo control if previous granule was * a short block. This is to avoid the situation: * frame0: quiet (very low masking) * frame1: surge (triggers short blocks) * frame2: regular frame. looks like pre-echo when compared to * frame0, but all pre-echo was in frame1. */ /* chn=0,1 L and R channels chn=2,3 S and M channels. */ if (psv->blocktype_old[chn & 0x01] == SHORT_TYPE) { FLOAT const ecb_limit = rpelev * psv->nb_l1[chn][b]; if (ecb_limit > 0) { thr[b] = Min(ecb, ecb_limit); } else { /* Robert 071209: Because we don't calculate long block psy when we know a granule should be of short blocks, we don't have any clue how the granule before would have looked like as a long block. So we have to guess a little bit for this END_TYPE block. Most of the time we get away with this sloppyness. (fingers crossed :) The speed increase is worth it. */ thr[b] = Min(ecb, eb_l[b] * NS_PREECHO_ATT2); } } else { FLOAT ecb_limit_2 = rpelev2 * psv->nb_l2[chn][b]; FLOAT ecb_limit_1 = rpelev * psv->nb_l1[chn][b]; FLOAT ecb_limit; if (ecb_limit_2 <= 0) { ecb_limit_2 = ecb; } if (ecb_limit_1 <= 0) { ecb_limit_1 = ecb; } if (psv->blocktype_old[chn & 0x01] == NORM_TYPE) { ecb_limit = Min(ecb_limit_1, ecb_limit_2); } else { ecb_limit = ecb_limit_1; } thr[b] = Min(ecb, ecb_limit); } psv->nb_l2[chn][b] = psv->nb_l1[chn][b]; psv->nb_l1[chn][b] = ecb; { /* if THR exceeds EB, the quantization routines will take the difference * from other bands. in case of strong tonal samples (tonaltest.wav) * this leads to heavy distortions. that's why we limit THR here. */ x = max[b]; x *= gdl->minval[b]; x *= avg_mask; if (thr[b] > x) { thr[b] = x; } } if (masking_lower > 1) { thr[b] *= masking_lower; } if (thr[b] > eb_l[b]) { thr[b] = eb_l[b]; } if (masking_lower < 1) { thr[b] *= masking_lower; } assert(thr[b] >= 0); } for (; b < CBANDS; ++b) { eb_l[b] = 0; thr[b] = 0; } } static void vbrpsy_compute_block_type(SessionConfig_t const *cfg, int *uselongblock) { int chn; if (cfg->short_blocks == short_block_coupled /* force both channels to use the same block type */ /* this is necessary if the frame is to be encoded in ms_stereo. */ /* But even without ms_stereo, FhG does this */ && !(uselongblock[0] && uselongblock[1])) uselongblock[0] = uselongblock[1] = 0; for (chn = 0; chn < cfg->channels_out; chn++) { /* disable short blocks */ if (cfg->short_blocks == short_block_dispensed) { uselongblock[chn] = 1; } if (cfg->short_blocks == short_block_forced) { uselongblock[chn] = 0; } } } static void vbrpsy_apply_block_type(PsyStateVar_t * psv, int nch, int const *uselongblock, int *blocktype_d) { int chn; /* update the blocktype of the previous granule, since it depends on what * happend in this granule */ for (chn = 0; chn < nch; chn++) { int blocktype = NORM_TYPE; /* disable short blocks */ if (uselongblock[chn]) { /* no attack : use long blocks */ assert(psv->blocktype_old[chn] != START_TYPE); if (psv->blocktype_old[chn] == SHORT_TYPE) blocktype = STOP_TYPE; } else { /* attack : use short blocks */ blocktype = SHORT_TYPE; if (psv->blocktype_old[chn] == NORM_TYPE) { psv->blocktype_old[chn] = START_TYPE; } if (psv->blocktype_old[chn] == STOP_TYPE) psv->blocktype_old[chn] = SHORT_TYPE; } blocktype_d[chn] = psv->blocktype_old[chn]; /* value returned to calling program */ psv->blocktype_old[chn] = blocktype; /* save for next call to l3psy_anal */ } } /*************************************************************** * compute M/S thresholds from Johnston & Ferreira 1992 ICASSP paper ***************************************************************/ static void vbrpsy_compute_MS_thresholds(const FLOAT eb[4][CBANDS], FLOAT thr[4][CBANDS], const FLOAT cb_mld[CBANDS], const FLOAT ath_cb[CBANDS], FLOAT athlower, FLOAT msfix, int n) { FLOAT const msfix2 = msfix * 2.f; FLOAT rside, rmid; int b; for (b = 0; b < n; ++b) { FLOAT const ebM = eb[2][b]; FLOAT const ebS = eb[3][b]; FLOAT const thmL = thr[0][b]; FLOAT const thmR = thr[1][b]; FLOAT thmM = thr[2][b]; FLOAT thmS = thr[3][b]; /* use this fix if L & R masking differs by 2db or less */ /* if db = 10*log10(x2/x1) < 2 */ /* if (x2 < 1.58*x1) { */ if (thmL <= 1.58f * thmR && thmR <= 1.58f * thmL) { FLOAT const mld_m = cb_mld[b] * ebS; FLOAT const mld_s = cb_mld[b] * ebM; FLOAT const tmp_m = Min(thmS, mld_m); FLOAT const tmp_s = Min(thmM, mld_s); rmid = Max(thmM, tmp_m); rside = Max(thmS, tmp_s); } else { rmid = thmM; rside = thmS; } if (msfix > 0.f) { /***************************************************************/ /* Adjust M/S maskings if user set "msfix" */ /***************************************************************/ /* Naoki Shibata 2000 */ FLOAT thmLR, thmMS; FLOAT const ath = ath_cb[b] * athlower; FLOAT const tmp_l = Max(thmL, ath); FLOAT const tmp_r = Max(thmR, ath); thmLR = Min(tmp_l, tmp_r); thmM = Max(rmid, ath); thmS = Max(rside, ath); thmMS = thmM + thmS; if (thmMS > 0.f && (thmLR * msfix2) < thmMS) { FLOAT const f = thmLR * msfix2 / thmMS; thmM *= f; thmS *= f; assert(thmMS > 0.f); } rmid = Min(thmM, rmid); rside = Min(thmS, rside); } if (rmid > ebM) { rmid = ebM; } if (rside > ebS) { rside = ebS; } thr[2][b] = rmid; thr[3][b] = rside; } } /* * NOTE: the bitrate reduction from the inter-channel masking effect is low * compared to the chance of getting annyoing artefacts. L3psycho_anal_vbr does * not use this feature. (Robert 071216) */ int L3psycho_anal_vbr(lame_internal_flags * gfc, const sample_t * const buffer[2], int gr_out, III_psy_ratio masking_ratio[2][2], III_psy_ratio masking_MS_ratio[2][2], FLOAT percep_entropy[2], FLOAT percep_MS_entropy[2], FLOAT energy[4], int blocktype_d[2]) { SessionConfig_t const *const cfg = &gfc->cfg; PsyStateVar_t *const psv = &gfc->sv_psy; PsyConst_CB2SB_t const *const gdl = &gfc->cd_psy->l; PsyConst_CB2SB_t const *const gds = &gfc->cd_psy->s; plotting_data *plt = cfg->analysis ? gfc->pinfo : 0; III_psy_xmin last_thm[4]; /* fft and energy calculation */ FLOAT(*wsamp_l)[BLKSIZE]; FLOAT(*wsamp_s)[3][BLKSIZE_s]; FLOAT