summaryrefslogtreecommitdiff
path: root/libavcodec/aaccoder.c (plain)
blob: abd0b9c636c1725f72d0a62fe4ee1e1aff223ce5
1/*
2 * AAC coefficients encoder
3 * Copyright (C) 2008-2009 Konstantin Shishkov
4 *
5 * This file is part of FFmpeg.
6 *
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
11 *
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
16 *
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22/**
23 * @file
24 * AAC coefficients encoder
25 */
26
27/***********************************
28 * TODOs:
29 * speedup quantizer selection
30 * add sane pulse detection
31 ***********************************/
32
33#include "libavutil/libm.h" // brought forward to work around cygwin header breakage
34
35#include <float.h>
36
37#include "libavutil/mathematics.h"
38#include "mathops.h"
39#include "avcodec.h"
40#include "put_bits.h"
41#include "aac.h"
42#include "aacenc.h"
43#include "aactab.h"
44#include "aacenctab.h"
45#include "aacenc_utils.h"
46#include "aacenc_quantization.h"
47
48#include "aacenc_is.h"
49#include "aacenc_tns.h"
50#include "aacenc_ltp.h"
51#include "aacenc_pred.h"
52
53#include "libavcodec/aaccoder_twoloop.h"
54
55/* Parameter of f(x) = a*(lambda/100), defines the maximum fourier spread
56 * beyond which no PNS is used (since the SFBs contain tone rather than noise) */
57#define NOISE_SPREAD_THRESHOLD 0.9f
58
59/* Parameter of f(x) = a*(100/lambda), defines how much PNS is allowed to
60 * replace low energy non zero bands */
61#define NOISE_LAMBDA_REPLACE 1.948f
62
63#include "libavcodec/aaccoder_trellis.h"
64
65/**
66 * structure used in optimal codebook search
67 */
68typedef struct BandCodingPath {
69 int prev_idx; ///< pointer to the previous path point
70 float cost; ///< path cost
71 int run;
72} BandCodingPath;
73
74/**
75 * Encode band info for single window group bands.
76 */
77static void encode_window_bands_info(AACEncContext *s, SingleChannelElement *sce,
78 int win, int group_len, const float lambda)
79{
80 BandCodingPath path[120][CB_TOT_ALL];
81 int w, swb, cb, start, size;
82 int i, j;
83 const int max_sfb = sce->ics.max_sfb;
84 const int run_bits = sce->ics.num_windows == 1 ? 5 : 3;
85 const int run_esc = (1 << run_bits) - 1;
86 int idx, ppos, count;
87 int stackrun[120], stackcb[120], stack_len;
88 float next_minrd = INFINITY;
89 int next_mincb = 0;
90
91 s->abs_pow34(s->scoefs, sce->coeffs, 1024);
92 start = win*128;
93 for (cb = 0; cb < CB_TOT_ALL; cb++) {
94 path[0][cb].cost = 0.0f;
95 path[0][cb].prev_idx = -1;
96 path[0][cb].run = 0;
97 }
98 for (swb = 0; swb < max_sfb; swb++) {
99 size = sce->ics.swb_sizes[swb];
100 if (sce->zeroes[win*16 + swb]) {
101 for (cb = 0; cb < CB_TOT_ALL; cb++) {
102 path[swb+1][cb].prev_idx = cb;
103 path[swb+1][cb].cost = path[swb][cb].cost;
104 path[swb+1][cb].run = path[swb][cb].run + 1;
105 }
106 } else {
107 float minrd = next_minrd;
108 int mincb = next_mincb;
109 next_minrd = INFINITY;
110 next_mincb = 0;
111 for (cb = 0; cb < CB_TOT_ALL; cb++) {
112 float cost_stay_here, cost_get_here;
113 float rd = 0.0f;
114 if (cb >= 12 && sce->band_type[win*16+swb] < aac_cb_out_map[cb] ||
115 cb < aac_cb_in_map[sce->band_type[win*16+swb]] && sce->band_type[win*16+swb] > aac_cb_out_map[cb]) {
116 path[swb+1][cb].prev_idx = -1;
117 path[swb+1][cb].cost = INFINITY;
118 path[swb+1][cb].run = path[swb][cb].run + 1;
119 continue;
120 }
121 for (w = 0; w < group_len; w++) {
122 FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[(win+w)*16+swb];
123 rd += quantize_band_cost(s, &sce->coeffs[start + w*128],
124 &s->scoefs[start + w*128], size,
125 sce->sf_idx[(win+w)*16+swb], aac_cb_out_map[cb],
126 lambda / band->threshold, INFINITY, NULL, NULL, 0);
127 }
128 cost_stay_here = path[swb][cb].cost + rd;
129 cost_get_here = minrd + rd + run_bits + 4;
130 if ( run_value_bits[sce->ics.num_windows == 8][path[swb][cb].run]
131 != run_value_bits[sce->ics.num_windows == 8][path[swb][cb].run+1])
132 cost_stay_here += run_bits;
133 if (cost_get_here < cost_stay_here) {
134 path[swb+1][cb].prev_idx = mincb;
135 path[swb+1][cb].cost = cost_get_here;
136 path[swb+1][cb].run = 1;
137 } else {
138 path[swb+1][cb].prev_idx = cb;
139 path[swb+1][cb].cost = cost_stay_here;
