1 |
|
|
/* |
2 |
|
|
* AAC encoder twoloop coder |
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 encoder twoloop coder |
25 |
|
|
* @author Konstantin Shishkov, Claudio Freire |
26 |
|
|
*/ |
27 |
|
|
|
28 |
|
|
/** |
29 |
|
|
* This file contains a template for the twoloop coder function. |
30 |
|
|
* It needs to be provided, externally, as an already included declaration, |
31 |
|
|
* the following functions from aacenc_quantization/util.h. They're not included |
32 |
|
|
* explicitly here to make it possible to provide alternative implementations: |
33 |
|
|
* - quantize_band_cost |
34 |
|
|
* - abs_pow34_v |
35 |
|
|
* - find_max_val |
36 |
|
|
* - find_min_book |
37 |
|
|
* - find_form_factor |
38 |
|
|
*/ |
39 |
|
|
|
40 |
|
|
#ifndef AVCODEC_AACCODER_TWOLOOP_H |
41 |
|
|
#define AVCODEC_AACCODER_TWOLOOP_H |
42 |
|
|
|
43 |
|
|
#include <float.h> |
44 |
|
|
#include "libavutil/mathematics.h" |
45 |
|
|
#include "mathops.h" |
46 |
|
|
#include "avcodec.h" |
47 |
|
|
#include "put_bits.h" |
48 |
|
|
#include "aac.h" |
49 |
|
|
#include "aacenc.h" |
50 |
|
|
#include "aactab.h" |
51 |
|
|
#include "aacenctab.h" |
52 |
|
|
|
53 |
|
|
/** Frequency in Hz for lower limit of noise substitution **/ |
54 |
|
|
#define NOISE_LOW_LIMIT 4000 |
55 |
|
|
|
56 |
|
|
#define sclip(x) av_clip(x,60,218) |
57 |
|
|
|
58 |
|
|
/* Reflects the cost to change codebooks */ |
59 |
|
|
static inline int ff_pns_bits(SingleChannelElement *sce, int w, int g) |
60 |
|
|
{ |
61 |
|
|
return (!g || !sce->zeroes[w*16+g-1] || !sce->can_pns[w*16+g-1]) ? 9 : 5; |
62 |
|
|
} |
63 |
|
|
|
64 |
|
|
/** |
65 |
|
|
* two-loop quantizers search taken from ISO 13818-7 Appendix C |
66 |
|
|
*/ |
67 |
|
|
static void search_for_quantizers_twoloop(AVCodecContext *avctx, |
68 |
|
|
AACEncContext *s, |
69 |
|
|
SingleChannelElement *sce, |
70 |
|
|
const float lambda) |
71 |
|
|
{ |
72 |
|
|
int start = 0, i, w, w2, g, recomprd; |
73 |
|
|
int destbits = avctx->bit_rate * 1024.0 / avctx->sample_rate |
74 |
|
|
/ ((avctx->flags & AV_CODEC_FLAG_QSCALE) ? 2.0f : avctx->channels) |
75 |
|
|
* (lambda / 120.f); |
76 |
|
|
int refbits = destbits; |
77 |
|
|
int toomanybits, toofewbits; |
78 |
|
|
char nzs[128]; |
79 |
|
|
uint8_t nextband[128]; |
80 |
|
|
int maxsf[128], minsf[128]; |
81 |
|
|
float dists[128] = { 0 }, qenergies[128] = { 0 }, uplims[128], euplims[128], energies[128]; |
82 |
|
|
float maxvals[128], spread_thr_r[128]; |
83 |
|
|
float min_spread_thr_r, max_spread_thr_r; |
84 |
|
|
|
85 |
|
|
/** |
86 |
|
|
* rdlambda controls the maximum tolerated distortion. Twoloop |
87 |
|
|
* will keep iterating until it fails to lower it or it reaches |
88 |
|
|
* ulimit * rdlambda. Keeping it low increases quality on difficult |
89 |
|
|
* signals, but lower it too much, and bits will be taken from weak |
90 |
|
|
* signals, creating "holes". A balance is necessary. |
91 |
|
|
* rdmax and rdmin specify the relative deviation from rdlambda |
92 |
|
|
* allowed for tonality compensation |
93 |
|
|
*/ |
94 |
|
|
float rdlambda = av_clipf(2.0f * 120.f / lambda, 0.0625f, 16.0f); |
95 |
|
|
const float nzslope = 1.5f; |
96 |
|
|
float rdmin = 0.03125f; |
97 |
|
|
float rdmax = 1.0f; |
98 |
|
|
|
99 |
|
|
/** |
100 |
|
|
* sfoffs controls an offset of optmium allocation that will be |
101 |
|
|
* applied based on lambda. Keep it real and modest, the loop |
102 |
|
|
* will take care of the rest, this just accelerates convergence |
103 |
|
|
*/ |
104 |
|
|
float sfoffs = av_clipf(log2f(120.0f / lambda) * 4.0f, -5, 10); |
105 |
|
|
|
106 |
|
|
int fflag, minscaler, maxscaler, nminscaler; |
107 |
|
|
int its = 0; |
108 |
|
|
int maxits = 30; |
109 |
|
|
int allz = 0; |
110 |
|
|
int tbits; |
111 |
|
|
int cutoff = 1024; |
112 |
|
|
int pns_start_pos; |
113 |
|
|
int prev; |
114 |
|
|
|
115 |
|
|
/** |
116 |
|
|
* zeroscale controls a multiplier of the threshold, if band energy |
117 |
|
|
