ffmpeg/libavcodec/aaccoder.c

852 lines
35 KiB
C
Raw Normal View History

/*
* AAC coefficients encoder
* Copyright (C) 2008-2009 Konstantin Shishkov
*
* This file is part of FFmpeg.
*
* FFmpeg 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,
* 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
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
/**
* @file
* AAC coefficients encoder
*/
/***********************************
* TODOs:
* speedup quantizer selection
* add sane pulse detection
***********************************/
#include "libavutil/libm.h" // brought forward to work around cygwin header breakage
#include <float.h>
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
#include "libavutil/mathematics.h"
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
#include "mathops.h"
#include "avcodec.h"
#include "put_bits.h"
#include "aac.h"
#include "aacenc.h"
#include "aactab.h"
#include "aacenctab.h"
#include "aacenc_utils.h"
#include "aacenc_quantization.h"
#include "aacenc_is.h"
#include "aacenc_tns.h"
#include "aacenc_ltp.h"
#include "aacenc_pred.h"
#include "libavcodec/aaccoder_twoloop.h"
/* Parameter of f(x) = a*(lambda/100), defines the maximum fourier spread
* beyond which no PNS is used (since the SFBs contain tone rather than noise) */
AAC encoder: improve SF range utilization This patch does 4 things, all of which interact and thus it woudln't be possible to commit them separately without causing either quality regressions or assertion failures. Fate comparison targets don't all reflect improvements in quality, yet listening tests show substantially improved quality and stability. 1. Increase SF range utilization. The spec requires SF delta values to be constrained within the range -60..60. The previous code was applying that range to the whole SF array and not only the deltas of consecutive values, because doing so requires smarter code: zeroing or otherwise skipping a band may invalidate lots of SF choices. This patch implements that logic to allow the coders to utilize the full dynamic range of scalefactors, increasing quality quite considerably, and fixing delta-SF-related assertion failures, since now the limitation is enforced rather than asserted. 2. PNS tweaks The previous modification makes big improvements in twoloop's efficiency, and every time that happens PNS logic needs to be tweaked accordingly to avoid it from stepping all over twoloop's decisions. This patch includes modifications of the sort. 3. Account for lowpass cutoff during PSY analysis The closer PSY's allocation is to final allocation the better the quality is, and given these modifications, twoloop is now very efficient at avoiding holes. Thus, to compute accurate thresholds, PSY needs to account for the lowpass applied implicitly during twoloop (by zeroing high bands). This patch makes twoloop set the cutoff in psymodel's context the first time it runs, and makes PSY account for it during threshold computation, making PE and threshold computations closer to the final allocation and thus achieving better subjective quality. 4. Tweaks to RC lambda tracking loop in relation to PNS Without this tweak some corner cases cause quality regressions. Basically, lambda needs to react faster to overall bitrate efficiency changes since now PNS can be quite successful in enforcing maximum bitrates, when PSY allocates too many bits to the lower bands, suppressing the signals RC logic uses to lower lambda in those cases and causing aggressive PNS. This tweak makes PNS much less aggressive, though it can still use some further tweaks. Also update MIPS specializations and adjust fuzz Also in lavc/mips/aacpsy_mips.h: remove trailing whitespace
2015-12-01 06:28:36 +00:00
#define NOISE_SPREAD_THRESHOLD 0.9f
/* Parameter of f(x) = a*(100/lambda), defines how much PNS is allowed to
* replace low energy non zero bands */
#define NOISE_LAMBDA_REPLACE 1.948f
aaccoder: Implement Perceptual Noise Substitution for AAC This commit implements the perceptual noise substitution AAC extension. This is a proof of concept implementation, and as such, is not enabled by default. This is the fourth revision of this patch, made after some problems were noted out. Any changes made since the previous revisions have been indicated. In order to extend the encoder to use an additional codebook, the array holding each codebook has been modified with two additional entries - 13 for the NOISE_BT codebook and 12 which has a placeholder function. The cost system was modified to skip the 12th entry using an array to map the input and outputs it has. It also does not accept using the 13th codebook for any band which is not marked as containing noise, thereby restricting its ability to arbitrarily choose it for bands. The use of arrays allows the system to be easily extended to allow for intensity stereo encoding, which uses additional codebooks. The 12th entry in the codebook function array points to a function which stops the execution of the program by calling an assert with an always 'false' argument. It was pointed out in an email discussion with Claudio Freire that having a 'NULL' entry can result in unexpected behaviour and could be used as a security hole. There is no danger of this function being called during encoding due to the codebook maps introduced. Another change from version 1 of the patch is the addition of an argument to the encoder, '-aac_pns' to enable and disable the PNS. This currently defaults to disable the PNS, as it is experimental. The switch will be removed in the future, when the algorithm to select noise bands has been improved. The current algorithm simply compares the energy to the threshold (multiplied by a constant) to determine noise, however the FFPsyBand structure contains other useful figures to determine which bands carry noise more accurately. Some of the sample files provided triggered an assertion when the parameter to tune the threshold was set to a value of '2.2'. Claudio Freire reported the problem's source could be in the range of the scalefactor indices for noise and advised to measure the minimal index and clip anything above the maximum allowed value. This has been implemented and all the files which used to trigger the asserion now encode without error. The third revision of the problem also removes unneded variabes and comparisons. All of them were redundant and were of little use for when the PNS implementation would be extended. The fourth revision moved the clipping of the noise scalefactors outside the second loop of the two-loop algorithm in order to prevent their redundant calculations. Also, freq_mult has been changed to a float variable due to the fact that rounding errors can prove to be a problem at low frequencies. Considerations were taken whether the entire expression could be evaluated inside the expression , but in the end it was decided that it would be for the best if just the type of the variable were to change. Claudio Freire reported the two problems. There is no change of functionality (except for low sampling frequencies) so the spectral demonstrations at the end of this commit's message were not updated. Finally, the way energy values are converted to scalefactor indices has changed since the first commit, as per the suggestion of Claudio Freire. This may still have some drawbacks, but unlike the first commit it works without having redundant offsets and outputs what the decoder expects to have, in terms of the ranges of the scalefactor indices. Some spectral comparisons: https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/Original.png (original), https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/PNS_NO.png (encoded without PNS), https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/PNS1.2.png (encoded with PNS, const = 1.2), https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/Difference1.png (spectral difference). The constant is the value which multiplies the threshold when it gets compared to the energy, larger values means more noise will be substituded by PNS values. Example when const = 2.2: https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/PNS_2.2.png Reviewed-by: Claudio Freire <klaussfreire@gmail.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-04-15 11:18:42 +00:00
#include "libavcodec/aaccoder_trellis.h"
/**
* structure used in optimal codebook search
*/
typedef struct BandCodingPath {
int prev_idx; ///< pointer to the previous path point
float cost; ///< path cost
int run;
} BandCodingPath;
/**
* Encode band info for single window group bands.