140 path[swb+1][cb].run = path[swb][cb].run + 1;
141 }
142 if (path[swb+1][cb].cost < next_minrd) {
143 next_minrd = path[swb+1][cb].cost;
144 next_mincb = cb;
145 }
146 }
147 }
148 start += sce->ics.swb_sizes[swb];
149 }
150
151 //convert resulting path from backward-linked list
152 stack_len = 0;
153 idx = 0;
154 for (cb = 1; cb < CB_TOT_ALL; cb++)
155 if (path[max_sfb][cb].cost < path[max_sfb][idx].cost)
156 idx = cb;
157 ppos = max_sfb;
158 while (ppos > 0) {
159 av_assert1(idx >= 0);
160 cb = idx;
161 stackrun[stack_len] = path[ppos][cb].run;
162 stackcb [stack_len] = cb;
163 idx = path[ppos-path[ppos][cb].run+1][cb].prev_idx;
164 ppos -= path[ppos][cb].run;
165 stack_len++;
166 }
167 //perform actual band info encoding
168 start = 0;
169 for (i = stack_len - 1; i >= 0; i--) {
170 cb = aac_cb_out_map[stackcb[i]];
171 put_bits(&s->pb, 4, cb);
172 count = stackrun[i];
173 memset(sce->zeroes + win*16 + start, !cb, count);
174 //XXX: memset when band_type is also uint8_t
175 for (j = 0; j < count; j++) {
176 sce->band_type[win*16 + start] = cb;
177 start++;
178 }
179 while (count >= run_esc) {
180 put_bits(&s->pb, run_bits, run_esc);
181 count -= run_esc;
182 }
183 put_bits(&s->pb, run_bits, count);
184 }
185}
186
187
188typedef struct TrellisPath {
189 float cost;
190 int prev;
191} TrellisPath;
192
193#define TRELLIS_STAGES 121
194#define TRELLIS_STATES (SCALE_MAX_DIFF+1)
195
196static void set_special_band_scalefactors(AACEncContext *s, SingleChannelElement *sce)
197{
198 int w, g;
199 int prevscaler_n = -255, prevscaler_i = 0;
200 int bands = 0;
201
202 for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
203 for (g = 0; g < sce->ics.num_swb; g++) {
204 if (sce->zeroes[w*16+g])
205 continue;
206 if (sce->band_type[w*16+g] == INTENSITY_BT || sce->band_type[w*16+g] == INTENSITY_BT2) {
207 sce->sf_idx[w*16+g] = av_clip(roundf(log2f(sce->is_ener[w*16+g])*2), -155, 100);
208 bands++;
209 } else if (sce->band_type[w*16+g] == NOISE_BT) {
210 sce->sf_idx[w*16+g] = av_clip(3+ceilf(log2f(sce->pns_ener[w*16+g])*2), -100, 155);
211 if (prevscaler_n == -255)
212 prevscaler_n = sce->sf_idx[w*16+g];
213 bands++;
214 }
215 }
216 }
217
218 if (!bands)
219 return;
220
221 /* Clip the scalefactor indices */
222 for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
223 for (g = 0; g < sce->ics.num_swb; g++) {
224 if (sce->zeroes[w*16+g])
225 continue;
226 if (sce->band_type[w*16+g] == INTENSITY_BT || sce->band_type[w*16+g] == INTENSITY_BT2) {
227 sce->sf_idx[w*16+g] = prevscaler_i = av_clip(sce->sf_idx[w*16+g], prevscaler_i - SCALE_MAX_DIFF, prevscaler_i + SCALE_MAX_DIFF);
228 } else if (sce->band_type[w*16+g] == NOISE_BT) {
229 sce->sf_idx[w*16+g] = prevscaler_n = av_clip(sce->sf_idx[w*16+g], prevscaler_n - SCALE_MAX_DIFF, prevscaler_n + SCALE_MAX_DIFF);
230 }
231 }
232 }
233}
234
235static void search_for_quantizers_anmr(AVCodecContext *avctx, AACEncContext *s,
236 SingleChannelElement *sce,
237 const float lambda)
238{
239 int q, w, w2, g, start = 0;
240 int i, j;
241 int idx;
242 TrellisPath paths[TRELLIS_STAGES][TRELLIS_STATES];
243 int bandaddr[TRELLIS_STAGES];
244 int minq;
245 float mincost;
246 float q0f = FLT_MAX, q1f = 0.0f, qnrgf = 0.0f;
247 int q0, q1, qcnt = 0;
248
249 for (i = 0; i < 1024; i++) {
250 float t = fabsf(sce->coeffs[i]);
251 if (t > 0.0f) {
252 q0f = FFMIN(q0f, t);
253 q1f = FFMAX(q1f, t);
254 qnrgf += t*t;
255 qcnt++;
256 }
257 }
258
259 if (!qcnt) {
260 memset(sce->sf_idx, 0, sizeof(sce->sf_idx));
261 memset(sce->zeroes, 1, sizeof(sce->zeroes));
262 return;
263 }
264
265 //minimum scalefactor index is when minimum nonzero coefficient after quantizing is not clipped
266 q0 = av_clip(coef2minsf(q0f), 0, SCALE_MAX_POS-1);
267 //maximum scalefactor index is when maximum coefficient after quantizing is still not zero
268 q1 = av_clip(coef2maxsf(q1f), 1, SCALE_MAX_POS);
269 if (q1 - q0 > 60) {
270 int q0low = q0;
271 int q1high = q1;
272 //minimum scalefactor index is when maximum nonzero coefficient after quantizing is not clipped
273 int qnrg = av_clip_uint8(log2f(sqrtf(qnrgf/qcnt))*4 - 31 + SCALE_ONE_POS - SCALE_DIV_512);
274 q1 = qnrg + 30;
275 q0 = qnrg - 30;
276 if (q0 < q0low) {
277 q1 += q0low - q0;
278 q0 = q0low;
279 } else if (q1 > q1high) {