* is below this, a zero is forced. Keep it lower than 1, unless |
118 |
|
|
* low lambda is used, because energy < threshold doesn't mean there's |
119 |
|
|
* no audible signal outright, it's just energy. Also make it rise |
120 |
|
|
* slower than rdlambda, as rdscale has due compensation with |
121 |
|
|
* noisy band depriorization below, whereas zeroing logic is rather dumb |
122 |
|
|
*/ |
123 |
|
|
float zeroscale; |
124 |
|
|
if (lambda > 120.f) { |
125 |
|
|
zeroscale = av_clipf(powf(120.f / lambda, 0.25f), 0.0625f, 1.0f); |
126 |
|
|
} else { |
127 |
|
|
zeroscale = 1.f; |
128 |
|
|
} |
129 |
|
|
|
130 |
|
|
if (s->psy.bitres.alloc >= 0) { |
131 |
|
|
/** |
132 |
|
|
* Psy granted us extra bits to use, from the reservoire |
133 |
|
|
* adjust for lambda except what psy already did |
134 |
|
|
*/ |
135 |
|
|
destbits = s->psy.bitres.alloc |
136 |
|
|
* (lambda / (avctx->global_quality ? avctx->global_quality : 120)); |
137 |
|
|
} |
138 |
|
|
|
139 |
|
|
if (avctx->flags & AV_CODEC_FLAG_QSCALE) { |
140 |
|
|
/** |
141 |
|
|
* Constant Q-scale doesn't compensate MS coding on its own |
142 |
|
|
* No need to be overly precise, this only controls RD |
143 |
|
|
* adjustment CB limits when going overboard |
144 |
|
|
*/ |
145 |
|
|
if (s->options.mid_side && s->cur_type == TYPE_CPE) |
146 |
|
|
destbits *= 2; |
147 |
|
|
|
148 |
|
|
/** |
149 |
|
|
* When using a constant Q-scale, don't adjust bits, just use RD |
150 |
|
|
* Don't let it go overboard, though... 8x psy target is enough |
151 |
|
|
*/ |
152 |
|
|
toomanybits = 5800; |
153 |
|
|
toofewbits = destbits / 16; |
154 |
|
|
|
155 |
|
|
/** Don't offset scalers, just RD */ |
156 |
|
|
sfoffs = sce->ics.num_windows - 1; |
157 |
|
|
rdlambda = sqrtf(rdlambda); |
158 |
|
|
|
159 |
|
|
/** search further */ |
160 |
|
|
maxits *= 2; |
161 |
|
|
} else { |
162 |
|
|
/* When using ABR, be strict, but a reasonable leeway is |
163 |
|
|
* critical to allow RC to smoothly track desired bitrate |
164 |
|
|
* without sudden quality drops that cause audible artifacts. |
165 |
|
|
* Symmetry is also desirable, to avoid systematic bias. |
166 |
|
|
*/ |
167 |
|
|
toomanybits = destbits + destbits/8; |
168 |
|
|
toofewbits = destbits - destbits/8; |
169 |
|
|
|
170 |
|
|
sfoffs = 0; |
171 |
|
|
rdlambda = sqrtf(rdlambda); |
172 |
|
|
} |
173 |
|
|
|
174 |
|
|
/** and zero out above cutoff frequency */ |
175 |
|
|
{ |
176 |
|
|
int wlen = 1024 / sce->ics.num_windows; |
177 |
|
|
int bandwidth; |
178 |
|
|
|
179 |
|
|
/** |
180 |
|
|
* Scale, psy gives us constant quality, this LP only scales |
181 |
|
|
* bitrate by lambda, so we save bits on subjectively unimportant HF |
182 |
|
|
* rather than increase quantization noise. Adjust nominal bitrate |
183 |
|
|
* to effective bitrate according to encoding parameters, |
184 |
|
|
* AAC_CUTOFF_FROM_BITRATE is calibrated for effective bitrate. |
185 |
|
|
*/ |
186 |
|
|
float rate_bandwidth_multiplier = 1.5f; |
187 |
|
|
int frame_bit_rate = (avctx->flags & AV_CODEC_FLAG_QSCALE) |
188 |
|
|
? (refbits * rate_bandwidth_multiplier * avctx->sample_rate / 1024) |
189 |
|
|
: (avctx->bit_rate / avctx->channels); |
190 |
|
|
|
191 |
|
|
/** Compensate for extensions that increase efficiency */ |
192 |
|
|
if (s->options.pns || s->options.intensity_stereo) |
193 |
|
|
frame_bit_rate *= 1.15f; |
194 |
|
|
|
195 |
|
|
if (avctx->cutoff > 0) { |
196 |
|
|
bandwidth = avctx->cutoff; |
197 |
|
|
} else { |
198 |
|
|
bandwidth = FFMAX(3000, AAC_CUTOFF_FROM_BITRATE(frame_bit_rate, 1, avctx->sample_rate)); |
199 |
|
|
s->psy.cutoff = bandwidth; |
200 |
|
|
} |
201 |
|
|
|
202 |
|
|
cutoff = bandwidth * 2 * wlen / avctx->sample_rate; |
203 |
|
|
pns_start_pos = NOISE_LOW_LIMIT * 2 * wlen / avctx->sample_rate; |
204 |
|
|
} |
205 |
|
|
|
206 |
|
|
/** |
207 |
|
|
* for values above this the decoder might end up in an endless loop |
208 |
|
|
* due to always having more bits than what can be encoded. |