*/
static void encode_window_bands_info(AACEncContext *s, SingleChannelElement *sce,
int win, int group_len, const float lambda)
{
BandCodingPath path[120][CB_TOT_ALL];
2011-06-01 16:26:27 +00:00
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;
int next_mincb = 0;
abs_pow34_v(s->scoefs, sce->coeffs, 1024);
start = win*128;
for (cb = 0; cb < CB_TOT_ALL; cb++) {
path[0][cb].cost = 0.0f;
path[0][cb].prev_idx = -1;
path[0][cb].run = 0;
}
for (swb = 0; swb < max_sfb; swb++) {
size = sce->ics.swb_sizes[swb];
if (sce->zeroes[win*16 + swb]) {
for (cb = 0; cb < CB_TOT_ALL; 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;
}
} else {
float minrd = next_minrd;
int mincb = next_mincb;
next_minrd = INFINITY;
next_mincb = 0;
for (cb = 0; cb < CB_TOT_ALL; cb++) {
float cost_stay_here, cost_get_here;
float rd = 0.0f;
if (cb >= 12 && sce->band_type[win*16+swb] < aac_cb_out_map[cb] ||
cb < aac_cb_in_map[sce->band_type[win*16+swb]] && sce->band_type[win*16+swb] > aac_cb_out_map[cb]) {
path[swb+1][cb].prev_idx = -1;
path[swb+1][cb].cost = INFINITY;
path[swb+1][cb].run = path[swb][cb].run + 1;
continue;
}
for (w = 0; w < group_len; w++) {
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,
aaccoder: Implement Perceptual Noise Substitution for AAC This commit implements the perceptual noise substitution AAC extension. This is a proof of concept implementation, and as such, is not enabled by default. This is the fourth revision of this patch, made after some problems were noted out. Any changes made since the previous revisions have been indicated. In order to extend the encoder to use an additional codebook, the array holding each codebook has been modified with two additional entries - 13 for the NOISE_BT codebook and 12 which has a placeholder function. The cost system was modified to skip the 12th entry using an array to map the input and outputs it has. It also does not accept using the 13th codebook for any band which is not marked as containing noise, thereby restricting its ability to arbitrarily choose it for bands. The use of arrays allows the system to be easily extended to allow for intensity stereo encoding, which uses additional codebooks. The 12th entry in the codebook function array points to a function which stops the execution of the program by calling an assert with an always 'false' argument. It was pointed out in an email discussion with Claudio Freire that having a 'NULL' entry can result in unexpected behaviour and could be used as a security hole. There is no danger of this function being called during encoding due to the codebook maps introduced. Another change from version 1 of the patch is the addition of an argument to the encoder, '-aac_pns' to enable and disable the PNS. This currently defaults to disable the PNS, as it is experimental. The switch will be removed in the future, when the algorithm to select noise bands has been improved. The current algorithm simply compares the energy to the threshold (multiplied by a constant) to determine noise, however the FFPsyBand structure contains other useful figures to determine which bands carry noise more accurately. Some of the sample files provided triggered an assertion when the parameter to tune the threshold was set to a value of '2.2'. Claudio Freire reported the problem's source could be in the range of the scalefactor indices for noise and advised to measure the minimal index and clip anything above the maximum allowed value. This has been implemented and all the files which used to trigger the asserion now encode without error. The third revision of the problem also removes unneded variabes and comparisons. All of them were redundant and were of little use for when the PNS implementation would be extended. The fourth revision moved the clipping of the noise scalefactors outside the second loop of the two-loop algorithm in order to prevent their redundant calculations. Also, freq_mult has been changed to a float variable due to the fact that rounding errors can prove to be a problem at low frequencies. Considerations were taken whether the entire expression could be evaluated inside the expression , but in the end it was decided that it would be for the best if just the type of the variable were to change. Claudio Freire reported the two problems. There is no change of functionality (except for low sampling frequencies) so the spectral demonstrations at the end of this commit's message were not updated. Finally, the way energy values are converted to scalefactor indices has changed since the first commit, as per the suggestion of Claudio Freire. This may still have some drawbacks, but unlike the first commit it works without having redundant offsets and outputs what the decoder expects to have, in terms of the ranges of the scalefactor indices. Some spectral comparisons: https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/Original.png (original), https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/PNS_NO.png (encoded without PNS), https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/PNS1.2.png (encoded with PNS, const = 1.2), https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/Difference1.png (spectral difference). The constant is the value which multiplies the threshold when it gets compared to the energy, larger values means more noise will be substituded by PNS values. Example when const = 2.2: https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/PNS_2.2.png Reviewed-by: Claudio Freire <klaussfreire@gmail.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-04-15 11:18:42 +00:00
sce->sf_idx[(win+w)*16+swb], aac_cb_out_map[cb],
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
lambda / band->threshold, INFINITY, NULL, NULL, 0);
}
cost_stay_here = path[swb][cb].cost + rd;
cost_get_here = minrd + rd + 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;
if (cost_get_here < cost_stay_here) {
path[swb+1][cb].prev_idx = mincb;
path[swb+1][cb].cost = cost_get_here;
path[swb+1][cb].run = 1;
} else {
path[swb+1][cb].prev_idx = cb;
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;
next_mincb = cb;
}
}
}
start += sce->ics.swb_sizes[swb];
}
//convert resulting path from backward-linked list
stack_len = 0;
idx = 0;
for (cb = 1; cb < CB_TOT_ALL; cb++)
if (path[max_sfb][cb].cost < path[max_sfb][idx].cost)
idx = cb;
ppos = max_sfb;
while (ppos > 0) {
av_assert1(idx >= 0);
cb = idx;
stackrun[stack_len] = path[ppos][cb].run;
stackcb [stack_len] = cb;
idx = path[ppos-path[ppos][cb].run+1][cb].prev_idx;
ppos -= path[ppos][cb].run;
stack_len++;
}
//perform actual band info encoding
start = 0;
for (i = stack_len - 1; i >= 0; i--) {
aaccoder: Implement Perceptual Noise Substitution for AAC This commit implements the perceptual noise substitution AAC extension. This is a proof of concept implementation, and as such, is not enabled by default. This is the fourth revision of this patch, made after some problems were noted out. Any changes made since the previous revisions have been indicated. In order to extend the encoder to use an additional codebook, the array holding each codebook has been modified with two additional entries - 13 for the NOISE_BT codebook and 12 which has a placeholder function. The cost system was modified to skip the 12th entry using an array to map the input and outputs it has. It also does not accept using the 13th codebook for any band which is not marked as containing noise, thereby restricting its ability to arbitrarily choose it for bands. The use of arrays allows the system to be easily extended to allow for intensity stereo encoding, which uses additional codebooks. The 12th entry in the codebook function array points to a function which stops the execution of the program by calling an assert with an always 'false' argument. It was pointed out in an email discussion with Claudio Freire that having