280 q0 -= q1 - q1high;
281 q1 = q1high;
282 }
283 }
284 // q0 == q1 isn't really a legal situation
285 if (q0 == q1) {
286 // the following is indirect but guarantees q1 != q0 && q1 near q0
287 q1 = av_clip(q0+1, 1, SCALE_MAX_POS);
288 q0 = av_clip(q1-1, 0, SCALE_MAX_POS - 1);
289 }
290
291 for (i = 0; i < TRELLIS_STATES; i++) {
292 paths[0][i].cost = 0.0f;
293 paths[0][i].prev = -1;
294 }
295 for (j = 1; j < TRELLIS_STAGES; j++) {
296 for (i = 0; i < TRELLIS_STATES; i++) {
297 paths[j][i].cost = INFINITY;
298 paths[j][i].prev = -2;
299 }
300 }
301 idx = 1;
302 s->abs_pow34(s->scoefs, sce->coeffs, 1024);
303 for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
304 start = w*128;
305 for (g = 0; g < sce->ics.num_swb; g++) {
306 const float *coefs = &sce->coeffs[start];
307 float qmin, qmax;
308 int nz = 0;
309
310 bandaddr[idx] = w * 16 + g;
311 qmin = INT_MAX;
312 qmax = 0.0f;
313 for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
314 FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g];
315 if (band->energy <= band->threshold || band->threshold == 0.0f) {
316 sce->zeroes[(w+w2)*16+g] = 1;
317 continue;
318 }
319 sce->zeroes[(w+w2)*16+g] = 0;
320 nz = 1;
321 for (i = 0; i < sce->ics.swb_sizes[g]; i++) {
322 float t = fabsf(coefs[w2*128+i]);
323 if (t > 0.0f)
324 qmin = FFMIN(qmin, t);
325 qmax = FFMAX(qmax, t);
326 }
327 }
328 if (nz) {
329 int minscale, maxscale;
330 float minrd = INFINITY;
331 float maxval;
332 //minimum scalefactor index is when minimum nonzero coefficient after quantizing is not clipped
333 minscale = coef2minsf(qmin);
334 //maximum scalefactor index is when maximum coefficient after quantizing is still not zero
335 maxscale = coef2maxsf(qmax);
336 minscale = av_clip(minscale - q0, 0, TRELLIS_STATES - 1);
337 maxscale = av_clip(maxscale - q0, 0, TRELLIS_STATES);
338 if (minscale == maxscale) {
339 maxscale = av_clip(minscale+1, 1, TRELLIS_STATES);
340 minscale = av_clip(maxscale-1, 0, TRELLIS_STATES - 1);
341 }
342 maxval = find_max_val(sce->ics.group_len[w], sce->ics.swb_sizes[g], s->scoefs+start);
343 for (q = minscale; q < maxscale; q++) {
344 float dist = 0;
345 int cb = find_min_book(maxval, sce->sf_idx[w*16+g]);
346 for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
347 FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g];
348 dist += quantize_band_cost(s, coefs + w2*128, s->scoefs + start + w2*128, sce->ics.swb_sizes[g],
349 q + q0, cb, lambda / band->threshold, INFINITY, NULL, NULL, 0);
350 }
351 minrd = FFMIN(minrd, dist);
352
353 for (i = 0; i < q1 - q0; i++) {
354 float cost;
355 cost = paths[idx - 1][i].cost + dist
356 + ff_aac_scalefactor_bits[q - i + SCALE_DIFF_ZERO];
357 if (cost < paths[idx][q].cost) {
358 paths[idx][q].cost = cost;
359 paths[idx][q].prev = i;
360 }
361 }
362 }
363 } else {
364 for (q = 0; q < q1 - q0; q++) {
365 paths[idx][q].cost = paths[idx - 1][q].cost + 1;
366 paths[idx][q].prev = q;
367 }
368 }
369 sce->zeroes[w*16+g] = !nz;
370 start += sce->ics.swb_sizes[g];
371 idx++;
372 }
373 }
374 idx--;
375 mincost = paths[idx][0].cost;
376 minq = 0;
377 for (i = 1; i < TRELLIS_STATES; i++) {
378 if (paths[idx][i].cost < mincost) {
379 mincost = paths[idx][i].cost;
380 minq = i;
381 }
382 }
383 while (idx) {
384 sce->sf_idx[bandaddr[idx]] = minq + q0;
385 minq = FFMAX(paths[idx][minq].prev, 0);
386 idx--;
387 }
388 //set the same quantizers inside window groups
389 for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w])
390 for (g = 0; g < sce->ics.num_swb; g++)
391 for (w2 = 1; w2 < sce->ics.group_len[w]; w2++)
392 sce->sf_idx[(w+w2)*16+g] = sce->sf_idx[w*16+g];
393}
394
395static void search_for_quantizers_fast(AVCodecContext *avctx, AACEncContext *s,
396 SingleChannelElement *sce,
397 const float lambda)
398{
399 int start = 0, i, w, w2, g;
400 int destbits = avctx->bit_rate * 1024.0 / avctx->sample_rate / avctx->channels * (lambda / 120.f);
401 float dists[128] = { 0 }, uplims[128] = { 0 };
402 float maxvals[128];
403 int fflag, minscaler;
404 int its = 0;
405 int allz = 0;
406 float minthr = INFINITY;
407
408 // for values above this the decoder might end up in an endless loop
409 // due to always having more bits than what can be encoded.