209 |
|
|
*/ |
210 |
|
|
destbits = FFMIN(destbits, 5800); |
211 |
|
|
toomanybits = FFMIN(toomanybits, 5800); |
212 |
|
|
toofewbits = FFMIN(toofewbits, 5800); |
213 |
|
|
/** |
214 |
|
|
* XXX: some heuristic to determine initial quantizers will reduce search time |
215 |
|
|
* determine zero bands and upper distortion limits |
216 |
|
|
*/ |
217 |
|
|
min_spread_thr_r = -1; |
218 |
|
|
max_spread_thr_r = -1; |
219 |
|
|
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) { |
220 |
|
|
for (g = start = 0; g < sce->ics.num_swb; start += sce->ics.swb_sizes[g++]) { |
221 |
|
|
int nz = 0; |
222 |
|
|
float uplim = 0.0f, energy = 0.0f, spread = 0.0f; |
223 |
|
|
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) { |
224 |
|
|
FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g]; |
225 |
|
|
if (start >= cutoff || band->energy <= (band->threshold * zeroscale) || band->threshold == 0.0f) { |
226 |
|
|
sce->zeroes[(w+w2)*16+g] = 1; |
227 |
|
|
continue; |
228 |
|
|
} |
229 |
|
|
nz = 1; |
230 |
|
|
} |
231 |
|
|
if (!nz) { |
232 |
|
|
uplim = 0.0f; |
233 |
|
|
} else { |
234 |
|
|
nz = 0; |
235 |
|
|
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) { |
236 |
|
|
FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g]; |
237 |
|
|
if (band->energy <= (band->threshold * zeroscale) || band->threshold == 0.0f) |
238 |
|
|
continue; |
239 |
|
|
uplim += band->threshold; |
240 |
|
|
energy += band->energy; |
241 |
|
|
spread += band->spread; |
242 |
|
|
nz++; |
243 |
|
|
} |
244 |
|
|
} |
245 |
|
|
uplims[w*16+g] = uplim; |
246 |
|
|
energies[w*16+g] = energy; |
247 |
|
|
nzs[w*16+g] = nz; |
248 |
|
|
sce->zeroes[w*16+g] = !nz; |
249 |
|
|
allz |= nz; |
250 |
|
|
if (nz && sce->can_pns[w*16+g]) { |
251 |
|
|
spread_thr_r[w*16+g] = energy * nz / (uplim * spread); |
252 |
|
|
if (min_spread_thr_r < 0) { |
253 |
|
|
min_spread_thr_r = max_spread_thr_r = spread_thr_r[w*16+g]; |
254 |
|
|
} else { |
255 |
|
|
min_spread_thr_r = FFMIN(min_spread_thr_r, spread_thr_r[w*16+g]); |
256 |
|
|
max_spread_thr_r = FFMAX(max_spread_thr_r, spread_thr_r[w*16+g]); |
257 |
|
|
} |
258 |
|
|
} |
259 |
|
|
} |
260 |
|
|
} |
261 |
|
|
|
262 |
|
|
/** Compute initial scalers */ |
263 |
|
|
minscaler = 65535; |
264 |
|
|
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) { |
265 |
|
|
for (g = 0; g < sce->ics.num_swb; g++) { |
266 |
|
|
if (sce->zeroes[w*16+g]) { |
267 |
|
|
sce->sf_idx[w*16+g] = SCALE_ONE_POS; |
268 |
|
|
continue; |
269 |
|
|
} |
270 |
|
|
/** |
271 |
|
|
* log2f-to-distortion ratio is, technically, 2 (1.5db = 4, but it's power vs level so it's 2). |
272 |
|
|
* But, as offsets are applied, low-frequency signals are too sensitive to the induced distortion, |
273 |
|
|
* so we make scaling more conservative by choosing a lower log2f-to-distortion ratio, and thus |
274 |
|
|
* more robust. |
275 |
|
|
*/ |
276 |
|
|
sce->sf_idx[w*16+g] = av_clip( |
277 |
|
|
SCALE_ONE_POS |
278 |
|
|
+ 1.75*log2f(FFMAX(0.00125f,uplims[w*16+g]) / sce->ics.swb_sizes[g]) |
279 |
|
|
+ sfoffs, |
280 |
|
|
60, SCALE_MAX_POS); |
281 |
|
|
minscaler = FFMIN(minscaler, sce->sf_idx[w*16+g]); |
282 |
|
|
} |
283 |
|
|
} |
284 |
|
|
|
285 |
|
|
/** Clip */ |
286 |
|
|
minscaler = av_clip(minscaler, SCALE_ONE_POS - SCALE_DIV_512, SCALE_MAX_POS - SCALE_DIV_512); |
287 |
|
|
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) |
288 |
|
|
for (g = 0; g < sce->ics.num_swb; g++) |
289 |
|
|
if (!sce->zeroes[w*16+g]) |
290 |
|
|
sce->sf_idx[w*16+g] = av_clip(sce->sf_idx[w*16+g], minscaler, minscaler + SCALE_MAX_DIFF - 1); |
291 |
|
|
|
292 |
|
|
if (!allz) |
293 |
|
|
return; |
294 |
|
|
s->abs_pow34(s->scoefs, sce->coeffs, 1024); |
295 |
|
|
ff_quantize_band_cost_cache_init(s); |
296 |
|
|
|
297 |
|
|
for (i = 0; i < sizeof(minsf) / sizeof(minsf[0]); ++i) |
298 |
|
|
minsf[i] = 0; |
299 |
|
|
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) { |
300 |
|
|
start = w*128; |
301 |
