* AAC coefficients encoder
* Copyright (C) 2008-2009 Konstantin Shishkov
*
- * This file is part of FFmpeg.
+ * This file is part of Libav.
*
- * FFmpeg is free software; you can redistribute it and/or
+ * Libav 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.
*
- * FFmpeg is distributed in the hope that it will be useful,
+ * Libav 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 FFmpeg; if not, write to the Free Software
+ * License along with Libav; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
* add sane pulse detection
***********************************/
+#include "libavutil/libm.h" // brought forward to work around cygwin header breakage
+
#include <float.h>
+#include "libavutil/mathematics.h"
#include "avcodec.h"
#include "put_bits.h"
#include "aac.h"
int *bits, int BT_ZERO, int BT_UNSIGNED,
int BT_PAIR, int BT_ESC)
{
- const float IQ = ff_aac_pow2sf_tab[200 + scale_idx - SCALE_ONE_POS + SCALE_DIV_512];
- const float Q = ff_aac_pow2sf_tab[200 - scale_idx + SCALE_ONE_POS - SCALE_DIV_512];
+ const int q_idx = POW_SF2_ZERO - scale_idx + SCALE_ONE_POS - SCALE_DIV_512;
+ const float Q = ff_aac_pow2sf_tab [q_idx];
+ const float Q34 = ff_aac_pow34sf_tab[q_idx];
+ const float IQ = ff_aac_pow2sf_tab [POW_SF2_ZERO + scale_idx - SCALE_ONE_POS + SCALE_DIV_512];
const float CLIPPED_ESCAPE = 165140.0f*IQ;
- int i, j, k;
+ int i, j;
float cost = 0;
const int dim = BT_PAIR ? 2 : 4;
int resbits = 0;
- const float Q34 = sqrtf(Q * sqrtf(Q));
const int range = aac_cb_range[cb];
const int maxval = aac_cb_maxval[cb];
int off;
curidx *= range;
curidx += quants[j] + off;
}
- curbits = ff_aac_spectral_bits[cb-1][curidx];
- vec = &ff_aac_codebook_vectors[cb-1][curidx*dim];
- if (BT_UNSIGNED) {
- for (k = 0; k < dim; k++) {
- float t = fabsf(in[i+k]);
- float di;
- if (BT_ESC && vec[k] == 64.0f) { //FIXME: slow
- if (t >= CLIPPED_ESCAPE) {
- di = t - CLIPPED_ESCAPE;
- curbits += 21;
- } else {
- int c = av_clip(quant(t, Q), 0, 8191);
- di = t - c*cbrtf(c)*IQ;
- curbits += av_log2(c)*2 - 4 + 1;
- }
+ curbits = ff_aac_spectral_bits[cb-1][curidx];
+ vec = &ff_aac_codebook_vectors[cb-1][curidx*dim];
+ if (BT_UNSIGNED) {
+ for (j = 0; j < dim; j++) {
+ float t = fabsf(in[i+j]);
+ float di;
+ if (BT_ESC && vec[j] == 64.0f) { //FIXME: slow
+ if (t >= CLIPPED_ESCAPE) {
+ di = t - CLIPPED_ESCAPE;
+ curbits += 21;
} else {
- di = t - vec[k]*IQ;
+ int c = av_clip(quant(t, Q), 0, 8191);
+ di = t - c*cbrtf(c)*IQ;
+ curbits += av_log2(c)*2 - 4 + 1;
}
- if (vec[k] != 0.0f)
- curbits++;
- rd += di*di;
- }
- } else {
- for (k = 0; k < dim; k++) {
- float di = in[i+k] - vec[k]*IQ;
- rd += di*di;
+ } else {
+ di = t - vec[j]*IQ;
}
+ if (vec[j] != 0.0f)
+ curbits++;
+ rd += di*di;
}
+ } else {
+ for (j = 0; j < dim; j++) {
+ float di = in[i+j] - vec[j]*IQ;
+ rd += di*di;
+ }
+ }