410 destbits = FFMIN(destbits, 5800);
411 //some heuristic to determine initial quantizers will reduce search time
412 //determine zero bands and upper limits
413 for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
414 start = 0;
415 for (g = 0; g < sce->ics.num_swb; g++) {
416 int nz = 0;
417 float uplim = 0.0f, energy = 0.0f;
418 for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
419 FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g];
420 uplim += band->threshold;
421 energy += band->energy;
422 if (band->energy <= band->threshold || band->threshold == 0.0f) {
423 sce->zeroes[(w+w2)*16+g] = 1;
424 continue;
425 }
426 nz = 1;
427 }
428 uplims[w*16+g] = uplim *512;
429 sce->band_type[w*16+g] = 0;
430 sce->zeroes[w*16+g] = !nz;
431 if (nz)
432 minthr = FFMIN(minthr, uplim);
433 allz |= nz;
434 start += sce->ics.swb_sizes[g];
435 }
436 }
437 for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
438 for (g = 0; g < sce->ics.num_swb; g++) {
439 if (sce->zeroes[w*16+g]) {
440 sce->sf_idx[w*16+g] = SCALE_ONE_POS;
441 continue;
442 }
443 sce->sf_idx[w*16+g] = SCALE_ONE_POS + FFMIN(log2f(uplims[w*16+g]/minthr)*4,59);
444 }
445 }
446
447 if (!allz)
448 return;
449 s->abs_pow34(s->scoefs, sce->coeffs, 1024);
450 ff_quantize_band_cost_cache_init(s);
451
452 for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
453 start = w*128;
454 for (g = 0; g < sce->ics.num_swb; g++) {
455 const float *scaled = s->scoefs + start;
456 maxvals[w*16+g] = find_max_val(sce->ics.group_len[w], sce->ics.swb_sizes[g], scaled);
457 start += sce->ics.swb_sizes[g];
458 }
459 }
460
461 //perform two-loop search
462 //outer loop - improve quality
463 do {
464 int tbits, qstep;
465 minscaler = sce->sf_idx[0];
466 //inner loop - quantize spectrum to fit into given number of bits
467 qstep = its ? 1 : 32;
468 do {
469 int prev = -1;
470 tbits = 0;
471 for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
472 start = w*128;
473 for (g = 0; g < sce->ics.num_swb; g++) {
474 const float *coefs = sce->coeffs + start;
475 const float *scaled = s->scoefs + start;
476 int bits = 0;
477 int cb;
478 float dist = 0.0f;
479
480 if (sce->zeroes[w*16+g] || sce->sf_idx[w*16+g] >= 218) {
481 start += sce->ics.swb_sizes[g];
482 continue;
483 }
484 minscaler = FFMIN(minscaler, sce->sf_idx[w*16+g]);
485 cb = find_min_book(maxvals[w*16+g], sce->sf_idx[w*16+g]);
486 for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
487 int b;
488 dist += quantize_band_cost_cached(s, w + w2, g,
489 coefs + w2*128,
490 scaled + w2*128,
491 sce->ics.swb_sizes[g],
492 sce->sf_idx[w*16+g],
493 cb, 1.0f, INFINITY,
494 &b, NULL, 0);
495 bits += b;
496 }
497 dists[w*16+g] = dist - bits;
498 if (prev != -1) {
499 bits += ff_aac_scalefactor_bits[sce->sf_idx[w*16+g] - prev + SCALE_DIFF_ZERO];
500 }
501 tbits += bits;
502 start += sce->ics.swb_sizes[g];
503 prev = sce->sf_idx[w*16+g];
504 }
505 }
506 if (tbits > destbits) {
507 for (i = 0; i < 128; i++)
508 if (sce->sf_idx[i] < 218 - qstep)
509 sce->sf_idx[i] += qstep;
510 } else {
511 for (i = 0; i < 128; i++)
512 if (sce->sf_idx[i] > 60 - qstep)
513 sce->sf_idx[i] -= qstep;
514 }
515 qstep >>= 1;
516 if (!qstep && tbits > destbits*1.02 && sce->sf_idx[0] < 217)
517 qstep = 1;
518 } while (qstep);
519
520 fflag = 0;
521 minscaler = av_clip(minscaler, 60, 255 - SCALE_MAX_DIFF);
522
523 for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
524 for (g = 0; g < sce->ics.num_swb; g++) {
525 int prevsc = sce->sf_idx[w*16+g];
526 if (dists[w*16+g] > uplims[w*16+g] && sce->sf_idx[w*16+g] > 60) {
527 if (find_min_book(maxvals[w*16+g], sce->sf_idx[w*16+g]-1))
528 sce->sf_idx[w*16+g]--;
529 else //Try to make sure there is some energy in every band
530 sce->sf_idx[w*16+g]-=2;
531 }
532 sce->sf_idx[w*16+g] = av_clip(sce->sf_idx[w*16+g], minscaler, minscaler + SCALE_MAX_DIFF);
533 sce->sf_idx[w*16+g] = FFMIN(sce->sf_idx[w*16+g], 219);
534 if (sce->sf_idx[w*16+g] != prevsc)
535 fflag = 1;