|
|
for (g = 0; g < sce->ics.num_swb; g++) { |
302 |
|
|
const float *scaled = s->scoefs + start; |
303 |
|
|
int minsfidx; |
304 |
|
|
maxvals[w*16+g] = find_max_val(sce->ics.group_len[w], sce->ics.swb_sizes[g], scaled); |
305 |
|
|
if (maxvals[w*16+g] > 0) { |
306 |
|
|
minsfidx = coef2minsf(maxvals[w*16+g]); |
307 |
|
|
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) |
308 |
|
|
minsf[(w+w2)*16+g] = minsfidx; |
309 |
|
|
} |
310 |
|
|
start += sce->ics.swb_sizes[g]; |
311 |
|
|
} |
312 |
|
|
} |
313 |
|
|
|
314 |
|
|
/** |
315 |
|
|
* Scale uplims to match rate distortion to quality |
316 |
|
|
* bu applying noisy band depriorization and tonal band priorization. |
317 |
|
|
* Maxval-energy ratio gives us an idea of how noisy/tonal the band is. |
318 |
|
|
* If maxval^2 ~ energy, then that band is mostly noise, and we can relax |
319 |
|
|
* rate distortion requirements. |
320 |
|
|
*/ |
321 |
|
|
memcpy(euplims, uplims, sizeof(euplims)); |
322 |
|
|
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) { |
323 |
|
|
/** psy already priorizes transients to some extent */ |
324 |
|
|
float de_psy_factor = (sce->ics.num_windows > 1) ? 8.0f / sce->ics.group_len[w] : 1.0f; |
325 |
|
|
start = w*128; |
326 |
|
|
for (g = 0; g < sce->ics.num_swb; g++) { |
327 |
|
|
if (nzs[g] > 0) { |
328 |
|
|
float cleanup_factor = ff_sqrf(av_clipf(start / (cutoff * 0.75f), 1.0f, 2.0f)); |
329 |
|
|
float energy2uplim = find_form_factor( |
330 |
|
|
sce->ics.group_len[w], sce->ics.swb_sizes[g], |
331 |
|
|
uplims[w*16+g] / (nzs[g] * sce->ics.swb_sizes[w]), |
332 |
|
|
sce->coeffs + start, |
333 |
|
|
nzslope * cleanup_factor); |
334 |
|
|
energy2uplim *= de_psy_factor; |
335 |
|
|
if (!(avctx->flags & AV_CODEC_FLAG_QSCALE)) { |
336 |
|
|
/** In ABR, we need to priorize less and let rate control do its thing */ |
337 |
|
|
energy2uplim = sqrtf(energy2uplim); |
338 |
|
|
} |
339 |
|
|
energy2uplim = FFMAX(0.015625f, FFMIN(1.0f, energy2uplim)); |
340 |
|
|
uplims[w*16+g] *= av_clipf(rdlambda * energy2uplim, rdmin, rdmax) |
341 |
|
|
* sce->ics.group_len[w]; |
342 |
|
|
|
343 |
|
|
energy2uplim = find_form_factor( |
344 |
|
|
sce->ics.group_len[w], sce->ics.swb_sizes[g], |
345 |
|
|
uplims[w*16+g] / (nzs[g] * sce->ics.swb_sizes[w]), |
346 |
|
|
sce->coeffs + start, |
347 |
|
|
2.0f); |
348 |
|
|
energy2uplim *= de_psy_factor; |
349 |
|
|
if (!(avctx->flags & AV_CODEC_FLAG_QSCALE)) { |
350 |
|
|
/** In ABR, we need to priorize less and let rate control do its thing */ |
351 |
|
|
energy2uplim = sqrtf(energy2uplim); |
352 |
|
|
} |
353 |
|
|
energy2uplim = FFMAX(0.015625f, FFMIN(1.0f, energy2uplim)); |
354 |
|
|
euplims[w*16+g] *= av_clipf(rdlambda * energy2uplim * sce->ics.group_len[w], |
355 |
|
|
0.5f, 1.0f); |
356 |
|
|
} |
357 |
|
|
start += sce->ics.swb_sizes[g]; |
358 |
|
|
} |
359 |
|
|
} |
360 |
|
|
|
361 |
|
|
for (i = 0; i < sizeof(maxsf) / sizeof(maxsf[0]); ++i) |
362 |
|
|
maxsf[i] = SCALE_MAX_POS; |
363 |
|
|
|
364 |
|
|
//perform two-loop search |
365 |
|
|
//outer loop - improve quality |
366 |
|
|
do { |
367 |
|
|
//inner loop - quantize spectrum to fit into given number of bits |
368 |
|
|
int overdist; |
369 |
|
|
int qstep = its ? 1 : 32; |
370 |
|
|
do { |
371 |
|
|
int changed = 0; |
372 |
|
|
prev = -1; |
373 |
|
|
recomprd = 0; |
374 |
|
|
tbits = 0; |
375 |
|
|
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) { |
376 |
|
|
start = w*128; |
377 |
|
|
for (g = 0; g < sce->ics.num_swb; g++) { |
378 |
|
|
const float *coefs = &sce->coeffs[start]; |
379 |
|
|
const float *scaled = &s->scoefs[start]; |
380 |
|
|
int bits = 0; |
381 |
|
|
int cb; |
382 |
|
|
float dist = 0.0f; |
383 |
|
|
float qenergy = 0.0f; |
384 |
|
|
|
385 |
|
|
if (sce->zeroes[w*16+g] || sce->sf_idx[w*16+g] >= 218) { |
386 |
|
|
start += sce->ics.swb_sizes[g]; |
387 |
|
|
if (sce->can_pns[w*16+g]) { |
388 |
|
|
/** PNS isn't free */ |
389 |
|
|
tbits += ff_pns_bits(sce, w, g); |