cost += rd * lambda + curbits;
resbits += curbits;
if (cost >= uplim)
}
static int find_min_book(float maxval, int sf) {
- float Q = ff_aac_pow2sf_tab[200 - sf + SCALE_ONE_POS - SCALE_DIV_512];
+ float Q = ff_aac_pow2sf_tab[POW_SF2_ZERO - sf + SCALE_ONE_POS - SCALE_DIV_512];
float Q34 = sqrtf(Q * sqrtf(Q));
int qmaxval, cb;
qmaxval = maxval * Q34 + 0.4054f;
int win, int group_len, const float lambda)
{
BandCodingPath path[120][12];
- int w, swb, cb, start, start2, size;
+ int w, swb, cb, start, size;
int i, j;
const int max_sfb = sce->ics.max_sfb;
const int run_bits = sce->ics.num_windows == 1 ? 5 : 3;
path[0][cb].run = 0;
}
for (swb = 0; swb < max_sfb; swb++) {
- start2 = start;
size = sce->ics.swb_sizes[swb];
if (sce->zeroes[win*16 + swb]) {
for (cb = 0; cb < 12; cb++) {
float cost_stay_here, cost_get_here;
float rd = 0.0f;
for (w = 0; w < group_len; w++) {
- FFPsyBand *band = &s->psy.psy_bands[s->cur_channel*PSY_MAX_BANDS+(win+w)*16+swb];
+ FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[(win+w)*16+swb];
rd += quantize_band_cost(s, sce->coeffs + start + w*128,
s->scoefs + start + w*128, size,
sce->sf_idx[(win+w)*16+swb], cb,
int win, int group_len, const float lambda)
{
BandCodingPath path[120][12];
- int w, swb, cb, start, start2, size;
+ int w, swb, cb, start, size;
int i, j;
const int max_sfb = sce->ics.max_sfb;
const int run_bits = sce->ics.num_windows == 1 ? 5 : 3;
const int run_esc = (1 << run_bits) - 1;
int idx, ppos, count;
int stackrun[120], stackcb[120], stack_len;
- float next_minrd = INFINITY;
+ float next_minbits = INFINITY;
int next_mincb = 0;
abs_pow34_v(s->scoefs, sce->coeffs, 1024);
path[0][cb].run = 0;
}
for (swb = 0; swb < max_sfb; swb++) {
- start2 = start;
size = sce->ics.swb_sizes[swb];
if (sce->zeroes[win*16 + swb]) {
- for (cb = 0; cb < 12; cb++) {
- path[swb+1][cb].prev_idx = cb;
- path[swb+1][cb].cost = path[swb][cb].cost;
- path[swb+1][cb].run = path[swb][cb].run + 1;
+ float cost_stay_here = path[swb][0].cost;
+ float cost_get_here = next_minbits + run_bits + 4;
+ if ( run_value_bits[sce->ics.num_windows == 8][path[swb][0].run]
+ != run_value_bits[sce->ics.num_windows == 8][path[swb][0].run+1])
+ cost_stay_here += run_bits;
+ if (cost_get_here < cost_stay_here) {
+ path[swb+1][0].prev_idx = next_mincb;
+ path[swb+1][0].cost = cost_get_here;
+ path[swb+1][0].run = 1;
+ } else {
+ path[swb+1][0].prev_idx = 0;
+ path[swb+1][0].cost = cost_stay_here;
+ path[swb+1][0].run = path[swb][0].run + 1;
+ }
+ next_minbits = path[swb+1][0].cost;
+ next_mincb = 0;
+ for (cb = 1; cb < 12; cb++) {
+ path[swb+1][cb].cost = 61450;
+ path[swb+1][cb].prev_idx = -1;
+ path[swb+1][cb].run = 0;
}
} else {
- float minrd = next_minrd;
+ float minbits = next_minbits;
int mincb = next_mincb;
int startcb = sce->band_type[win*16+swb];
- next_minrd = INFINITY;