536 sce->band_type[w*16+g] = find_min_book(maxvals[w*16+g], sce->sf_idx[w*16+g]);
537 }
538 }
539 its++;
540 } while (fflag && its < 10);
541}
542
543static void search_for_pns(AACEncContext *s, AVCodecContext *avctx, SingleChannelElement *sce)
544{
545 FFPsyBand *band;
546 int w, g, w2, i;
547 int wlen = 1024 / sce->ics.num_windows;
548 int bandwidth, cutoff;
549 float *PNS = &s->scoefs[0*128], *PNS34 = &s->scoefs[1*128];
550 float *NOR34 = &s->scoefs[3*128];
551 uint8_t nextband[128];
552 const float lambda = s->lambda;
553 const float freq_mult = avctx->sample_rate*0.5f/wlen;
554 const float thr_mult = NOISE_LAMBDA_REPLACE*(100.0f/lambda);
555 const float spread_threshold = FFMIN(0.75f, NOISE_SPREAD_THRESHOLD*FFMAX(0.5f, lambda/100.f));
556 const float dist_bias = av_clipf(4.f * 120 / lambda, 0.25f, 4.0f);
557 const float pns_transient_energy_r = FFMIN(0.7f, lambda / 140.f);
558
559 int refbits = avctx->bit_rate * 1024.0 / avctx->sample_rate
560 / ((avctx->flags & AV_CODEC_FLAG_QSCALE) ? 2.0f : avctx->channels)
561 * (lambda / 120.f);
562
563 /** Keep this in sync with twoloop's cutoff selection */
564 float rate_bandwidth_multiplier = 1.5f;
565 int prev = -1000, prev_sf = -1;
566 int frame_bit_rate = (avctx->flags & AV_CODEC_FLAG_QSCALE)
567 ? (refbits * rate_bandwidth_multiplier * avctx->sample_rate / 1024)
568 : (avctx->bit_rate / avctx->channels);
569
570 frame_bit_rate *= 1.15f;
571
572 if (avctx->cutoff > 0) {
573 bandwidth = avctx->cutoff;
574 } else {
575 bandwidth = FFMAX(3000, AAC_CUTOFF_FROM_BITRATE(frame_bit_rate, 1, avctx->sample_rate));
576 }
577
578 cutoff = bandwidth * 2 * wlen / avctx->sample_rate;
579
580 memcpy(sce->band_alt, sce->band_type, sizeof(sce->band_type));
581 ff_init_nextband_map(sce, nextband);
582 for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
583 int wstart = w*128;
584 for (g = 0; g < sce->ics.num_swb; g++) {
585 int noise_sfi;
586 float dist1 = 0.0f, dist2 = 0.0f, noise_amp;
587 float pns_energy = 0.0f, pns_tgt_energy, energy_ratio, dist_thresh;
588 float sfb_energy = 0.0f, threshold = 0.0f, spread = 2.0f;
589 float min_energy = -1.0f, max_energy = 0.0f;
590 const int start = wstart+sce->ics.swb_offset[g];
591 const float freq = (start-wstart)*freq_mult;
592 const float freq_boost = FFMAX(0.88f*freq/NOISE_LOW_LIMIT, 1.0f);
593 if (freq < NOISE_LOW_LIMIT || (start-wstart) >= cutoff) {
594 if (!sce->zeroes[w*16+g])
595 prev_sf = sce->sf_idx[w*16+g];
596 continue;
597 }
598 for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
599 band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g];
600 sfb_energy += band->energy;
601 spread = FFMIN(spread, band->spread);
602 threshold += band->threshold;
603 if (!w2) {
604 min_energy = max_energy = band->energy;
605 } else {
606 min_energy = FFMIN(min_energy, band->energy);
607 max_energy = FFMAX(max_energy, band->energy);
608 }
609 }
610
611 /* Ramps down at ~8000Hz and loosens the dist threshold */
612 dist_thresh = av_clipf(2.5f*NOISE_LOW_LIMIT/freq, 0.5f, 2.5f) * dist_bias;
613
614 /* PNS is acceptable when all of these are true:
615 * 1. high spread energy (noise-like band)
616 * 2. near-threshold energy (high PE means the random nature of PNS content will be noticed)
617 * 3. on short window groups, all windows have similar energy (variations in energy would be destroyed by PNS)
618 *
619 * At this stage, point 2 is relaxed for zeroed bands near the noise threshold (hole avoidance is more important)
620 */
621 if ((!sce->zeroes[w*16+g] && !ff_sfdelta_can_remove_band(sce, nextband, prev_sf, w*16+g)) ||
622 ((sce->zeroes[w*16+g] || !sce->band_alt[w*16+g]) && sfb_energy < threshold*sqrtf(1.0f/freq_boost)) || spread < spread_threshold ||
623 (!sce->zeroes[w*16+g] && sce->band_alt[w*16+g] && sfb_energy > threshold*thr_mult*freq_boost) ||
624 min_energy < pns_transient_energy_r * max_energy ) {
625 sce->pns_ener[w*16+g] = sfb_energy;
626 if (!sce->zeroes[w*16+g])
627 prev_sf = sce->sf_idx[w*16+g];
628 continue;
629 }
630