390 |
|
|
} |
391 |
|
|
continue; |
392 |
|
|
} |
393 |
|
|
cb = find_min_book(maxvals[w*16+g], sce->sf_idx[w*16+g]); |
394 |
|
|
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) { |
395 |
|
|
int b; |
396 |
|
|
float sqenergy; |
397 |
|
|
dist += quantize_band_cost_cached(s, w + w2, g, coefs + w2*128, |
398 |
|
|
scaled + w2*128, |
399 |
|
|
sce->ics.swb_sizes[g], |
400 |
|
|
sce->sf_idx[w*16+g], |
401 |
|
|
cb, |
402 |
|
|
1.0f, |
403 |
|
|
INFINITY, |
404 |
|
|
&b, &sqenergy, |
405 |
|
|
0); |
406 |
|
|
bits += b; |
407 |
|
|
qenergy += sqenergy; |
408 |
|
|
} |
409 |
|
|
dists[w*16+g] = dist - bits; |
410 |
|
|
qenergies[w*16+g] = qenergy; |
411 |
|
|
if (prev != -1) { |
412 |
|
|
int sfdiff = av_clip(sce->sf_idx[w*16+g] - prev + SCALE_DIFF_ZERO, 0, 2*SCALE_MAX_DIFF); |
413 |
|
|
bits += ff_aac_scalefactor_bits[sfdiff]; |
414 |
|
|
} |
415 |
|
|
tbits += bits; |
416 |
|
|
start += sce->ics.swb_sizes[g]; |
417 |
|
|
prev = sce->sf_idx[w*16+g]; |
418 |
|
|
} |
419 |
|
|
} |
420 |
|
|
if (tbits > toomanybits) { |
421 |
|
|
recomprd = 1; |
422 |
|
|
for (i = 0; i < 128; i++) { |
423 |
|
|
if (sce->sf_idx[i] < (SCALE_MAX_POS - SCALE_DIV_512)) { |
424 |
|
|
int maxsf_i = (tbits > 5800) ? SCALE_MAX_POS : maxsf[i]; |
425 |
|
|
int new_sf = FFMIN(maxsf_i, sce->sf_idx[i] + qstep); |
426 |
|
|
if (new_sf != sce->sf_idx[i]) { |
427 |
|
|
sce->sf_idx[i] = new_sf; |
428 |
|
|
changed = 1; |
429 |
|
|
} |
430 |
|
|
} |
431 |
|
|
} |
432 |
|
|
} else if (tbits < toofewbits) { |
433 |
|
|
recomprd = 1; |
434 |
|
|
for (i = 0; i < 128; i++) { |
435 |
|
|
if (sce->sf_idx[i] > SCALE_ONE_POS) { |
436 |
|
|
int new_sf = FFMAX3(minsf[i], SCALE_ONE_POS, sce->sf_idx[i] - qstep); |
437 |
|
|
if (new_sf != sce->sf_idx[i]) { |
438 |
|
|
sce->sf_idx[i] = new_sf; |
439 |
|
|
changed = 1; |
440 |
|
|
} |
441 |
|
|
} |
442 |
|
|
} |
443 |
|
|
} |
444 |
|
|
qstep >>= 1; |
445 |
|
|
if (!qstep && tbits > toomanybits && sce->sf_idx[0] < 217 && changed) |
446 |
|
|
qstep = 1; |
447 |
|
|
} while (qstep); |
448 |
|
|
|
449 |
|
|
overdist = 1; |
450 |
|
|
fflag = tbits < toofewbits; |
451 |
|
|
for (i = 0; i < 2 && (overdist || recomprd); ++i) { |
452 |
|
|
if (recomprd) { |
453 |
|
|
/** Must recompute distortion */ |
454 |
|
|
prev = -1; |
455 |
|
|
tbits = 0; |
456 |
|
|
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) { |
457 |
|
|
start = w*128; |
458 |
|
|
for (g = 0; g < sce->ics.num_swb; g++) { |
459 |
|
|
const float *coefs = sce->coeffs + start; |
460 |
|
|
const float *scaled = s->scoefs + start; |
461 |
|
|
int bits = 0; |
462 |
|
|
int cb; |
463 |
|
|
float dist = 0.0f; |
464 |
|
|
float qenergy = 0.0f; |
465 |
|
|
|
466 |
|
|
if (sce->zeroes[w*16+g] || sce->sf_idx[w*16+g] >= 218) { |
467 |
|
|
start += sce->ics.swb_sizes[g]; |
468 |
|
|
if (sce->can_pns[w*16+g]) { |
469 |
|
|
/** PNS isn't free */ |
470 |
|
|
tbits += ff_pns_bits(sce, w, g); |
471 |
|
|
} |
472 |
|
|
continue; |
473 |
|
|
} |
474 |
|
|
cb = find_min_book(maxvals[w*16+g], sce->sf_idx[w*16+g]); |
475 |
|
|
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) { |
476 |
|
|
int b; |
477 |
|
|
float sqenergy; |
478 |
|
|
dist += quantize_band_cost_cached(s, w + w2, g, coefs + w2*128, |
479 |
|
|
scaled + w2*128, |
480 |
|
|
sce->ics.swb_sizes[g], |
481 |
|
|
sce->sf_idx[w*16+g], |
482 |
|
|
cb, |
483 |
|
|
1.0f, |
484 |
|
|
INFINITY, |
485 |
|
|
&b, &sqenergy, |
486 |
|
|
0); |
487 |
|
|
bits += b; |
488 |
|
|
qenergy += sqenergy; |
489 |
|
|
} |
490 |
|
|
dists[w*16+g] = dist - bits; |
491 |
|
|
qenergies[w*16+g] = qenergy; |
492 |
|
|
if (prev != -1) { |
493 |
|
|
int sfdiff = av_clip(sce->sf_idx[w*16+g] - prev + SCALE_DIFF_ZERO, 0, 2*SCALE_MAX_DIFF); |
494 |
|
|
bits += ff_aac_scalefactor_bits[sfdiff]; |
495 |
|
|
} |
496 |
|
|
tbits += bits; |
497 |
|
|
start += sce->ics.swb_sizes[g]; |
498 |
|
|
prev = sce->sf_idx[w*16+g]; |
499 |
|
|
} |
500 |
|
|
} |
501 |
|
|
} |
502 |
|
|