+ next_minbits = INFINITY;
next_mincb = 0;
for (cb = 0; cb < startcb; cb++) {
path[swb+1][cb].cost = 61450;
}
for (cb = startcb; cb < 12; cb++) {
float cost_stay_here, cost_get_here;
- float rd = 0.0f;
+ float bits = 0.0f;
for (w = 0; w < group_len; w++) {
- rd += quantize_band_cost(s, sce->coeffs + start + w*128,
- s->scoefs + start + w*128, size,
- sce->sf_idx[(win+w)*16+swb], cb,
- 0, INFINITY, NULL);
+ bits += quantize_band_cost(s, sce->coeffs + start + w*128,
+ s->scoefs + start + w*128, size,
+ sce->sf_idx[(win+w)*16+swb], cb,
+ 0, INFINITY, NULL);
}
- cost_stay_here = path[swb][cb].cost + rd;
- cost_get_here = minrd + rd + run_bits + 4;
+ cost_stay_here = path[swb][cb].cost + bits;
+ cost_get_here = minbits + bits + run_bits + 4;
if ( run_value_bits[sce->ics.num_windows == 8][path[swb][cb].run]
!= run_value_bits[sce->ics.num_windows == 8][path[swb][cb].run+1])
cost_stay_here += run_bits;
path[swb+1][cb].cost = cost_stay_here;
path[swb+1][cb].run = path[swb][cb].run + 1;
}
- if (path[swb+1][cb].cost < next_minrd) {
- next_minrd = path[swb+1][cb].cost;
+ if (path[swb+1][cb].cost < next_minbits) {
+ next_minbits = path[swb+1][cb].cost;
next_mincb = cb;
}
}
/** Return the minimum scalefactor where the quantized coef does not clip. */
static av_always_inline uint8_t coef2minsf(float coef) {
- return av_clip_uint8(log2(coef)*4 - 69 + SCALE_ONE_POS - SCALE_DIV_512);
+ return av_clip_uint8(log2f(coef)*4 - 69 + SCALE_ONE_POS - SCALE_DIV_512);
}
/** Return the maximum scalefactor where the quantized coef is not zero. */
static av_always_inline uint8_t coef2maxsf(float coef) {
- return av_clip_uint8(log2(coef)*4 + 6 + SCALE_ONE_POS - SCALE_DIV_512);
+ return av_clip_uint8(log2f(coef)*4 + 6 + SCALE_ONE_POS - SCALE_DIV_512);
}
typedef struct TrellisPath {
int q0low = q0;
int q1high = q1;
//minimum scalefactor index is when maximum nonzero coefficient after quantizing is not clipped
- int qnrg = av_clip_uint8(log2(sqrt(qnrgf/qcnt))*4 - 31 + SCALE_ONE_POS - SCALE_DIV_512);
+ int qnrg = av_clip_uint8(log2f(sqrtf(qnrgf/qcnt))*4 - 31 + SCALE_ONE_POS - SCALE_DIV_512);
q1 = qnrg + 30;
q0 = qnrg - 30;
- //av_log(NULL, AV_LOG_ERROR, "q0 %d, q1 %d\n", q0, q1);
+ //av_log(NULL, AV_LOG_ERROR, "q0 %d, q1 %d\n", q0, q1);
if (q0 < q0low) {
q1 += q0low - q0;
q0 = q0low;
qmin = INT_MAX;
qmax = 0.0f;
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
- FFPsyBand *band = &s->psy.psy_bands[s->cur_channel*PSY_MAX_BANDS+(w+w2)*16+g];
+ FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g];
if (band->energy <= band->threshold || band->threshold == 0.0f) {
sce->zeroes[(w+w2)*16+g] = 1;
continue;
float dist = 0;
int cb = find_min_book(maxval, sce->sf_idx[w*16+g]);
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
- FFPsyBand *band = &s->psy.psy_bands[s->cur_channel*PSY_MAX_BANDS+(w+w2)*16+g];
+ FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g];
dist += quantize_band_cost(s, coefs + w2*128, s->scoefs + start + w2*128, sce->ics.swb_sizes[g],
q + q0, cb, lambda / band->threshold, INFINITY, NULL);
}
int start = 0, i, w, w2, g;
int destbits = avctx->bit_rate * 1024.0 / avctx->sample_rate / avctx->channels;
float dists[128], uplims[128];
+ float maxvals[128];
int fflag, minscaler;
int its = 0;
int allz = 0;
int nz = 0;
float uplim = 0.0f;
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
- FFPsyBand *band = &s->psy.psy_bands[s->cur_channel*PSY_MAX_BANDS+(w+w2)*16+g];
+ FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g];
uplim += band->threshold;
if (band->energy <= band->threshold || band->threshold == 0.0f) {
sce->zeroes[(w+w2)*16+g] = 1;
sce->zeroes[w*16+g] = !nz;
if (nz)
minthr = FFMIN(minthr, uplim);
- allz = FFMAX(allz, nz);
+ allz |= nz;
}
}
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
sce->sf_idx[w*16+g] = SCALE_ONE_POS;
continue;
}
- sce->sf_idx[w*16+g] = SCALE_ONE_POS + FFMIN(log2(uplims[w*16+g]/minthr)*4,59);
+ sce->sf_idx[w*16+g] = SCALE_ONE_POS + FFMIN(log2f(uplims[w*16+g]/minthr)*4,59);
}
}
if (!allz)
return;
abs_pow34_v(s->scoefs, sce->coeffs, 1024);
+
+ for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
+ start = w*128;
+ for (g = 0; g < sce->ics.num_swb; g++) {
+ const float *scaled = s->scoefs + start;
+ maxvals[w*16+g] = find_max_val(sce->ics.group_len[w], sce->ics.swb_sizes[g], scaled);
+ start += sce->ics.swb_sizes[g];
+ }
+ }
+
//perform two-loop search
//outer loop - improve quality
do {
continue;
}
minscaler = FFMIN(minscaler, sce->sf_idx[w*16+g]);
- cb = find_min_book(find_max_val(sce->ics.group_len[w], sce->ics.swb_sizes[g], scaled), sce->sf_idx[w*16+g]);
+ cb = find_min_book(maxvals[w*16+g], sce->sf_idx[w*16+g]);
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
int b;
dist += quantize_band_cost(s, coefs + w2*128,
sce->sf_idx[i] -= qstep;
}
qstep >>= 1;
- if (!qstep && tbits > destbits*1.02)
+ if (!qstep && tbits > destbits*1.02 && sce->sf_idx[0] < 217)
qstep = 1;
- if (sce->sf_idx[0] >= 217)
- break;
} while (qstep);
fflag = 0;
minscaler = av_clip(minscaler, 60, 255 - SCALE_MAX_DIFF);
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
- start = w*128;
for (g = 0; g < sce->ics.num_swb; g++) {
int prevsc = sce->sf_idx[w*16+g];
- const float *scaled = s->scoefs + start;
- if (dists[w*16+g] > uplims[w*16+g] && sce->sf_idx[w*16+g] > 60)
- sce->sf_idx[w*16+g]--;
+ if (dists[w*16+g] > uplims[w*16+g] && sce->sf_idx[w*16+g] > 60) {
+ if (find_min_book(maxvals[w*16+g], sce->sf_idx[w*16+g]-1))
+ sce->sf_idx[w*16+g]--;
+ else //Try to make sure there is some energy in every band
+ sce->sf_idx[w*16+g]-=2;
+ }
sce->sf_idx[w*16+g] = av_clip(sce->sf_idx[w*16+g], minscaler, minscaler + SCALE_MAX_DIFF);