631 pns_tgt_energy = sfb_energy*FFMIN(1.0f, spread*spread);
632 noise_sfi = av_clip(roundf(log2f(pns_tgt_energy)*2), -100, 155); /* Quantize */
633 noise_amp = -ff_aac_pow2sf_tab[noise_sfi + POW_SF2_ZERO]; /* Dequantize */
634 if (prev != -1000) {
635 int noise_sfdiff = noise_sfi - prev + SCALE_DIFF_ZERO;
636 if (noise_sfdiff < 0 || noise_sfdiff > 2*SCALE_MAX_DIFF) {
637 if (!sce->zeroes[w*16+g])
638 prev_sf = sce->sf_idx[w*16+g];
639 continue;
640 }
641 }
642 for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
643 float band_energy, scale, pns_senergy;
644 const int start_c = (w+w2)*128+sce->ics.swb_offset[g];
645 band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g];
646 for (i = 0; i < sce->ics.swb_sizes[g]; i++) {
647 s->random_state = lcg_random(s->random_state);
648 PNS[i] = s->random_state;
649 }
650 band_energy = s->fdsp->scalarproduct_float(PNS, PNS, sce->ics.swb_sizes[g]);
651 scale = noise_amp/sqrtf(band_energy);
652 s->fdsp->vector_fmul_scalar(PNS, PNS, scale, sce->ics.swb_sizes[g]);
653 pns_senergy = s->fdsp->scalarproduct_float(PNS, PNS, sce->ics.swb_sizes[g]);
654 pns_energy += pns_senergy;
655 s->abs_pow34(NOR34, &sce->coeffs[start_c], sce->ics.swb_sizes[g]);
656 s->abs_pow34(PNS34, PNS, sce->ics.swb_sizes[g]);
657 dist1 += quantize_band_cost(s, &sce->coeffs[start_c],
658 NOR34,
659 sce->ics.swb_sizes[g],
660 sce->sf_idx[(w+w2)*16+g],
661 sce->band_alt[(w+w2)*16+g],
662 lambda/band->threshold, INFINITY, NULL, NULL, 0);
663 /* Estimate rd on average as 5 bits for SF, 4 for the CB, plus spread energy * lambda/thr */
664 dist2 += band->energy/(band->spread*band->spread)*lambda*dist_thresh/band->threshold;
665 }
666 if (g && sce->band_type[w*16+g-1] == NOISE_BT) {
667 dist2 += 5;
668 } else {
669 dist2 += 9;
670 }
671 energy_ratio = pns_tgt_energy/pns_energy; /* Compensates for quantization error */
672 sce->pns_ener[w*16+g] = energy_ratio*pns_tgt_energy;
673 if (sce->zeroes[w*16+g] || !sce->band_alt[w*16+g] || (energy_ratio > 0.85f && energy_ratio < 1.25f && dist2 < dist1)) {
674 sce->band_type[w*16+g] = NOISE_BT;
675 sce->zeroes[w*16+g] = 0;
676 prev = noise_sfi;
677 } else {
678 if (!sce->zeroes[w*16+g])
679 prev_sf = sce->sf_idx[w*16+g];
680 }
681 }
682 }
683}
684
685static void mark_pns(AACEncContext *s, AVCodecContext *avctx, SingleChannelElement *sce)
686{
687 FFPsyBand *band;
688 int w, g, w2;
689 int wlen = 1024 / sce->ics.num_windows;
690 int bandwidth, cutoff;
691 const float lambda = s->lambda;
692 const float freq_mult = avctx->sample_rate*0.5f/wlen;
693 const float spread_threshold = FFMIN(0.75f, NOISE_SPREAD_THRESHOLD*FFMAX(0.5f, lambda/100.f));
694 const float pns_transient_energy_r = FFMIN(0.7f, lambda / 140.f);
695
696 int refbits = avctx->bit_rate * 1024.0 / avctx->sample_rate
697 / ((avctx->flags & AV_CODEC_FLAG_QSCALE) ? 2.0f : avctx->channels)
698 * (lambda / 120.f);
699
700 /** Keep this in sync with twoloop's cutoff selection */
701 float rate_bandwidth_multiplier = 1.5f;
702 int frame_bit_rate = (avctx->flags & AV_CODEC_FLAG_QSCALE)
703 ? (refbits * rate_bandwidth_multiplier * avctx->sample_rate / 1024)
704 : (avctx->bit_rate / avctx->channels);
705
706 frame_bit_rate *= 1.15f;
707
708 if (avctx->cutoff > 0) {
709 bandwidth = avctx->cutoff;
710 } else {
711 bandwidth = FFMAX(3000, AAC_CUTOFF_FROM_BITRATE(frame_bit_rate, 1, avctx->sample_rate));
712 }
713
714 cutoff = bandwidth * 2 * wlen / avctx->sample_rate;
715
716 memcpy(sce->band_alt, sce->band_type, sizeof(sce->band_type));
717 for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
718 for (g = 0; g < sce->ics.num_swb; g++) {
719 float sfb_energy = 0.0f, threshold = 0.0f, spread = 2.0f;
720 float min_energy = -1.0f, max_energy = 0.0f;
721 const int start = sce->ics.swb_offset[g];
722 const float freq = start*freq_mult;
723 const float freq_boost = FFMAX(0.88f*freq/NOISE_LOW_LIMIT, 1.0f);
724 if (freq < NOISE_LOW_LIMIT || start >= cutoff) {
725 sce->can_pns[w*16+g] = 0;
726 continue;
727 }
728 for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