if (!i && s->options.pns && its > maxits/2 && tbits > toofewbits) { |
503 |
|
|
float maxoverdist = 0.0f; |
504 |
|
|
float ovrfactor = 1.f+(maxits-its)*16.f/maxits; |
505 |
|
|
overdist = recomprd = 0; |
506 |
|
|
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) { |
507 |
|
|
for (g = start = 0; g < sce->ics.num_swb; start += sce->ics.swb_sizes[g++]) { |
508 |
|
|
if (!sce->zeroes[w*16+g] && sce->sf_idx[w*16+g] > SCALE_ONE_POS && dists[w*16+g] > uplims[w*16+g]*ovrfactor) { |
509 |
|
|
float ovrdist = dists[w*16+g] / FFMAX(uplims[w*16+g],euplims[w*16+g]); |
510 |
|
|
maxoverdist = FFMAX(maxoverdist, ovrdist); |
511 |
|
|
overdist++; |
512 |
|
|
} |
513 |
|
|
} |
514 |
|
|
} |
515 |
|
|
if (overdist) { |
516 |
|
|
/* We have overdistorted bands, trade for zeroes (that can be noise) |
517 |
|
|
* Zero the bands in the lowest 1.25% spread-energy-threshold ranking |
518 |
|
|
*/ |
519 |
|
|
float minspread = max_spread_thr_r; |
520 |
|
|
float maxspread = min_spread_thr_r; |
521 |
|
|
float zspread; |
522 |
|
|
int zeroable = 0; |
523 |
|
|
int zeroed = 0; |
524 |
|
|
int maxzeroed, zloop; |
525 |
|
|
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) { |
526 |
|
|
for (g = start = 0; g < sce->ics.num_swb; start += sce->ics.swb_sizes[g++]) { |
527 |
|
|
if (start >= pns_start_pos && !sce->zeroes[w*16+g] && sce->can_pns[w*16+g]) { |
528 |
|
|
minspread = FFMIN(minspread, spread_thr_r[w*16+g]); |
529 |
|
|
maxspread = FFMAX(maxspread, spread_thr_r[w*16+g]); |
530 |
|
|
zeroable++; |
531 |
|
|
} |
532 |
|
|
} |
533 |
|
|
} |
534 |
|
|
zspread = (maxspread-minspread) * 0.0125f + minspread; |
535 |
|
|
/* Don't PNS everything even if allowed. It suppresses bit starvation signals from RC, |
536 |
|
|
* and forced the hand of the later search_for_pns step. |
537 |
|
|
* Instead, PNS a fraction of the spread_thr_r range depending on how starved for bits we are, |
538 |
|
|
* and leave further PNSing to search_for_pns if worthwhile. |
539 |
|
|
*/ |
540 |
|
|
zspread = FFMIN3(min_spread_thr_r * 8.f, zspread, |
541 |
|
|
((toomanybits - tbits) * min_spread_thr_r + (tbits - toofewbits) * max_spread_thr_r) / (toomanybits - toofewbits + 1)); |
542 |
|
|
maxzeroed = FFMIN(zeroable, FFMAX(1, (zeroable * its + maxits - 1) / (2 * maxits))); |
543 |
|
|
for (zloop = 0; zloop < 2; zloop++) { |
544 |
|
|
/* Two passes: first distorted stuff - two birds in one shot and all that, |
545 |
|
|
* then anything viable. Viable means not zero, but either CB=zero-able |
546 |
|
|
* (too high SF), not SF <= 1 (that means we'd be operating at very high |
547 |
|
|
* quality, we don't want PNS when doing VHQ), PNS allowed, and within |
548 |
|
|
* the lowest ranking percentile. |
549 |
|
|
*/ |
550 |
|
|
float loopovrfactor = (zloop) ? 1.0f : ovrfactor; |
551 |
|
|
int loopminsf = (zloop) ? (SCALE_ONE_POS - SCALE_DIV_512) : SCALE_ONE_POS; |
552 |
|
|
int mcb; |
553 |
|
|
for (g = sce->ics.num_swb-1; g > 0 && zeroed < maxzeroed; g--) { |
554 |
|
|
if (sce->ics.swb_offset[g] < pns_start_pos) |
555 |
|
|
continue; |
556 |
|
|
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) { |
557 |
|
|
if (!sce->zeroes[w*16+g] && sce->can_pns[w*16+g] && spread_thr_r[w*16+g] <= zspread |
558 |
|
|
&& sce->sf_idx[w*16+g] > loopminsf |
559 |
|
|
&& (dists[w*16+g] > loopovrfactor*uplims[w*16+g] || !(mcb = find_min_book(maxvals[w*16+g], sce->sf_idx[w*16+g])) |
560 |
|
|
|| (mcb <= 1 && dists[w*16+g] > FFMIN(uplims[w*16+g], euplims[w*16+g]))) ) { |
561 |
|
|
sce->zeroes[w*16+g] = 1; |
562 |
|
|
sce->band_type[w*16+g] = 0; |
563 |
|
|
zeroed++; |
564 |
|
|
} |
565 |
|
|
} |
566 |
|
|
} |
567 |
|
|
} |
568 |
|
|
if (zeroed) |
569 |
|
|
recomprd = fflag = 1; |
570 |
|
|
} else { |
571 |
|
|
overdist = 0; |
572 |
|
|
} |
573 |
|
|
} |
574 |
|
|
} |
575 |
|
|
|
576 |
|
|
minscaler = SCALE_MAX_POS; |
577 |
|
|
maxscaler = 0; |
578 |
|
|
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) { |
579 |
|
|