a 'NULL' entry can result in unexpected behaviour and could be used as a security hole. There is no danger of this function being called during encoding due to the codebook maps introduced. Another change from version 1 of the patch is the addition of an argument to the encoder, '-aac_pns' to enable and disable the PNS. This currently defaults to disable the PNS, as it is experimental. The switch will be removed in the future, when the algorithm to select noise bands has been improved. The current algorithm simply compares the energy to the threshold (multiplied by a constant) to determine noise, however the FFPsyBand structure contains other useful figures to determine which bands carry noise more accurately. Some of the sample files provided triggered an assertion when the parameter to tune the threshold was set to a value of '2.2'. Claudio Freire reported the problem's source could be in the range of the scalefactor indices for noise and advised to measure the minimal index and clip anything above the maximum allowed value. This has been implemented and all the files which used to trigger the asserion now encode without error. The third revision of the problem also removes unneded variabes and comparisons. All of them were redundant and were of little use for when the PNS implementation would be extended. The fourth revision moved the clipping of the noise scalefactors outside the second loop of the two-loop algorithm in order to prevent their redundant calculations. Also, freq_mult has been changed to a float variable due to the fact that rounding errors can prove to be a problem at low frequencies. Considerations were taken whether the entire expression could be evaluated inside the expression , but in the end it was decided that it would be for the best if just the type of the variable were to change. Claudio Freire reported the two problems. There is no change of functionality (except for low sampling frequencies) so the spectral demonstrations at the end of this commit's message were not updated. Finally, the way energy values are converted to scalefactor indices has changed since the first commit, as per the suggestion of Claudio Freire. This may still have some drawbacks, but unlike the first commit it works without having redundant offsets and outputs what the decoder expects to have, in terms of the ranges of the scalefactor indices. Some spectral comparisons: https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/Original.png (original), https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/PNS_NO.png (encoded without PNS), https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/PNS1.2.png (encoded with PNS, const = 1.2), https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/Difference1.png (spectral difference). The constant is the value which multiplies the threshold when it gets compared to the energy, larger values means more noise will be substituded by PNS values. Example when const = 2.2: https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/PNS_2.2.png Reviewed-by: Claudio Freire <klaussfreire@gmail.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-04-15 11:18:42 +00:00
cb = aac_cb_out_map[stackcb[i]];
put_bits(&s->pb, 4, cb);
count = stackrun[i];
aaccoder: Implement Perceptual Noise Substitution for AAC This commit implements the perceptual noise substitution AAC extension. This is a proof of concept implementation, and as such, is not enabled by default. This is the fourth revision of this patch, made after some problems were noted out. Any changes made since the previous revisions have been indicated. In order to extend the encoder to use an additional codebook, the array holding each codebook has been modified with two additional entries - 13 for the NOISE_BT codebook and 12 which has a placeholder function. The cost system was modified to skip the 12th entry using an array to map the input and outputs it has. It also does not accept using the 13th codebook for any band which is not marked as containing noise, thereby restricting its ability to arbitrarily choose it for bands. The use of arrays allows the system to be easily extended to allow for intensity stereo encoding, which uses additional codebooks. The 12th entry in the codebook function array points to a function which stops the execution of the program by calling an assert with an always 'false' argument. It was pointed out in an email discussion with Claudio Freire that having a 'NULL' entry can result in unexpected behaviour and could be used as a security hole. There is no danger of this function being called during encoding due to the codebook maps introduced. Another change from version 1 of the patch is the addition of an argument to the encoder, '-aac_pns' to enable and disable the PNS. This currently defaults to disable the PNS, as it is experimental. The switch will be removed in the future, when the algorithm to select noise bands has been improved. The current algorithm simply compares the energy to the threshold (multiplied by a constant) to determine noise, however the FFPsyBand structure contains other useful figures to determine which bands carry noise more accurately. Some of the sample files provided triggered an assertion when the parameter to tune the threshold was set to a value of '2.2'. Claudio Freire reported the problem's source could be in the range of the scalefactor indices for noise and advised to measure the minimal index and clip anything above the maximum allowed value. This has been implemented and all the files which used to trigger the asserion now encode without error. The third revision of the problem also removes unneded variabes and comparisons. All of them were redundant and were of little use for when the PNS implementation would be extended. The fourth revision moved the clipping of the noise scalefactors outside the second loop of the two-loop algorithm in order to prevent their redundant calculations. Also, freq_mult has been changed to a float variable due to the fact that rounding errors can prove to be a problem at low frequencies. Considerations were taken whether the entire expression could be evaluated inside the expression , but in the end it was decided that it would be for the best if just the type of the variable were to change. Claudio Freire reported the two problems. There is no change of functionality (except for low sampling frequencies) so the spectral demonstrations at the end of this commit's message were not updated. Finally, the way energy values are converted to scalefactor indices has changed since the first commit, as per the suggestion of Claudio Freire. This may still have some drawbacks, but unlike the first commit it works without having redundant offsets and outputs what the decoder expects to have, in terms of the ranges of the scalefactor indices. Some spectral comparisons: https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/Original.png (original), https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/PNS_NO.png (encoded without PNS), https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/PNS1.2.png (encoded with PNS, const = 1.2), https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/Difference1.png (spectral difference). The constant is the value which multiplies the threshold when it gets compared to the energy, larger values means more noise will be substituded by PNS values. Example when const = 2.2: https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/PNS_2.2.png Reviewed-by: Claudio Freire <klaussfreire@gmail.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-04-15 11:18:42 +00:00
memset(sce->zeroes + win*16 + start, !cb, count);
//XXX: memset when band_type is also uint8_t
for (j = 0; j < count; j++) {