729 band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g];
730 sfb_energy += band->energy;
731 spread = FFMIN(spread, band->spread);
732 threshold += band->threshold;
733 if (!w2) {
734 min_energy = max_energy = band->energy;
735 } else {
736 min_energy = FFMIN(min_energy, band->energy);
737 max_energy = FFMAX(max_energy, band->energy);
738 }
739 }
740
741 /* PNS is acceptable when all of these are true:
742 * 1. high spread energy (noise-like band)
743 * 2. near-threshold energy (high PE means the random nature of PNS content will be noticed)
744 * 3. on short window groups, all windows have similar energy (variations in energy would be destroyed by PNS)
745 */
746 sce->pns_ener[w*16+g] = sfb_energy;
747 if (sfb_energy < threshold*sqrtf(1.5f/freq_boost) || spread < spread_threshold || min_energy < pns_transient_energy_r * max_energy) {
748 sce->can_pns[w*16+g] = 0;
749 } else {
750 sce->can_pns[w*16+g] = 1;
751 }
752 }
753 }
754}
755
756static void search_for_ms(AACEncContext *s, ChannelElement *cpe)
757{
758 int start = 0, i, w, w2, g, sid_sf_boost, prev_mid, prev_side;
759 uint8_t nextband0[128], nextband1[128];
760 float *M = s->scoefs + 128*0, *S = s->scoefs + 128*1;
761 float *L34 = s->scoefs + 128*2, *R34 = s->scoefs + 128*3;
762 float *M34 = s->scoefs + 128*4, *S34 = s->scoefs + 128*5;
763 const float lambda = s->lambda;
764 const float mslambda = FFMIN(1.0f, lambda / 120.f);
765 SingleChannelElement *sce0 = &cpe->ch[0];
766 SingleChannelElement *sce1 = &cpe->ch[1];
767 if (!cpe->common_window)
768 return;
769
770 /** Scout out next nonzero bands */
771 ff_init_nextband_map(sce0, nextband0);
772 ff_init_nextband_map(sce1, nextband1);
773
774 prev_mid = sce0->sf_idx[0];
775 prev_side = sce1->sf_idx[0];
776 for (w = 0; w < sce0->ics.num_windows; w += sce0->ics.group_len[w]) {
777 start = 0;
778 for (g = 0; g < sce0->ics.num_swb; g++) {
779 float bmax = bval2bmax(g * 17.0f / sce0->ics.num_swb) / 0.0045f;
780 if (!cpe->is_mask[w*16+g])
781 cpe->ms_mask[w*16+g] = 0;
782 if (!sce0->zeroes[w*16+g] && !sce1->zeroes[w*16+g] && !cpe->is_mask[w*16+g]) {
783 float Mmax = 0.0f, Smax = 0.0f;
784
785 /* Must compute mid/side SF and book for the whole window group */
786 for (w2 = 0; w2 < sce0->ics.group_len[w]; w2++) {
787 for (i = 0; i < sce0->ics.swb_sizes[g]; i++) {
788 M[i] = (sce0->coeffs[start+(w+w2)*128+i]
789 + sce1->coeffs[start+(w+w2)*128+i]) * 0.5;
790 S[i] = M[i]
791 - sce1->coeffs[start+(w+w2)*128+i];
792 }
793 s->abs_pow34(M34, M, sce0->ics.swb_sizes[g]);
794 s->abs_pow34(S34, S, sce0->ics.swb_sizes[g]);
795 for (i = 0; i < sce0->ics.swb_sizes[g]; i++ ) {
796 Mmax = FFMAX(Mmax, M34[i]);
797 Smax = FFMAX(Smax, S34[i]);
798 }
799 }
800
801 for (sid_sf_boost = 0; sid_sf_boost < 4; sid_sf_boost++) {
802 float dist1 = 0.0f, dist2 = 0.0f;
803 int B0 = 0, B1 = 0;
804 int minidx;
805 int mididx, sididx;
806 int midcb, sidcb;
807
808 minidx = FFMIN(sce0->sf_idx[w*16+g], sce1->sf_idx[w*16+g]);
809 mididx = av_clip(minidx, 0, SCALE_MAX_POS - SCALE_DIV_512);
810 sididx = av_clip(minidx - sid_sf_boost * 3, 0, SCALE_MAX_POS - SCALE_DIV_512);
811 if (sce0->band_type[w*16+g] != NOISE_BT && sce1->band_type[w*16+g] != NOISE_BT
812 && ( !ff_sfdelta_can_replace(sce0, nextband0, prev_mid, mididx, w*16+g)
813 || !ff_sfdelta_can_replace(sce1, nextband1, prev_side, sididx, w*16+g))) {
814 /* scalefactor range violation, bad stuff, will decrease quality unacceptably */
815 continue;
816 }
817
818 midcb = find_min_book(Mmax, mididx);
819 sidcb = find_min_book(Smax, sididx);
820
821 /* No CB can be zero */
822 midcb = FFMAX(1,midcb);
823 sidcb = FFMAX(1,sidcb);
824
825 for (w2 = 0; w2 < sce0->ics.group_len[w]; w2++) {
826 FFPsyBand *band0 = &s->psy.ch[s->cur_channel+0].psy_bands[(w+w2)*16+g];
827 FFPsyBand *band1 = &s->psy.ch[s->cur_channel+1].psy_bands[(w+w2)*16+g];
828 float minthr = FFMIN(band0->threshold, band1->threshold);
829 int b1,b2,b3,b4;
830 for (i = 0; i < sce0->ics.swb_sizes[g]; i++) {