for (g = 0; g < sce->ics.num_swb; g++) { |
580 |
|
|
if (!sce->zeroes[w*16+g]) { |
581 |
|
|
minscaler = FFMIN(minscaler, sce->sf_idx[w*16+g]); |
582 |
|
|
maxscaler = FFMAX(maxscaler, sce->sf_idx[w*16+g]); |
583 |
|
|
} |
584 |
|
|
} |
585 |
|
|
} |
586 |
|
|
|
587 |
|
|
minscaler = nminscaler = av_clip(minscaler, SCALE_ONE_POS - SCALE_DIV_512, SCALE_MAX_POS - SCALE_DIV_512); |
588 |
|
|
prev = -1; |
589 |
|
|
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) { |
590 |
|
|
/** Start with big steps, end up fine-tunning */ |
591 |
|
|
int depth = (its > maxits/2) ? ((its > maxits*2/3) ? 1 : 3) : 10; |
592 |
|
|
int edepth = depth+2; |
593 |
|
|
float uplmax = its / (maxits*0.25f) + 1.0f; |
594 |
|
|
uplmax *= (tbits > destbits) ? FFMIN(2.0f, tbits / (float)FFMAX(1,destbits)) : 1.0f; |
595 |
|
|
start = w * 128; |
596 |
|
|
for (g = 0; g < sce->ics.num_swb; g++) { |
597 |
|
|
int prevsc = sce->sf_idx[w*16+g]; |
598 |
|
|
if (prev < 0 && !sce->zeroes[w*16+g]) |
599 |
|
|
prev = sce->sf_idx[0]; |
600 |
|
|
if (!sce->zeroes[w*16+g]) { |
601 |
|
|
const float *coefs = sce->coeffs + start; |
602 |
|
|
const float *scaled = s->scoefs + start; |
603 |
|
|
int cmb = find_min_book(maxvals[w*16+g], sce->sf_idx[w*16+g]); |
604 |
|
|
int mindeltasf = FFMAX(0, prev - SCALE_MAX_DIFF); |
605 |
|
|
int maxdeltasf = FFMIN(SCALE_MAX_POS - SCALE_DIV_512, prev + SCALE_MAX_DIFF); |
606 |
|
|
if ((!cmb || dists[w*16+g] > uplims[w*16+g]) && sce->sf_idx[w*16+g] > FFMAX(mindeltasf, minsf[w*16+g])) { |
607 |
|
|
/* Try to make sure there is some energy in every nonzero band |
608 |
|
|
* NOTE: This algorithm must be forcibly imbalanced, pushing harder |
609 |
|
|
* on holes or more distorted bands at first, otherwise there's |
610 |
|
|
* no net gain (since the next iteration will offset all bands |
611 |
|
|
* on the opposite direction to compensate for extra bits) |
612 |
|
|
*/ |
613 |
|
|
for (i = 0; i < edepth && sce->sf_idx[w*16+g] > mindeltasf; ++i) { |
614 |
|
|
int cb, bits; |
615 |
|
|
float dist, qenergy; |
616 |
|
|
int mb = find_min_book(maxvals[w*16+g], sce->sf_idx[w*16+g]-1); |
617 |
|
|
cb = find_min_book(maxvals[w*16+g], sce->sf_idx[w*16+g]); |
618 |
|
|
dist = qenergy = 0.f; |
619 |
|
|
bits = 0; |
620 |
|
|
if (!cb) { |
621 |
|
|
maxsf[w*16+g] = FFMIN(sce->sf_idx[w*16+g]-1, maxsf[w*16+g]); |
622 |
|
|
} else if (i >= depth && dists[w*16+g] < euplims[w*16+g]) { |
623 |
|
|
break; |
624 |
|
|
} |
625 |
|
|
/* !g is the DC band, it's important, since quantization error here |
626 |
|
|
* applies to less than a cycle, it creates horrible intermodulation |
627 |
|
|
* distortion if it doesn't stick to what psy requests |
628 |
|
|
*/ |
629 |
|
|
if (!g && sce->ics.num_windows > 1 && dists[w*16+g] >= euplims[w*16+g]) |
630 |
|
|
maxsf[w*16+g] = FFMIN(sce->sf_idx[w*16+g], maxsf[w*16+g]); |
631 |
|
|
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) { |
632 |
|
|
int b; |
633 |
|
|
float sqenergy; |
634 |
|
|
dist += quantize_band_cost_cached(s, w + w2, g, coefs + w2*128, |
635 |
|
|
scaled + w2*128, |
636 |
|
|
sce->ics.swb_sizes[g], |
637 |
|
|
sce->sf_idx[w*16+g]-1, |
638 |
|
|
cb, |
639 |
|
|
1.0f, |
640 |
|
|
INFINITY, |
641 |
|
|
&b, &sqenergy, |
642 |
|
|
0); |
643 |
|
|
bits += b; |
644 |
|
|
qenergy += sqenergy; |
645 |
|
|
} |
646 |
|
|
sce->sf_idx[w*16+g]--; |
647 |
|
|
dists[w*16+g] = dist - bits; |
648 |
|
|
qenergies[w*16+g] = qenergy; |
649 |
|
|
if (mb && (sce->sf_idx[w*16+g] < mindeltasf || ( |
650 |
|
|
(dists[w*16+g] < FFMIN(uplmax*uplims[w*16+g], euplims[w*16+g])) |
651 |
|
|
&& (fabsf(qenergies[w*16+g]-energies[w*16+g]) < euplims[w*16+g]) |
652 |
|
|
) )) { |
653 |
|
|
break; |
654 |
|
|
} |
655 |
|
|
} |
656 |
|
|
} else if (tbits > toofewbits && sce->sf_idx[w*16+g] < FFMIN(maxdeltasf, maxsf[w*16+g]) |
657 |
|
|
&& (dists[w*16+g] < FFMIN(euplims[w*16+g], uplims[w*16+g])) |
658 |
|
|
&& (fabsf(qenergies[w*16+g]-energies[w*16+g]) < euplims[w*16+g]) |
659 |
|
|