aaccoder: Implement Perceptual Noise Substitution for AAC This commit implements the perceptual noise substitution AAC extension. This is a proof of concept implementation, and as such, is not enabled by default. This is the fourth revision of this patch, made after some problems were noted out. Any changes made since the previous revisions have been indicated. In order to extend the encoder to use an additional codebook, the array holding each codebook has been modified with two additional entries - 13 for the NOISE_BT codebook and 12 which has a placeholder function. The cost system was modified to skip the 12th entry using an array to map the input and outputs it has. It also does not accept using the 13th codebook for any band which is not marked as containing noise, thereby restricting its ability to arbitrarily choose it for bands. The use of arrays allows the system to be easily extended to allow for intensity stereo encoding, which uses additional codebooks. The 12th entry in the codebook function array points to a function which stops the execution of the program by calling an assert with an always 'false' argument. It was pointed out in an email discussion with Claudio Freire that having a 'NULL' entry can result in unexpected behaviour and could be used as a security hole. There is no danger of this function being called during encoding due to the codebook maps introduced. Another change from version 1 of the patch is the addition of an argument to the encoder, '-aac_pns' to enable and disable the PNS. This currently defaults to disable the PNS, as it is experimental. The switch will be removed in the future, when the algorithm to select noise bands has been improved. The current algorithm simply compares the energy to the threshold (multiplied by a constant) to determine noise, however the FFPsyBand structure contains other useful figures to determine which bands carry noise more accurately. Some of the sample files provided triggered an assertion when the parameter to tune the threshold was set to a value of '2.2'. Claudio Freire reported the problem's source could be in the range of the scalefactor indices for noise and advised to measure the minimal index and clip anything above the maximum allowed value. This has been implemented and all the files which used to trigger the asserion now encode without error. The third revision of the problem also removes unneded variabes and comparisons. All of them were redundant and were of little use for when the PNS implementation would be extended. The fourth revision moved the clipping of the noise scalefactors outside the second loop of the two-loop algorithm in order to prevent their redundant calculations. Also, freq_mult has been changed to a float variable due to the fact that rounding errors can prove to be a problem at low frequencies. Considerations were taken whether the entire expression could be evaluated inside the expression , but in the end it was decided that it would be for the best if just the type of the variable were to change. Claudio Freire reported the two problems. There is no change of functionality (except for low sampling frequencies) so the spectral demonstrations at the end of this commit's message were not updated. Finally, the way energy values are converted to scalefactor indices has changed since the first commit, as per the suggestion of Claudio Freire. This may still have some drawbacks, but unlike the first commit it works without having redundant offsets and outputs what the decoder expects to have, in terms of the ranges of the scalefactor indices. Some spectral comparisons: https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/Original.png (original), https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/PNS_NO.png (encoded without PNS), https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/PNS1.2.png (encoded with PNS, const = 1.2), https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/Difference1.png (spectral difference). The constant is the value which multiplies the threshold when it gets compared to the energy, larger values means more noise will be substituded by PNS values. Example when const = 2.2: https://trac.ffmpeg.org/attachment/wiki/Encode/AAC/PNS_2.2.png Reviewed-by: Claudio Freire <klaussfreire@gmail.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-04-15 11:18:42 +00:00
sce->band_type[win*16 + start] = cb;
start++;
}
while (count >= run_esc) {
put_bits(&s->pb, run_bits, run_esc);
count -= run_esc;
}
put_bits(&s->pb, run_bits, count);
}
}
typedef struct TrellisPath {
float cost;
int prev;
} TrellisPath;
#define TRELLIS_STAGES 121
#define TRELLIS_STATES (SCALE_MAX_DIFF+1)
static void set_special_band_scalefactors(AACEncContext *s, SingleChannelElement *sce)
{
int w, g;
int prevscaler_n = -255, prevscaler_i = 0;
int bands = 0;
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
for (g = 0; g < sce->ics.num_swb; g++) {
if (sce->zeroes[w*16+g])
continue;
if (sce->band_type[w*16+g] == INTENSITY_BT || sce->band_type[w*16+g] == INTENSITY_BT2) {
sce->sf_idx[w*16+g] = av_clip(roundf(log2f(sce->is_ener[w*16+g])*2), -155, 100);
bands++;
} else if (sce->band_type[w*16+g] == NOISE_BT) {
sce->sf_idx[w*16+g] = av_clip(3+ceilf(log2f(sce->pns_ener[w*16+g])*2), -100, 155);
if (prevscaler_n == -255)
prevscaler_n = sce->sf_idx[w*16+g];
bands++;
}
}
}
if (!bands)
return;
/* Clip the scalefactor indices */
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
for (g = 0; g < sce->ics.num_swb; g++) {
if (sce->zeroes[w*16+g])
continue;
if (sce->band_type[w*16+g] == INTENSITY_BT || sce->band_type[w*16+g] == INTENSITY_BT2) {
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);
} else if (sce->band_type[w*16+g] == NOISE_BT) {
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);
}
}
}
}
static void search_for_quantizers_anmr(AVCodecContext *avctx, AACEncContext *s,
SingleChannelElement *sce,
const float lambda)
{
int q, w, w2, g, start = 0;
int i, j;
int idx;
TrellisPath paths[TRELLIS_STAGES][TRELLIS_STATES];
int bandaddr[TRELLIS_STAGES];
int minq;
float mincost;
float q0f = FLT_MAX, q1f = 0.0f, qnrgf = 0.0f;
int q0, q1, qcnt = 0;
for (i = 0; i < 1024; i++) {
float t = fabsf(sce->coeffs[i]);
if (t > 0.0f) {
q0f = FFMIN(q0f, t);
q1f = FFMAX(q1f, t);
qnrgf += t*t;
qcnt++;
}
}
if (!qcnt) {
memset(sce->sf_idx, 0, sizeof(sce->sf_idx));
memset(sce->zeroes, 1, sizeof(sce->zeroes));
return;
}
//minimum scalefactor index is when minimum nonzero coefficient after quantizing is not clipped
q0 = av_clip(coef2minsf(q0f), 0, SCALE_MAX_POS-1);
//maximum scalefactor index is when maximum coefficient after quantizing is still not zero
q1 = av_clip(coef2maxsf(q1f), 1, SCALE_MAX_POS);
if (q1 - q0 > 60) {
int q0low = q0;
int q1high = q1;
//minimum scalefactor index is when maximum nonzero coefficient after quantizing is not clipped
int qnrg = av_clip_uint8(log2f(sqrtf(qnrgf/qcnt))*4 - 31 + SCALE_ONE_POS - SCALE_DIV_512);
q1 = qnrg + 30;
q0 = qnrg - 30;
if (q0 < q0low) {
q1 += q0low - q0;
q0 = q0low;
} else if (q1 > q1high) {
q0 -= q1 - q1high;
q1 = q1high;
}
}
// q0 == q1 isn't really a legal situation
if (q0 == q1) {
// the following is indirect but guarantees q1 != q0 && q1 near q0
q1 = av_clip(q0+1, 1, SCALE_MAX_POS);
q0 = av_clip(q1-1, 0, SCALE_MAX_POS - 1);
}
for (i = 0; i < TRELLIS_STATES; i++) {
paths[0][i].cost = 0.0f;
paths[0][i].prev = -1;
}
for (j = 1; j < TRELLIS_STAGES; j++) {
for (i = 0; i < TRELLIS_STATES; i++) {
paths[j][i].cost = INFINITY;
paths[j][i].prev = -2;
}
}
idx = 1;
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 *coefs = &sce->coeffs[start];
float qmin, qmax;
int nz = 0;
bandaddr[idx] = w * 16 + g;
qmin = INT_MAX;
qmax = 0.0f;
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
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;
}
sce->zeroes[(w+w2)*16+g] = 0;
nz = 1;
for (i = 0; i < sce->ics.swb_sizes[g]; i++) {
float t = fabsf(coefs[w2*128+i]);
if (t > 0.0f)
qmin = FFMIN(qmin, t);
qmax = FFMAX(qmax, t);