sce->sf_idx[w*16+g] = FFMIN(sce->sf_idx[w*16+g], 219);
if (sce->sf_idx[w*16+g] != prevsc)
fflag = 1;
- sce->band_type[w*16+g] = find_min_book(find_max_val(sce->ics.group_len[w], sce->ics.swb_sizes[g], scaled), sce->sf_idx[w*16+g]);
- start += sce->ics.swb_sizes[g];
+ sce->band_type[w*16+g] = find_min_book(maxvals[w*16+g], sce->sf_idx[w*16+g]);
}
}
its++;
continue;
}
sce->zeroes[w*16+g] = 0;
- scf = prev_scf = av_clip(SCALE_ONE_POS - SCALE_DIV_512 - log2(1/maxq[w*16+g])*16/3, 60, 218);
+ scf = prev_scf = av_clip(SCALE_ONE_POS - SCALE_DIV_512 - log2f(1/maxq[w*16+g])*16/3, 60, 218);
step = 16;
for (;;) {
float dist = 0.0f;
dist -= b;
}
dist *= 1.0f / 512.0f / lambda;
- quant_max = quant(maxq[w*16+g], ff_aac_pow2sf_tab[200 - scf + SCALE_ONE_POS - SCALE_DIV_512]);
+ quant_max = quant(maxq[w*16+g], ff_aac_pow2sf_tab[POW_SF2_ZERO - scf + SCALE_ONE_POS - SCALE_DIV_512]);
if (quant_max >= 8191) { // too much, return to the previous quantizer
sce->sf_idx[w*16+g] = prev_scf;
break;
if (curdiff <= 1.0f)
step = 0;
else
- step = log2(curdiff);
+ step = log2f(curdiff);
if (dist > uplim[w*16+g])
step = -step;
scf += step;
SingleChannelElement *sce,
const float lambda)
{
- int start = 0, i, w, w2, g;
+ int i, w, w2, g;
int minq = 255;
memset(sce->sf_idx, 0, sizeof(sce->sf_idx));
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
- start = w*128;
for (g = 0; g < sce->ics.num_swb; g++) {
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
- FFPsyBand *band = &s->psy.psy_bands[s->cur_channel*PSY_MAX_BANDS+(w+w2)*16+g];
+ FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g];
if (band->energy <= band->threshold) {
sce->sf_idx[(w+w2)*16+g] = 218;
sce->zeroes[(w+w2)*16+g] = 1;
} else {
- sce->sf_idx[(w+w2)*16+g] = av_clip(SCALE_ONE_POS - SCALE_DIV_512 + log2(band->threshold), 80, 218);
+ sce->sf_idx[(w+w2)*16+g] = av_clip(SCALE_ONE_POS - SCALE_DIV_512 + log2f(band->threshold), 80, 218);
sce->zeroes[(w+w2)*16+g] = 0;
}
minq = FFMIN(minq, sce->sf_idx[(w+w2)*16+g]);
if (!cpe->ch[0].zeroes[w*16+g] && !cpe->ch[1].zeroes[w*16+g]) {
float dist1 = 0.0f, dist2 = 0.0f;
for (w2 = 0; w2 < sce0->ics.group_len[w]; w2++) {
- FFPsyBand *band0 = &s->psy.psy_bands[(s->cur_channel+0)*PSY_MAX_BANDS+(w+w2)*16+g];
- FFPsyBand *band1 = &s->psy.psy_bands[(s->cur_channel+1)*PSY_MAX_BANDS+(w+w2)*16+g];
+ FFPsyBand *band0 = &s->psy.ch[s->cur_channel+0].psy_bands[(w+w2)*16+g];
+ FFPsyBand *band1 = &s->psy.ch[s->cur_channel+1].psy_bands[(w+w2)*16+g];
float minthr = FFMIN(band0->threshold, band1->threshold);
float maxthr = FFMAX(band0->threshold, band1->threshold);
for (i = 0; i < sce0->ics.swb_sizes[g]; i++) {
M[i] = (sce0->coeffs[start+w2*128+i]
+ sce1->coeffs[start+w2*128+i]) * 0.5;
- S[i] = sce0->coeffs[start+w2*128+i]
+ S[i] = M[i]
- sce1->coeffs[start+w2*128+i];
}
abs_pow34_v(L34, sce0->coeffs+start+w2*128, sce0->ics.swb_sizes[g]);