) { |
660 |
|
|
/** Um... over target. Save bits for more important stuff. */ |
661 |
|
|
for (i = 0; i < depth && sce->sf_idx[w*16+g] < maxdeltasf; ++i) { |
662 |
|
|
int cb, bits; |
663 |
|
|
float dist, qenergy; |
664 |
|
|
cb = find_min_book(maxvals[w*16+g], sce->sf_idx[w*16+g]+1); |
665 |
|
|
if (cb > 0) { |
666 |
|
|
dist = qenergy = 0.f; |
667 |
|
|
bits = 0; |
668 |
|
|
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) { |
669 |
|
|
int b; |
670 |
|
|
float sqenergy; |
671 |
|
|
dist += quantize_band_cost_cached(s, w + w2, g, coefs + w2*128, |
672 |
|
|
scaled + w2*128, |
673 |
|
|
sce->ics.swb_sizes[g], |
674 |
|
|
sce->sf_idx[w*16+g]+1, |
675 |
|
|
cb, |
676 |
|
|
1.0f, |
677 |
|
|
INFINITY, |
678 |
|
|
&b, &sqenergy, |
679 |
|
|
0); |
680 |
|
|
bits += b; |
681 |
|
|
qenergy += sqenergy; |
682 |
|
|
} |
683 |
|
|
dist -= bits; |
684 |
|
|
if (dist < FFMIN(euplims[w*16+g], uplims[w*16+g])) { |
685 |
|
|
sce->sf_idx[w*16+g]++; |
686 |
|
|
dists[w*16+g] = dist; |
687 |
|
|
qenergies[w*16+g] = qenergy; |
688 |
|
|
} else { |
689 |
|
|
break; |
690 |
|
|
} |
691 |
|
|
} else { |
692 |
|
|
maxsf[w*16+g] = FFMIN(sce->sf_idx[w*16+g], maxsf[w*16+g]); |
693 |
|
|
break; |
694 |
|
|
} |
695 |
|
|
} |
696 |
|
|
} |
697 |
|
|
prev = sce->sf_idx[w*16+g] = av_clip(sce->sf_idx[w*16+g], mindeltasf, maxdeltasf); |
698 |
|
|
if (sce->sf_idx[w*16+g] != prevsc) |
699 |
|
|
fflag = 1; |
700 |
|
|
nminscaler = FFMIN(nminscaler, sce->sf_idx[w*16+g]); |
701 |
|
|
sce->band_type[w*16+g] = find_min_book(maxvals[w*16+g], sce->sf_idx[w*16+g]); |
702 |
|
|
} |
703 |
|
|
start += sce->ics.swb_sizes[g]; |
704 |
|
|
} |
705 |
|
|
} |
706 |
|
|
|
707 |
|
|
/** SF difference limit violation risk. Must re-clamp. */ |
708 |
|
|
prev = -1; |
709 |
|
|
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) { |
710 |
|
|
for (g = 0; g < sce->ics.num_swb; g++) { |
711 |
|
|
if (!sce->zeroes[w*16+g]) { |
712 |
|
|
int prevsf = sce->sf_idx[w*16+g]; |
713 |
|
|
if (prev < 0) |
714 |
|
|
prev = prevsf; |
715 |
|
|
sce->sf_idx[w*16+g] = av_clip(sce->sf_idx[w*16+g], prev - SCALE_MAX_DIFF, prev + SCALE_MAX_DIFF); |
716 |
|
|
sce->band_type[w*16+g] = find_min_book(maxvals[w*16+g], sce->sf_idx[w*16+g]); |
717 |
|
|
prev = sce->sf_idx[w*16+g]; |
718 |
|
|
if (!fflag && prevsf != sce->sf_idx[w*16+g]) |
719 |
|
|
fflag = 1; |
720 |
|
|
} |
721 |
|
|
} |
722 |
|
|
} |
723 |
|
|
|
724 |
|
|
its++; |
725 |
|
|
} while (fflag && its < maxits); |
726 |
|
|
|
727 |
|
|
/** Scout out next nonzero bands */ |
728 |
|
|
ff_init_nextband_map(sce, nextband); |
729 |
|
|
|
730 |
|
|
prev = -1; |
731 |
|
|
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) { |
732 |
|
|
/** Make sure proper codebooks are set */ |
733 |
|
|
for (g = 0; g < sce->ics.num_swb; g++) { |
734 |
|
|
if (!sce->zeroes[w*16+g]) { |
735 |
|
|
sce->band_type[w*16+g] = find_min_book(maxvals[w*16+g], sce->sf_idx[w*16+g]); |
736 |
|
|
if (sce->band_type[w*16+g] <= 0) { |
737 |
|
|
if (!ff_sfdelta_can_remove_band(sce, nextband, prev, w*16+g)) { |
738 |
|
|
/** Cannot zero out, make sure it's not attempted */ |
739 |
|
|
sce->band_type[w*16+g] = 1; |
740 |
|
|
} else { |
741 |
|
|
sce->zeroes[w*16+g] = 1; |
742 |
|
|
sce->band_type[w*16+g] = 0; |
743 |
|
|
} |
744 |
|
|
} |
745 |
|
|
} else { |
746 |
|
|
sce->band_type[w*16+g] = 0; |
747 |
|
|
} |
748 |
|
|
/** Check that there's no SF delta range violations */ |
749 |
|
|
if (!sce->zeroes[w*16+g]) { |
750 |
|
|
if (prev != -1) { |
751 |
|
|
av_unused int sfdiff = sce->sf_idx[w*16+g] - prev + SCALE_DIFF_ZERO; |
752 |
|
|
av_assert1(sfdiff >= 0 && sfdiff <= 2*SCALE_MAX_DIFF); |
753 |
|
|
} else if (sce->zeroes[0]) { |
754 |
|
|
/** Set global gain to something useful */ |
755 |
|
|
sce->sf_idx[0] = sce->sf_idx[w*16+g]; |
756 |
|
|
} |
757 |
|
|
prev = sce->sf_idx[w*16+g]; |
758 |
|
|
} |
759 |
|
|
} |
760 |
|
|
} |
761 |
|
|
} |
762 |
|
|
|
763 |
|
|
#endif /* AVCODEC_AACCODER_TWOLOOP_H */ |