}
}
if (nz) {
int minscale, maxscale;
float minrd = INFINITY;
float maxval;
//minimum scalefactor index is when minimum nonzero coefficient after quantizing is not clipped
minscale = coef2minsf(qmin);
//maximum scalefactor index is when maximum coefficient after quantizing is still not zero
maxscale = coef2maxsf(qmax);
minscale = av_clip(minscale - q0, 0, TRELLIS_STATES - 1);
maxscale = av_clip(maxscale - q0, 0, TRELLIS_STATES);
if (minscale == maxscale) {
maxscale = av_clip(minscale+1, 1, TRELLIS_STATES);
minscale = av_clip(maxscale-1, 0, TRELLIS_STATES - 1);
}
maxval = find_max_val(sce->ics.group_len[w], sce->ics.swb_sizes[g], s->scoefs+start);
for (q = minscale; q < maxscale; q++) {
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.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],
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
q + q0, cb, lambda / band->threshold, INFINITY, NULL, NULL, 0);
}
minrd = FFMIN(minrd, dist);
for (i = 0; i < q1 - q0; i++) {
float cost;
cost = paths[idx - 1][i].cost + dist
+ ff_aac_scalefactor_bits[q - i + SCALE_DIFF_ZERO];
if (cost < paths[idx][q].cost) {
paths[idx][q].cost = cost;
paths[idx][q].prev = i;
}
}
}
} else {
for (q = 0; q < q1 - q0; q++) {
paths[idx][q].cost = paths[idx - 1][q].cost + 1;
paths[idx][q].prev = q;
}
}
sce->zeroes[w*16+g] = !nz;
start += sce->ics.swb_sizes[g];
idx++;
}
}
idx--;
mincost = paths[idx][0].cost;
minq = 0;
for (i = 1; i < TRELLIS_STATES; i++) {
if (paths[idx][i].cost < mincost) {
mincost = paths[idx][i].cost;
minq = i;
}
}
while (idx) {
sce->sf_idx[bandaddr[idx]] = minq + q0;
minq = FFMAX(paths[idx][minq].prev, 0);
idx--;
}
//set the same quantizers inside window groups
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w])
for (g = 0; g < sce->ics.num_swb; g++)
for (w2 = 1; w2 < sce->ics.group_len[w]; w2++)
sce->sf_idx[(w+w2)*16+g] = sce->sf_idx[w*16+g];
}
static void search_for_quantizers_fast(AVCodecContext *avctx, AACEncContext *s,
SingleChannelElement *sce,
const float lambda)
{
2011-06-01 16:26:27 +00:00
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]) {
for (g = 0; g < sce->ics.num_swb; g++) {
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
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 + log2f(band->threshold), 80, 218);
sce->zeroes[(w+w2)*16+g] = 0;
}
minq = FFMIN(minq, sce->sf_idx[(w+w2)*16+g]);
}
}
}
for (i = 0; i < 128; i++) {
sce->sf_idx[i] = 140;
//av_clip(sce->sf_idx[i], minq, minq + SCALE_MAX_DIFF - 1);
}
//set the same quantizers inside window groups
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w])
for (g = 0; g < sce->ics.num_swb; g++)
for (w2 = 1; w2 < sce->ics.group_len[w]; w2++)
sce->sf_idx[(w+w2)*16+g] = sce->sf_idx[w*16+g];
}
static void search_for_pns(AACEncContext *s, AVCodecContext *avctx, SingleChannelElement *sce)
{
FFPsyBand *band;
int w, g, w2, i;
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
int wlen = 1024 / sce->ics.num_windows;
int bandwidth, cutoff;
float *PNS = &s->scoefs[0*128], *PNS34 = &s->scoefs[1*128];
float *NOR34 = &s->scoefs[3*128];
AAC encoder: improve SF range utilization This patch does 4 things, all of which interact and thus it woudln't be possible to commit them separately without causing either quality regressions or assertion failures. Fate comparison targets don't all reflect improvements in quality, yet listening tests show substantially improved quality and stability. 1. Increase SF range utilization. The spec requires SF delta values to be constrained within the range -60..60. The previous code was applying that range to the whole SF array and not only the deltas of consecutive values, because doing so requires smarter code: zeroing or otherwise skipping a band may invalidate lots of SF choices. This patch implements that logic to allow the coders to utilize the full dynamic range of scalefactors, increasing quality quite considerably, and fixing delta-SF-related assertion failures, since now the limitation is enforced rather than asserted. 2. PNS tweaks The previous modification makes big improvements in twoloop's efficiency, and every time that happens PNS logic needs to be tweaked accordingly to avoid it from stepping all over twoloop's decisions. This patch includes modifications of the sort. 3. Account for lowpass cutoff during PSY analysis The closer PSY's allocation is to final allocation the better the quality is, and given these modifications, twoloop is now very efficient at avoiding holes. Thus, to compute accurate thresholds, PSY needs to account for the lowpass applied implicitly during twoloop (by zeroing high bands). This patch makes twoloop set the cutoff in psymodel's context the first time it runs, and makes PSY account for it during threshold computation, making PE and threshold computations closer to the final allocation and thus achieving better subjective quality. 4. Tweaks to RC lambda tracking loop in relation to PNS Without this tweak some corner cases cause quality regressions. Basically, lambda needs to react faster to overall bitrate efficiency changes since now PNS can be quite successful in enforcing maximum bitrates, when PSY allocates too many bits to the lower bands, suppressing the signals RC logic uses to lower lambda in those cases and causing aggressive PNS. This tweak makes PNS much less aggressive, though it can still use some further tweaks. Also update MIPS specializations and adjust fuzz Also in lavc/mips/aacpsy_mips.h: remove trailing whitespace
2015-12-01 06:28:36 +00:00
uint8_t nextband[128];
const float lambda = s->lambda;
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
const float freq_mult = avctx->sample_rate*0.5f/wlen;
const float thr_mult = NOISE_LAMBDA_REPLACE*(100.0f/lambda);
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
const float spread_threshold = FFMIN(0.75f, NOISE_SPREAD_THRESHOLD*FFMAX(0.5f, lambda/100.f));
const float dist_bias = av_clipf(4.f * 120 / lambda, 0.25f, 4.0f);
const float pns_transient_energy_r = FFMIN(0.7f, lambda / 140.f);
int refbits = avctx->bit_rate * 1024.0 / avctx->sample_rate
/ ((avctx->flags & CODEC_FLAG_QSCALE) ? 2.0f : avctx->channels)
* (lambda / 120.f);
/** Keep this in sync with twoloop's cutoff selection */
float rate_bandwidth_multiplier = 1.5f;
AAC encoder: improve SF range utilization This patch does 4 things, all of which interact and thus it woudln't be possible to commit them separately without causing either quality regressions or assertion failures. Fate comparison targets don't all reflect improvements in quality, yet listening tests show substantially improved quality and stability. 1. Increase SF range utilization. The spec requires SF delta values to be constrained within the range -60..60. The previous code was applying that range to the whole SF array and not only the deltas of consecutive values, because doing so requires smarter code: zeroing or otherwise skipping a band may invalidate lots of SF choices. This patch implements that logic to allow the coders to utilize the full dynamic range of scalefactors, increasing quality quite considerably, and fixing delta-SF-related assertion failures, since now the limitation is enforced rather than asserted. 2. PNS tweaks The previous modification makes big improvements in twoloop's efficiency, and every time that happens PNS logic needs to be tweaked accordingly to avoid it from stepping all over twoloop's decisions. This patch includes modifications of the sort. 3. Account for lowpass cutoff during PSY analysis The closer PSY's allocation is to final allocation the better the quality is, and given these modifications, twoloop is now very efficient at avoiding holes. Thus, to compute accurate thresholds, PSY needs to account for the lowpass applied implicitly during twoloop (by zeroing high bands). This patch makes twoloop set the cutoff in psymodel's context the first time it runs, and makes PSY account for it during threshold computation, making PE and threshold computations closer to the final allocation and thus achieving better subjective quality. 4. Tweaks to RC lambda tracking loop in relation to PNS Without this tweak some corner cases cause quality regressions. Basically, lambda needs to react faster to overall bitrate efficiency changes since now PNS can be quite successful in enforcing maximum bitrates, when PSY allocates too many bits to the lower bands, suppressing the signals RC logic uses to lower lambda in those cases and causing aggressive PNS. This tweak makes PNS much less aggressive, though it can still use some further tweaks. Also update MIPS specializations and adjust fuzz Also in lavc/mips/aacpsy_mips.h: remove trailing whitespace