fftenergy[HBLKSIZE]; FLOAT fftenergy_s[3][HBLKSIZE_s]; FLOAT wsamp_L[2][BLKSIZE]; FLOAT wsamp_S[2][3][BLKSIZE_s]; FLOAT eb[4][CBANDS], thr[4][CBANDS]; FLOAT sub_short_factor[4][3]; FLOAT thmm; FLOAT const pcfact = 0.6f; FLOAT const ath_factor = (cfg->msfix > 0.f) ? (cfg->ATH_offset_factor * gfc->ATH->adjust_factor) : 1.f; const FLOAT(*const_eb)[CBANDS] = (const FLOAT(*)[CBANDS]) eb; const FLOAT(*const_fftenergy_s)[HBLKSIZE_s] = (const FLOAT(*)[HBLKSIZE_s]) fftenergy_s; /* block type */ int ns_attacks[4][4] = { {0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0} }; int uselongblock[2]; /* usual variables like loop indices, etc.. */ int chn, sb, sblock; /* chn=2 and 3 = Mid and Side channels */ int const n_chn_psy = (cfg->mode == JOINT_STEREO) ? 4 : cfg->channels_out; memcpy(&last_thm[0], &psv->thm[0], sizeof(last_thm)); vbrpsy_attack_detection(gfc, buffer, gr_out, masking_ratio, masking_MS_ratio, energy, sub_short_factor, ns_attacks, uselongblock); vbrpsy_compute_block_type(cfg, uselongblock); /* LONG BLOCK CASE */ { for (chn = 0; chn < n_chn_psy; chn++) { int const ch01 = chn & 0x01; wsamp_l = wsamp_L + ch01; vbrpsy_compute_fft_l(gfc, buffer, chn, gr_out, fftenergy, wsamp_l); vbrpsy_compute_loudness_approximation_l(gfc, gr_out, chn, fftenergy); vbrpsy_compute_masking_l(gfc, fftenergy, eb[chn], thr[chn], chn); } if (cfg->mode == JOINT_STEREO) { if ((uselongblock[0] + uselongblock[1]) == 2) { vbrpsy_compute_MS_thresholds(const_eb, thr, gdl->mld_cb, gfc->ATH->cb_l, ath_factor, cfg->msfix, gdl->npart); } } /* TODO: apply adaptive ATH masking here ?? */ for (chn = 0; chn < n_chn_psy; chn++) { convert_partition2scalefac_l(gfc, eb[chn], thr[chn], chn); convert_partition2scalefac_l_to_s(gfc, eb[chn], thr[chn], chn); } } /* SHORT BLOCKS CASE */ { int const force_short_block_calc = gfc->cd_psy->force_short_block_calc; for (sblock = 0; sblock < 3; sblock++) { for (chn = 0; chn < n_chn_psy; ++chn) { int const ch01 = chn & 0x01; if (uselongblock[ch01] && !force_short_block_calc) { vbrpsy_skip_masking_s(gfc, chn, sblock); } else { /* compute masking thresholds for short blocks */ wsamp_s = wsamp_S + ch01; vbrpsy_compute_fft_s(gfc, buffer, chn, sblock, fftenergy_s, wsamp_s); vbrpsy_compute_masking_s(gfc, const_fftenergy_s, eb[chn], thr[chn], chn, sblock); } } if (cfg->mode == JOINT_STEREO) { if ((uselongblock[0] + uselongblock[1]) == 0) { vbrpsy_compute_MS_thresholds(const_eb, thr, gds->mld_cb, gfc->ATH->cb_s, ath_factor, cfg->msfix, gds->npart); } } /* TODO: apply adaptive ATH masking here ?? */ for (chn = 0; chn < n_chn_psy; ++chn) { int const ch01 = chn & 0x01; if (!uselongblock[ch01] || force_short_block_calc) { convert_partition2scalefac_s(gfc, eb[chn], thr[chn], chn, sblock); } } } /**** short block pre-echo control ****/ for (chn = 0; chn < n_chn_psy; chn++) { for (sb = 0; sb < SBMAX_s; sb++) { FLOAT new_thmm[3], prev_thm, t1, t2; for (sblock = 0; sblock < 3; sblock++) { thmm = psv->thm[chn].s[sb][sblock]; thmm *= NS_PREECHO_ATT0; t1 = t2 = thmm; if (sblock > 0) { prev_thm = new_thmm[sblock - 1]; } else { prev_thm = last_thm[chn].s[sb][2]; } if (ns_attacks[chn][sblock] >= 2 || ns_attacks[chn][sblock + 1] == 1) { t1 = NS_INTERP(prev_thm, thmm, NS_PREECHO_ATT1 * pcfact); } thmm = Min(t1, thmm); if (ns_attacks[chn][sblock] == 1) { t2 = NS_INTERP(prev_thm, thmm, NS_PREECHO_ATT2 * pcfact); } else if ((sblock == 0 && psv->last_attacks[chn] == 3) || (sblock > 0 && ns_attacks[chn][sblock - 1] == 3)) { /* 2nd preceeding block */ switch (sblock) { case 0: prev_thm = last_thm[chn].s[sb][1]; break; case 1: prev_thm = last_thm[chn].s[sb][2]; break; case 2: prev_thm = new_thmm[0]; break; } t2 = NS_INTERP(prev_thm, thmm, NS_PREECHO_ATT2 * pcfact); } thmm = Min(t1, thmm); thmm = Min(t2, thmm); /* pulse like signal detection for fatboy.wav and so on */ thmm *= sub_short_factor[chn][sblock]; new_thmm[sblock] = thmm; } for (sblock = 0; sblock < 3; sblock++) { psv->thm[chn].s[sb][sblock] = new_thmm[sblock]; } } } } for (chn = 0; chn < n_chn_psy; chn++) { psv->last_attacks[chn] = ns_attacks[chn][2]; } /*************************************************************** * determine final block type ***************************************************************/ vbrpsy_apply_block_type(psv, cfg->channels_out, uselongblock, blocktype_d); /********************************************************************* * compute the value of PE to return ... no delay and advance *********************************************************************/ for (chn = 0; chn < n_chn_psy; chn++) { FLOAT *ppe; int type; III_psy_ratio const *mr; if (chn > 1) { ppe = percep_MS_entropy - 2; type = NORM_TYPE; if (blocktype_d[0] == SHORT_TYPE || blocktype_d[1] == SHORT_TYPE) type = SHORT_TYPE; mr = &masking_MS_ratio[gr_out][chn - 2]; } else { ppe = percep_entropy; type = blocktype_d[chn]; mr = &masking_ratio[gr_out][chn]; } if (type == SHORT_TYPE) { ppe[chn] = pecalc_s(mr, gfc->sv_qnt.masking_lower); } else { ppe[chn] = pecalc_l(mr, gfc->sv_qnt.masking_lower); } if (plt) { plt->pe[gr_out][chn] = ppe[chn]; } } return 0; } /* * The spreading function. Values returned in units of energy */ static FLOAT s3_func(FLOAT bark) { FLOAT tempx, x, tempy, temp; tempx = bark; if (tempx >= 0) tempx *= 3; else tempx *= 1.5; if (tempx >= 0.5 && tempx <= 2.5) { temp = tempx - 0.5; x = 8.0 * (temp * temp - 2.0 * temp); } else x = 0.0; tempx += 0.474; tempy = 15.811389 + 7.5 * tempx - 17.5 * sqrt(1.0 + tempx * tempx); if (tempy <= -60.0) return 0.0; tempx = exp((x + tempy) * LN_TO_LOG10); /* Normalization. The spreading function should be normalized so that: +inf / | s3 [ bark ] d(bark) = 1 / -inf */ tempx /= .6609193; return tempx; } #if 0 static FLOAT norm_s3_func(void) { double lim_a = 0, lim_b = 0; double x = 0, l, h; for (x = 0; s3_func(x) > 1e-20; x -= 1); l = x; h = 0; while (fabs(h - l) > 1e-12) { x = (h + l) / 2; if (s3_func(x) > 0) { h = x; } else { l = x; } } lim_a = l; for (x = 0; s3_func(x) > 1e-20; x += 1); l = 0; h = x; while (fabs(h - l) > 1e-12) { x = (h + l) / 2; if (s3_func(x) > 0) { l = x; } else { h = x; } } lim_b = h; { double sum = 0; int const m = 1000; int i; for (i = 0; i <= m; ++i) { double x = lim_a + i * (lim_b - lim_a) / m; double y = s3_func(x); sum += y; } { double norm = (m + 1) / (sum * (lim_b - lim_a)); /*printf( "norm = %lf\n",norm); */ return norm; } } } #endif static FLOAT stereo_demask(double f) { /* setup stereo demasking thresholds */ /* formula reverse enginerred from plot in paper */ double arg = freq2bark(f); arg = (Min(arg, 15.5) / 15.5); return pow(10.0, 1.25 * (1 - cos(PI * arg)) - 2.5); } static void init_numline(PsyConst_CB2SB_t * gd, FLOAT sfreq, int fft_size, int mdct_size, int sbmax, int const *scalepos) { FLOAT b_frq[CBANDS + 1]; FLOAT const mdct_freq_frac = sfreq / (2.0f * mdct_size); FLOAT const deltafreq = fft_size / (2.0f * mdct_size); int partition[HBLKSIZE] = { 0 }; int i, j, ni; int sfb; sfreq /= fft_size; j = 0; ni = 0; /* compute numlines, the number of spectral lines in each partition band */ /* each partition band should be about DELBARK wide. */ for (i = 0; i < CBANDS; i++) { FLOAT bark1; int j2, nl; bark1 = freq2bark(sfreq * j); b_frq[i] = sfreq * j; for (j2 = j; freq2bark(sfreq * j2) - bark1 < DELBARK && j2 <= fft_size / 2; j2++); nl = j2 - j; gd->numlines[i] = nl; gd->rnumlines[i] = (nl > 0) ? (1.0f / nl) : 0; ni = i + 1; while (j < j2) { assert(j < HBLKSIZE); partition[j++] = i; } if (j > fft_size / 2) { j = fft_size / 2; ++i; break; } } assert(i < CBANDS); b_frq[i] = sfreq * j; gd->n_sb = sbmax; gd->npart = ni; { j = 0; for (i = 0; i < gd->npart; i++) { int const nl = gd->numlines[i]; FLOAT const freq = sfreq * (j + nl / 2); gd->mld_cb[i] = stereo_demask(freq); j += nl; } for (; i < CBANDS; ++i) { gd->mld_cb[i] = 1; } } for (sfb = 0; sfb < sbmax; sfb++) { int i1, i2, bo; int start = scalepos[sfb]; int end = scalepos[sfb + 1]; i1 = floor(.5 + deltafreq * (start - .5)); if (i1 < 0) i1 = 0; i2 = floor(.5 + deltafreq * (end - .5)); if (i2 > fft_size / 2) i2 = fft_size / 2; bo = partition[i2]; gd->bm[sfb] = (partition[i1] + partition[i2]) / 2; gd->bo[sfb] = bo; /* calculate how much of this band belongs to current scalefactor band */ { FLOAT const f_tmp = mdct_freq_frac * end; FLOAT bo_w = (f_tmp - b_frq[bo]) / (b_frq[bo + 1] - b_frq[bo]); if (bo_w < 0) { bo_w = 0; } else { if (bo_w > 1) { bo_w = 1; } } gd->bo_weight[sfb] = bo_w; } gd->mld[sfb] = stereo_demask(mdct_freq_frac * start); } } static void compute_bark_values(PsyConst_CB2SB_t const *gd, FLOAT sfreq, int fft_size, FLOAT * bval, FLOAT * bval_width) { /* compute bark values of each critical band */ int k, j = 0, ni = gd->npart; sfreq /= fft_size; for (k = 0; k < ni; k++) { int const w = gd->numlines[k]; FLOAT bark1, bark2; bark1 = freq2bark(sfreq * (j)); bark2 = freq2bark(sfreq * (j + w - 1)); bval[k] = .5 * (bark1 + bark2); bark1 = freq2bark(sfreq * (j - .5)); bark2 = freq2bark(sfreq * (j + w - .5)); bval_width[k] = bark2 - bark1; j += w; } } static int init_s3_values(FLOAT ** p, int (*s3ind)[2], int npart, FLOAT const *bval, FLOAT const *bval_width, FLOAT const *norm) { FLOAT s3[CBANDS][CBANDS]; /* The s3 array is not linear in the bark scale. * bval[x] should be used to get the bark value. */ int i, j, k; int numberOfNoneZero = 0; memset(&s3[0][0], 0, sizeof(s3)); /* s[i][j], the value of the spreading function, * centered at band j (masker), for band i (maskee) * * i.e.: sum over j to spread into signal barkval=i * NOTE: i and j are used opposite as in the ISO docs */ for (i = 0; i < npart; i++) { for (j = 0; j < npart; j++) { FLOAT v = s3_func(bval[i] - bval[j]) * bval_width[j]; s3[i][j] = v * norm[i]; } } for (i = 0; i < npart; i++) { for (j = 0; j < npart; j++) { if (s3[i][j] > 0.0f) break; } s3ind[i][0] = j; for (j = npart - 1; j > 0; j--) { if (s3[i][j] > 0.0f) break; } s3ind[i][1] = j; numberOfNoneZero += (s3ind[i][1] - s3ind[i][0] + 1); } *p = malloc(sizeof(FLOAT) * numberOfNoneZero); if (!*p) return -1; k = 0; for (i = 0; i < npart; i++) for (j = s3ind[i][0]; j <= s3ind[i][1]; j++) (*p)[k++] = s3[i][j]; return 0; } int psymodel_init(lame_global_flags const *gfp) { lame_internal_flags *const gfc = gfp->internal_flags; SessionConfig_t *const cfg = &gfc->cfg; PsyStateVar_t *const psv = &gfc->sv_psy; PsyConst_t *gd; int i, j, b, sb, k; FLOAT bvl_a = 13, bvl_b = 24; FLOAT snr_l_a = 0, snr_l_b = 0; FLOAT snr_s_a = -8.25, snr_s_b = -4.5; FLOAT bval[CBANDS]; FLOAT bval_width[CBANDS]; FLOAT norm[CBANDS]; FLOAT const sfreq = cfg->samplerate_out; FLOAT xav = 10, xbv = 12; FLOAT const minval_low = (0.f - cfg->minval); if (gfc->cd_psy != 0) { return 0; } memset(norm, 0, sizeof(norm)); gd = calloc(1, sizeof(PsyConst_t)); gfc->cd_psy = gd; gd->force_short_block_calc = gfp->experimentalZ; psv->blocktype_old[0] = psv->blocktype_old[1] = NORM_TYPE; /* the vbr header is long blocks */ for (i = 0; i < 4; ++i) { for (j = 0; j < CBANDS; ++j) { psv->nb_l1[i][j] = 1e20; psv->nb_l2[i][j] = 1e20; psv->nb_s1[i][j] = psv->nb_s2[i][j] = 1.0; } for (sb = 0; sb < SBMAX_l; sb++) { psv->en[i].l[sb] = 1e20; psv->thm[i].l[sb] = 1e20; } for (j = 0; j < 3; ++j) { for (sb = 0; sb < SBMAX_s; sb++) { psv->en[i].s[sb][j] = 1e20; psv->thm[i].s[sb][j] = 1e20; } psv->last_attacks[i] = 0; } for (j = 0; j < 9; j++) psv->last_en_subshort[i][j] = 10.; } /* init. for loudness approx. -jd 2001 mar 27 */ psv->loudness_sq_save[0] = psv->loudness_sq_save[1] = 0.0; /************************************************************************* * now compute the psychoacoustic model specific constants ************************************************************************/ /* compute numlines, bo, bm, bval, bval_width, mld */ init_numline(&gd->l, sfreq, BLKSIZE, 576, SBMAX_l, gfc->scalefac_band.l); assert(gd->l.npart < CBANDS); compute_bark_values(&gd->l, sfreq, BLKSIZE, bval, bval_width); /* compute the spreading function */ for (i = 0; i < gd->l.npart; i++) { double snr = snr_l_a; if (bval[i] >= bvl_a) { snr = snr_l_b * (bval[i] - bvl_a) / (bvl_b - bvl_a) + snr_l_a * (bvl_b - bval[i]) / (bvl_b - bvl_a); } norm[i] = pow(10.0, snr / 10.0); } i = init_s3_values(&gd->l.s3, gd->l.s3ind, gd->l.npart, bval, bval_width, norm); if (i) return i; /* compute long block specific values, ATH and MINVAL */ j = 0; for (i = 0; i < gd->l.npart; i++) { double x; /* ATH */ x = FLOAT_MAX; for (k = 0; k < gd->l.numlines[i]; k++, j++) { FLOAT const freq = sfreq * j / (1000.0 * BLKSIZE); FLOAT level; /* freq = Min(.1,freq); *//* ATH below 100 Hz constant, not further climbing */ level = ATHformula(cfg, freq * 1000) - 20; /* scale to FFT units; returned value is in dB */ level = pow(10., 0.1 * level); /* convert from dB -> energy */ level *= gd->l.numlines[i]; if (x > level) x = level; } gfc->ATH->cb_l[i] = x; /* MINVAL. For low freq, the strength of the masking is limited by minval this is an ISO MPEG1 thing, dont know if it is really needed */ /* FIXME: it does work to reduce low-freq problems in S53-Wind-Sax and lead-voice samples, but introduces some 3 kbps bit bloat too. TODO: Further refinement of the shape of this hack. */ x = 20.0 * (bval[i] / xav - 1.0); if (x > 6) { x = 30; } if (x < minval_low) { x = minval_low; } if (cfg->samplerate_out < 44000) { x = 30; } x -= 8.; gd->l.minval[i] = pow(10.0, x / 10.) * gd->l.numlines[i]; } /************************************************************************ * do the same things for short blocks ************************************************************************/ init_numline(&gd->s, sfreq, BLKSIZE_s, 192, SBMAX_s, gfc->scalefac_band.s); assert(gd->s.npart < CBANDS); compute_bark_values(&gd->s, sfreq, BLKSIZE_s, bval, bval_width); /* SNR formula. short block is normalized by SNR. is it still right ? */ j = 0; for (i = 0; i < gd->s.npart; i++) { double x; double snr = snr_s_a; if (bval[i] >= bvl_a) { snr = snr_s_b * (bval[i] - bvl_a) / (bvl_b - bvl_a) + snr_s_a * (bvl_b - bval[i]) / (bvl_b - bvl_a); } norm[i] = pow(10.0, snr / 10.0); /* ATH */ x = FLOAT_MAX; for (k = 0; k < gd->s.numlines[i]; k++, j++) { FLOAT const freq = sfreq * j / (1000.0 * BLKSIZE_s); FLOAT level; /* freq = Min(.1,freq); *//* ATH below 100 Hz constant, not further climbing */ level = ATHformula(cfg, freq * 1000) - 20; /* scale to FFT units; returned value is in dB */ level = pow(10., 0.1 * level); /* convert from dB -> energy */ level *= gd->s.numlines[i]; if (x > level) x = level; } gfc->ATH->cb_s[i] = x; /* MINVAL. For low freq, the strength of the masking is limited by minval this is an ISO MPEG1 thing, dont know if it is really needed */ x = 7.0 * (bval[i] / xbv - 1.0); if (bval[i] > xbv) { x *= 1 + log(1 + x) * 3.1; } if (bval[i] < xbv) { x *= 1 + log(1 - x) * 2.3; } if (x > 6) { x = 30; } if (x < minval_low) { x = minval_low; } if (cfg->samplerate_out < 44000) { x = 30; } x -= 8; gd->s.minval[i] = pow(10.0, x / 