831 M[i] = (sce0->coeffs[start+(w+w2)*128+i]
832 + sce1->coeffs[start+(w+w2)*128+i]) * 0.5;
833 S[i] = M[i]
834 - sce1->coeffs[start+(w+w2)*128+i];
835 }
836
837 s->abs_pow34(L34, sce0->coeffs+start+(w+w2)*128, sce0->ics.swb_sizes[g]);
838 s->abs_pow34(R34, sce1->coeffs+start+(w+w2)*128, sce0->ics.swb_sizes[g]);
839 s->abs_pow34(M34, M, sce0->ics.swb_sizes[g]);
840 s->abs_pow34(S34, S, sce0->ics.swb_sizes[g]);
841 dist1 += quantize_band_cost(s, &sce0->coeffs[start + (w+w2)*128],
842 L34,
843 sce0->ics.swb_sizes[g],
844 sce0->sf_idx[w*16+g],
845 sce0->band_type[w*16+g],
846 lambda / band0->threshold, INFINITY, &b1, NULL, 0);
847 dist1 += quantize_band_cost(s, &sce1->coeffs[start + (w+w2)*128],
848 R34,
849 sce1->ics.swb_sizes[g],
850 sce1->sf_idx[w*16+g],
851 sce1->band_type[w*16+g],
852 lambda / band1->threshold, INFINITY, &b2, NULL, 0);
853 dist2 += quantize_band_cost(s, M,
854 M34,
855 sce0->ics.swb_sizes[g],
856 mididx,
857 midcb,
858 lambda / minthr, INFINITY, &b3, NULL, 0);
859 dist2 += quantize_band_cost(s, S,
860 S34,
861 sce1->ics.swb_sizes[g],
862 sididx,
863 sidcb,
864 mslambda / (minthr * bmax), INFINITY, &b4, NULL, 0);
865 B0 += b1+b2;
866 B1 += b3+b4;
867 dist1 -= b1+b2;
868 dist2 -= b3+b4;
869 }
870 cpe->ms_mask[w*16+g] = dist2 <= dist1 && B1 < B0;
871 if (cpe->ms_mask[w*16+g]) {
872 if (sce0->band_type[w*16+g] != NOISE_BT && sce1->band_type[w*16+g] != NOISE_BT) {
873 sce0->sf_idx[w*16+g] = mididx;
874 sce1->sf_idx[w*16+g] = sididx;
875 sce0->band_type[w*16+g] = midcb;
876 sce1->band_type[w*16+g] = sidcb;
877 } else if ((sce0->band_type[w*16+g] != NOISE_BT) ^ (sce1->band_type[w*16+g] != NOISE_BT)) {
878 /* ms_mask unneeded, and it confuses some decoders */
879 cpe->ms_mask[w*16+g] = 0;
880 }
881 break;
882 } else if (B1 > B0) {
883 /* More boost won't fix this */
884 break;
885 }
886 }
887 }
888 if (!sce0->zeroes[w*16+g] && sce0->band_type[w*16+g] < RESERVED_BT)
889 prev_mid = sce0->sf_idx[w*16+g];
890 if (!sce1->zeroes[w*16+g] && !cpe->is_mask[w*16+g] && sce1->band_type[w*16+g] < RESERVED_BT)
891 prev_side = sce1->sf_idx[w*16+g];
892 start += sce0->ics.swb_sizes[g];
893 }
894 }
895}
896
897AACCoefficientsEncoder ff_aac_coders[AAC_CODER_NB] = {
898 [AAC_CODER_ANMR] = {
899 search_for_quantizers_anmr,
900 encode_window_bands_info,
901 quantize_and_encode_band,
902 ff_aac_encode_tns_info,
903 ff_aac_encode_ltp_info,
904 ff_aac_encode_main_pred,
905 ff_aac_adjust_common_pred,
906 ff_aac_adjust_common_ltp,
907 ff_aac_apply_main_pred,
908 ff_aac_apply_tns,
909 ff_aac_update_ltp,
910 ff_aac_ltp_insert_new_frame,
911 set_special_band_scalefactors,
912 search_for_pns,
913 mark_pns,
914 ff_aac_search_for_tns,
915 ff_aac_search_for_ltp,
916 search_for_ms,
917 ff_aac_search_for_is,
918 ff_aac_search_for_pred,
919 },
920 [AAC_CODER_TWOLOOP] = {
921 search_for_quantizers_twoloop,
922 codebook_trellis_rate,
923 quantize_and_encode_band,
924 ff_aac_encode_tns_info,
925 ff_aac_encode_ltp_info,
926 ff_aac_encode_main_pred,
927 ff_aac_adjust_common_pred,
928 ff_aac_adjust_common_ltp,
929 ff_aac_apply_main_pred,
930 ff_aac_apply_tns,
931 ff_aac_update_ltp,
932 ff_aac_ltp_insert_new_frame,
933 set_special_band_scalefactors,
934 search_for_pns,
935 mark_pns,
936 ff_aac_search_for_tns,
937 ff_aac_search_for_ltp,
938 search_for_ms,
939 ff_aac_search_for_is,
940 ff_aac_search_for_pred,
941 },
942 [AAC_CODER_FAST] = {
943 search_for_quantizers_fast,
944 codebook_trellis_rate,
945 quantize_and_encode_band,
946 ff_aac_encode_tns_info,
947 ff_aac_encode_ltp_info,
948 ff_aac_encode_main_pred,
949 ff_aac_adjust_common_pred,
950 ff_aac_adjust_common_ltp,
951 ff_aac_apply_main_pred,
952 ff_aac_apply_tns,
953 ff_aac_update_ltp,
954 ff_aac_ltp_insert_new_frame,
955 set_special_band_scalefactors,
956 search_for_pns,
957 mark_pns,
958 ff_aac_search_for_tns,
959 ff_aac_search_for_ltp,
960 search_for_ms,
961 ff_aac_search_for_is,
962 ff_aac_search_for_pred,
963 },
964};
965
generated by cgit at 2024-05-18 08:47:19 (GMT)