2015-12-01 06:28:36 +00:00
int prev = -1000, prev_sf = -1;
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
int frame_bit_rate = (avctx->flags & CODEC_FLAG_QSCALE)
? (refbits * rate_bandwidth_multiplier * avctx->sample_rate / 1024)
: (avctx->bit_rate / avctx->channels);
frame_bit_rate *= 1.15f;
if (avctx->cutoff > 0) {
bandwidth = avctx->cutoff;
} else {
bandwidth = FFMAX(3000, AAC_CUTOFF_FROM_BITRATE(frame_bit_rate, 1, avctx->sample_rate));
}
cutoff = bandwidth * 2 * wlen / avctx->sample_rate;
memcpy(sce->band_alt, sce->band_type, sizeof(sce->band_type));
AAC encoder: improve SF range utilization This patch does 4 things, all of which interact and thus it woudln't be possible to commit them separately without causing either quality regressions or assertion failures. Fate comparison targets don't all reflect improvements in quality, yet listening tests show substantially improved quality and stability. 1. Increase SF range utilization. The spec requires SF delta values to be constrained within the range -60..60. The previous code was applying that range to the whole SF array and not only the deltas of consecutive values, because doing so requires smarter code: zeroing or otherwise skipping a band may invalidate lots of SF choices. This patch implements that logic to allow the coders to utilize the full dynamic range of scalefactors, increasing quality quite considerably, and fixing delta-SF-related assertion failures, since now the limitation is enforced rather than asserted. 2. PNS tweaks The previous modification makes big improvements in twoloop's efficiency, and every time that happens PNS logic needs to be tweaked accordingly to avoid it from stepping all over twoloop's decisions. This patch includes modifications of the sort. 3. Account for lowpass cutoff during PSY analysis The closer PSY's allocation is to final allocation the better the quality is, and given these modifications, twoloop is now very efficient at avoiding holes. Thus, to compute accurate thresholds, PSY needs to account for the lowpass applied implicitly during twoloop (by zeroing high bands). This patch makes twoloop set the cutoff in psymodel's context the first time it runs, and makes PSY account for it during threshold computation, making PE and threshold computations closer to the final allocation and thus achieving better subjective quality. 4. Tweaks to RC lambda tracking loop in relation to PNS Without this tweak some corner cases cause quality regressions. Basically, lambda needs to react faster to overall bitrate efficiency changes since now PNS can be quite successful in enforcing maximum bitrates, when PSY allocates too many bits to the lower bands, suppressing the signals RC logic uses to lower lambda in those cases and causing aggressive PNS. This tweak makes PNS much less aggressive, though it can still use some further tweaks. Also update MIPS specializations and adjust fuzz Also in lavc/mips/aacpsy_mips.h: remove trailing whitespace
2015-12-01 06:28:36 +00:00
ff_init_nextband_map(sce, nextband);
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
int wstart = w*128;
for (g = 0; g < sce->ics.num_swb; g++) {
int noise_sfi;
float dist1 = 0.0f, dist2 = 0.0f, noise_amp;
float pns_energy = 0.0f, pns_tgt_energy, energy_ratio, dist_thresh;
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
float sfb_energy = 0.0f, threshold = 0.0f, spread = 2.0f;
float min_energy = -1.0f, max_energy = 0.0f;
const int start = wstart+sce->ics.swb_offset[g];
const float freq = (start-wstart)*freq_mult;
const float freq_boost = FFMAX(0.88f*freq/NOISE_LOW_LIMIT, 1.0f);
if (freq < NOISE_LOW_LIMIT || (start-wstart) >= cutoff) {
if (!sce->zeroes[w*16+g])
prev_sf = sce->sf_idx[w*16+g];
continue;
}
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g];
sfb_energy += band->energy;
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
spread = FFMIN(spread, band->spread);
threshold += band->threshold;
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
if (!w2) {
min_energy = max_energy = band->energy;
} else {
min_energy = FFMIN(min_energy, band->energy);
max_energy = FFMAX(max_energy, band->energy);
}
}
/* Ramps down at ~8000Hz and loosens the dist threshold */
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
dist_thresh = av_clipf(2.5f*NOISE_LOW_LIMIT/freq, 0.5f, 2.5f) * dist_bias;
/* PNS is acceptable when all of these are true:
* 1. high spread energy (noise-like band)
* 2. near-threshold energy (high PE means the random nature of PNS content will be noticed)
* 3. on short window groups, all windows have similar energy (variations in energy would be destroyed by PNS)
*
* At this stage, point 2 is relaxed for zeroed bands near the noise threshold (hole avoidance is more important)
*/
AAC encoder: improve SF range utilization This patch does 4 things, all of which interact and thus it woudln't be possible to commit them separately without causing either quality regressions or assertion failures. Fate comparison targets don't all reflect improvements in quality, yet listening tests show substantially improved quality and stability. 1. Increase SF range utilization. The spec requires SF delta values to be constrained within the range -60..60. The previous code was applying that range to the whole SF array and not only the deltas of consecutive values, because doing so requires smarter code: zeroing or otherwise skipping a band may invalidate lots of SF choices. This patch implements that logic to allow the coders to utilize the full dynamic range of scalefactors, increasing quality quite considerably, and fixing delta-SF-related assertion failures, since now the limitation is enforced rather than asserted. 2. PNS tweaks The previous modification makes big improvements in twoloop's efficiency, and every time that happens PNS logic needs to be tweaked accordingly to avoid it from stepping all over twoloop's decisions. This patch includes modifications of the sort. 3. Account for lowpass cutoff during PSY analysis The closer PSY's allocation is to final allocation the better the quality is, and given these modifications, twoloop is now very efficient at avoiding holes. Thus, to compute accurate thresholds, PSY needs to account for the lowpass applied implicitly during twoloop (by zeroing high bands). This patch makes twoloop set the cutoff in psymodel's context the first time it runs, and makes PSY account for it during threshold computation, making PE and threshold computations closer to the final allocation and thus achieving better subjective quality. 4. Tweaks to RC lambda tracking loop in relation to PNS Without this tweak some corner cases cause quality regressions. Basically, lambda needs to react faster to overall bitrate efficiency changes since now PNS can be quite successful in enforcing maximum bitrates, when PSY allocates too many bits to the lower bands, suppressing the signals RC logic uses to lower lambda in those cases and causing aggressive PNS. This tweak makes PNS much less aggressive, though it can still use some further tweaks. Also update MIPS specializations and adjust fuzz Also in lavc/mips/aacpsy_mips.h: remove trailing whitespace