10) * gd->s.numlines[i]; } i = init_s3_values(&gd->s.s3, gd->s.s3ind, gd->s.npart, bval, bval_width, norm); if (i) return i; init_mask_add_max_values(); init_fft(gfc); /* setup temporal masking */ gd->decay = exp(-1.0 * LOG10 / (temporalmask_sustain_sec * sfreq / 192.0)); { FLOAT msfix; msfix = NS_MSFIX; if (cfg->use_safe_joint_stereo) msfix = 1.0; if (fabs(cfg->msfix) > 0.0) msfix = cfg->msfix; cfg->msfix = msfix; /* spread only from npart_l bands. Normally, we use the spreading * function to convolve from npart_l down to npart_l bands */ for (b = 0; b < gd->l.npart; b++) if (gd->l.s3ind[b][1] > gd->l.npart - 1) gd->l.s3ind[b][1] = gd->l.npart - 1; } /* prepare for ATH auto adjustment: * we want to decrease the ATH by 12 dB per second */ #define frame_duration (576. * cfg->mode_gr / sfreq) gfc->ATH->decay = pow(10., -12. / 10. * frame_duration); gfc->ATH->adjust_factor = 0.01; /* minimum, for leading low loudness */ gfc->ATH->adjust_limit = 1.0; /* on lead, allow adjust up to maximum */ #undef frame_duration assert(gd->l.bo[SBMAX_l - 1] <= gd->l.npart); assert(gd->s.bo[SBMAX_s - 1] <= gd->s.npart); if (cfg->ATHtype != -1) { /* compute equal loudness weights (eql_w) */ FLOAT freq; FLOAT const freq_inc = (FLOAT) cfg->samplerate_out / (FLOAT) (BLKSIZE); FLOAT eql_balance = 0.0; freq = 0.0; for (i = 0; i < BLKSIZE / 2; ++i) { /* convert ATH dB to relative power (not dB) */ /* to determine eql_w */ freq += freq_inc; gfc->ATH->eql_w[i] = 1. / pow(10, ATHformula(cfg, freq) / 10); eql_balance += gfc->ATH->eql_w[i]; } eql_balance = 1.0 / eql_balance; for (i = BLKSIZE / 2; --i >= 0;) { /* scale weights */ gfc->ATH->eql_w[i] *= eql_balance; } } { for (b = j = 0; b < gd->s.npart; ++b) { for (i = 0; i < gd->s.numlines[b]; ++i) { ++j; } } assert(j == 129); for (b = j = 0; b < gd->l.npart; ++b) { for (i = 0; i < gd->l.numlines[b]; ++i) { ++j; } } assert(j == 513); } /* short block attack threshold */ { float x = gfp->attackthre; float y = gfp->attackthre_s; if (x < 0) { x = NSATTACKTHRE; } if (y < 0) { y = NSATTACKTHRE_S; } gd->attack_threshold[0] = gd->attack_threshold[1] = gd->attack_threshold[2] = x; gd->attack_threshold[3] = y; } { float sk_s = -10.f, sk_l = -4.7f; static float const sk[] = { -7.4, -7.4, -7.4, -9.5, -7.4, -6.1, -5.5, -4.7, -4.7, -4.7, -4.7 }; if (gfp->VBR_q < 4) { sk_l = sk_s = sk[0]; } else { sk_l = sk_s = sk[gfp->VBR_q] + gfp->VBR_q_frac * (sk[gfp->VBR_q] - sk[gfp->VBR_q + 1]); } b = 0; for (; b < gd->s.npart; b++) { float m = (float) (gd->s.npart - b) / gd->s.npart; gd->s.masking_lower[b] = powf(10.f, sk_s * m * 0.1f); } for (; b < CBANDS; ++b) { gd->s.masking_lower[b] = 1.f; } b = 0; for (; b < gd->l.npart; b++) { float m = (float) (gd->l.npart - b) / gd->l.npart; gd->l.masking_lower[b] = powf(10.f, sk_l * m * 0.1f); } for (; b < CBANDS; ++b) { gd->l.masking_lower[b] = 1.f; } } memcpy(&gd->l_to_s, &gd->l, sizeof(gd->l_to_s)); init_numline(&gd->l_to_s, sfreq, BLKSIZE, 192, SBMAX_s, gfc->scalefac_band.s); return 0; }