2015-12-01 06:28:36 +00:00
if ((!sce->zeroes[w*16+g] && !ff_sfdelta_can_remove_band(sce, nextband, prev_sf, w*16+g)) ||
((sce->zeroes[w*16+g] || !sce->band_alt[w*16+g]) && sfb_energy < threshold*sqrtf(1.0f/freq_boost)) || spread < spread_threshold ||
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
(!sce->zeroes[w*16+g] && sce->band_alt[w*16+g] && sfb_energy > threshold*thr_mult*freq_boost) ||
min_energy < pns_transient_energy_r * max_energy ) {
sce->pns_ener[w*16+g] = sfb_energy;
AAC encoder: improve SF range utilization This patch does 4 things, all of which interact and thus it woudln't be possible to commit them separately without causing either quality regressions or assertion failures. Fate comparison targets don't all reflect improvements in quality, yet listening tests show substantially improved quality and stability. 1. Increase SF range utilization. The spec requires SF delta values to be constrained within the range -60..60. The previous code was applying that range to the whole SF array and not only the deltas of consecutive values, because doing so requires smarter code: zeroing or otherwise skipping a band may invalidate lots of SF choices. This patch implements that logic to allow the coders to utilize the full dynamic range of scalefactors, increasing quality quite considerably, and fixing delta-SF-related assertion failures, since now the limitation is enforced rather than asserted. 2. PNS tweaks The previous modification makes big improvements in twoloop's efficiency, and every time that happens PNS logic needs to be tweaked accordingly to avoid it from stepping all over twoloop's decisions. This patch includes modifications of the sort. 3. Account for lowpass cutoff during PSY analysis The closer PSY's allocation is to final allocation the better the quality is, and given these modifications, twoloop is now very efficient at avoiding holes. Thus, to compute accurate thresholds, PSY needs to account for the lowpass applied implicitly during twoloop (by zeroing high bands). This patch makes twoloop set the cutoff in psymodel's context the first time it runs, and makes PSY account for it during threshold computation, making PE and threshold computations closer to the final allocation and thus achieving better subjective quality. 4. Tweaks to RC lambda tracking loop in relation to PNS Without this tweak some corner cases cause quality regressions. Basically, lambda needs to react faster to overall bitrate efficiency changes since now PNS can be quite successful in enforcing maximum bitrates, when PSY allocates too many bits to the lower bands, suppressing the signals RC logic uses to lower lambda in those cases and causing aggressive PNS. This tweak makes PNS much less aggressive, though it can still use some further tweaks. Also update MIPS specializations and adjust fuzz Also in lavc/mips/aacpsy_mips.h: remove trailing whitespace
2015-12-01 06:28:36 +00:00
if (!sce->zeroes[w*16+g])
prev_sf = sce->sf_idx[w*16+g];
continue;
}
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
pns_tgt_energy = sfb_energy*FFMIN(1.0f, spread*spread);
noise_sfi = av_clip(roundf(log2f(pns_tgt_energy)*2), -100, 155); /* Quantize */
noise_amp = -ff_aac_pow2sf_tab[noise_sfi + POW_SF2_ZERO]; /* Dequantize */
AAC encoder: improve SF range utilization This patch does 4 things, all of which interact and thus it woudln't be possible to commit them separately without causing either quality regressions or assertion failures. Fate comparison targets don't all reflect improvements in quality, yet listening tests show substantially improved quality and stability. 1. Increase SF range utilization. The spec requires SF delta values to be constrained within the range -60..60. The previous code was applying that range to the whole SF array and not only the deltas of consecutive values, because doing so requires smarter code: zeroing or otherwise skipping a band may invalidate lots of SF choices. This patch implements that logic to allow the coders to utilize the full dynamic range of scalefactors, increasing quality quite considerably, and fixing delta-SF-related assertion failures, since now the limitation is enforced rather than asserted. 2. PNS tweaks The previous modification makes big improvements in twoloop's efficiency, and every time that happens PNS logic needs to be tweaked accordingly to avoid it from stepping all over twoloop's decisions. This patch includes modifications of the sort. 3. Account for lowpass cutoff during PSY analysis The closer PSY's allocation is to final allocation the better the quality is, and given these modifications, twoloop is now very efficient at avoiding holes. Thus, to compute accurate thresholds, PSY needs to account for the lowpass applied implicitly during twoloop (by zeroing high bands). This patch makes twoloop set the cutoff in psymodel's context the first time it runs, and makes PSY account for it during threshold computation, making PE and threshold computations closer to the final allocation and thus achieving better subjective quality. 4. Tweaks to RC lambda tracking loop in relation to PNS Without this tweak some corner cases cause quality regressions. Basically, lambda needs to react faster to overall bitrate efficiency changes since now PNS can be quite successful in enforcing maximum bitrates, when PSY allocates too many bits to the lower bands, suppressing the signals RC logic uses to lower lambda in those cases and causing aggressive PNS. This tweak makes PNS much less aggressive, though it can still use some further tweaks. Also update MIPS specializations and adjust fuzz Also in lavc/mips/aacpsy_mips.h: remove trailing whitespace
2015-12-01 06:28:36 +00:00
if (prev != -1000) {
int noise_sfdiff = noise_sfi - prev + SCALE_DIFF_ZERO;
if (noise_sfdiff < 0 || noise_sfdiff > 2*SCALE_MAX_DIFF) {
if (!sce->zeroes[w*16+g])
prev_sf = sce->sf_idx[w*16+g];
continue;
}
}
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
float band_energy, scale, pns_senergy;
const int start_c = (w+w2)*128+sce->ics.swb_offset[g];
band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g];
for (i = 0; i < sce->ics.swb_sizes[g]; i+=2) {
double rnd[2];
av_bmg_get(&s->lfg, rnd);
PNS[i+0] = (float)rnd[0];
PNS[i+1] = (float)rnd[1];
}
band_energy = s->fdsp->scalarproduct_float(PNS, PNS, sce->ics.swb_sizes[g]);
scale = noise_amp/sqrtf(band_energy);
s->fdsp->vector_fmul_scalar(PNS, PNS, scale, sce->ics.swb_sizes[g]);
pns_senergy = s->fdsp->scalarproduct_float(PNS, PNS, sce->ics.swb_sizes[g]);
pns_energy += pns_senergy;
abs_pow34_v(NOR34, &sce->coeffs[start_c], sce->ics.swb_sizes[g]);
abs_pow34_v(PNS34, PNS, sce->ics.swb_sizes[g]);
dist1 += quantize_band_cost(s, &sce->coeffs[start_c],
NOR34,
sce->ics.swb_sizes[g],
sce->sf_idx[(w+w2)*16+g],
sce->band_alt[(w+w2)*16+g],
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
lambda/band->threshold, INFINITY, NULL, NULL, 0);
/* Estimate rd on average as 5 bits for SF, 4 for the CB, plus spread energy * lambda/thr */
dist2 += band->energy/(band->spread*band->spread)*lambda*dist_thresh/band->threshold;
}
if (g && sce->band_type[w*16+g-1] == NOISE_BT) {
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
dist2 += 5;
} else {
dist2 += 9;
}
energy_ratio = pns_tgt_energy/pns_energy; /* Compensates for quantization error */
sce->pns_ener[w*16+g] = energy_ratio*pns_tgt_energy;
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
if (sce->zeroes[w*16+g] || !sce->band_alt[w*16+g] || (energy_ratio > 0.85f && energy_ratio < 1.25f && dist2 < dist1)) {
sce->band_type[w*16+g] = NOISE_BT;
sce->zeroes[w*16+g] = 0;
AAC encoder: improve SF range utilization This patch does 4 things, all of which interact and thus it woudln't be possible to commit them separately without causing either quality regressions or assertion failures. Fate comparison targets don't all reflect improvements in quality, yet listening tests show substantially improved quality and stability. 1. Increase SF range utilization. The spec requires SF delta values to be constrained within the range -60..60. The previous code was applying that range to the whole SF array and not only the deltas of consecutive values, because doing so requires smarter code: zeroing or otherwise skipping a band may invalidate lots of SF choices. This patch implements that logic to allow the coders to utilize the full dynamic range of scalefactors, increasing quality quite considerably, and fixing delta-SF-related assertion failures, since now the limitation is enforced rather than asserted. 2. PNS tweaks The previous modification makes big improvements in twoloop's efficiency, and every time that happens PNS logic needs to be tweaked accordingly to avoid it from stepping all over twoloop's decisions. This patch includes modifications of the sort. 3. Account for lowpass cutoff during PSY analysis The closer PSY's allocation is to final allocation the better the quality is, and given these modifications, twoloop is now very efficient at avoiding holes. Thus, to compute accurate thresholds, PSY needs to account for the lowpass applied implicitly during twoloop (by zeroing high bands). This patch makes twoloop set the cutoff in psymodel's context the first time it runs, and makes PSY account for it during threshold computation, making PE and threshold computations closer to the final allocation and thus achieving better subjective quality. 4. Tweaks to RC lambda tracking loop in relation to PNS Without this tweak some corner cases cause quality regressions. Basically, lambda needs to react faster to overall bitrate efficiency changes since now PNS can be quite successful in enforcing maximum bitrates, when PSY allocates too many bits to the lower bands, suppressing the signals RC logic uses to lower lambda in those cases and causing aggressive PNS. This tweak makes PNS much less aggressive, though it can still use some further tweaks. Also update MIPS specializations and adjust fuzz Also in lavc/mips/aacpsy_mips.h: remove trailing whitespace
2015-12-01 06:28:36 +00:00
prev = noise_sfi;
} else {
if (!sce->zeroes[w*16+g])
prev_sf = sce->sf_idx[w*16+g];
}
}
}
}
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
static void mark_pns(AACEncContext *s, AVCodecContext *avctx, SingleChannelElement *sce)
{
FFPsyBand *band;
int w, g, w2;
int wlen = 1024 / sce->ics.num_windows;
int bandwidth, cutoff;
const float lambda = s->lambda;
const float freq_mult = avctx->sample_rate*0.5f/wlen;
const float spread_threshold = FFMIN(0.75f, NOISE_SPREAD_THRESHOLD*FFMAX(0.5f, lambda/100.f));
const float pns_transient_energy_r = FFMIN(0.7f, lambda / 140.f);
int refbits = avctx->bit_rate * 1024.0 / avctx->sample_rate
/ ((avctx->flags & CODEC_FLAG_QSCALE) ? 2.0f : avctx->channels)
* (lambda / 120.f);
/** Keep this in sync with twoloop's cutoff selection */
float rate_bandwidth_multiplier = 1.5f;
int frame_bit_rate = (avctx->flags & CODEC_FLAG_QSCALE)
? (refbits * rate_bandwidth_multiplier * avctx->sample_rate / 1024)
: (avctx->bit_rate / avctx->channels);
frame_bit_rate *= 1.15f;
if (avctx->cutoff > 0) {
bandwidth = avctx->cutoff;
} else {
bandwidth = FFMAX(3000, AAC_CUTOFF_FROM_BITRATE(frame_bit_rate, 1, avctx->sample_rate));
}
cutoff = bandwidth * 2 * wlen / avctx->sample_rate;
memcpy(sce->band_alt, sce->band_type, sizeof(sce->band_type));
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
for (g = 0; g < sce->ics.num_swb; g++) {
float sfb_energy = 0.0f, threshold = 0.0f, spread = 2.0f;
float min_energy = -1.0f, max_energy = 0.0f;
const int start = sce->ics.swb_offset[g];
const float freq = start*freq_mult;
const float freq_boost = FFMAX(0.88f*freq/NOISE_LOW_LIMIT, 1.0f);
if (freq < NOISE_LOW_LIMIT || start >= cutoff) {
sce->can_pns[w*16+g] = 0;
continue;
}
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g];
sfb_energy += band->energy;
spread = FFMIN(spread, band->spread);
threshold += band->threshold;
if (!w2) {
min_energy = max_energy = band->energy;
} else {
min_energy = FFMIN(min_energy, band->energy);
max_energy = FFMAX(max_energy, band->energy);
}
}
/* PNS is acceptable when all of these are true:
* 1. high spread energy (noise-like band)
* 2. near-threshold energy (high PE means the random nature of PNS content will be noticed)
* 3. on short window groups, all windows have similar energy (variations in energy would be destroyed by PNS)
*/
sce->pns_ener[w*16+g] = sfb_energy;
if (sfb_energy < threshold*sqrtf(1.5f/freq_boost) || spread < spread_threshold || min_energy < pns_transient_energy_r * max_energy) {
sce->can_pns[w*16+g] = 0;
} else {
sce->can_pns[w*16+g] = 1;
}
}
}
}
static void search_for_ms(AACEncContext *s, ChannelElement *cpe)
{
AAC encoder: improve SF range utilization This patch does 4 things, all of which interact and thus it woudln't be possible to commit them separately without causing either quality regressions or assertion failures. Fate comparison targets don't all reflect improvements in quality, yet listening tests show substantially improved quality and stability. 1. Increase SF range utilization. The spec requires SF delta values to be constrained within the range -60..60. The previous code was applying that range to the whole SF array and not only the deltas of consecutive values, because doing so requires smarter code: zeroing or otherwise skipping a band may invalidate lots of SF choices. This patch implements that logic to allow the coders to utilize the full dynamic range of scalefactors, increasing quality quite considerably, and fixing delta-SF-related assertion failures, since now the limitation is enforced rather than asserted. 2. PNS tweaks The previous modification makes big improvements in twoloop's efficiency, and every time that happens PNS logic needs to be tweaked accordingly to avoid it from stepping all over twoloop's decisions. This patch includes modifications of the sort. 3. Account for lowpass cutoff during PSY analysis The closer PSY's allocation is to final allocation the better the quality is, and given these modifications, twoloop is now very efficient at avoiding holes. Thus, to compute accurate thresholds, PSY needs to account for the lowpass applied implicitly during twoloop (by zeroing high bands). This patch makes twoloop set the cutoff in psymodel's context the first time it runs, and makes PSY account for it during threshold computation, making PE and threshold computations closer to the final allocation and thus achieving better subjective quality. 4. Tweaks to RC lambda tracking loop in relation to PNS Without this tweak some corner cases cause quality regressions. Basically, lambda needs to react faster to overall bitrate efficiency changes since now PNS can be quite successful in enforcing maximum bitrates, when PSY allocates too many bits to the lower bands, suppressing the signals RC logic uses to lower lambda in those cases and causing aggressive PNS. This tweak makes PNS much less aggressive, though it can still use some further tweaks. Also update MIPS specializations and adjust fuzz Also in lavc/mips/aacpsy_mips.h: remove trailing whitespace
2015-12-01 06:28:36 +00:00
int start = 0, i, w, w2, g, sid_sf_boost, prev_mid, prev_side;
uint8_t nextband0[128], nextband1[128];
float M[128], S[128];
float *L34 = s->scoefs, *R34 = s->scoefs + 128, *M34 = s->scoefs + 128*2, *S34 = s->scoefs + 128*3;
const float lambda = s->lambda;
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
const float mslambda = FFMIN(1.0f, lambda / 120.f);
SingleChannelElement *sce0 = &cpe->ch[0];
SingleChannelElement *sce1 = &cpe->ch[1];
if (!cpe->common_window)
return;
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
AAC encoder: improve SF range utilization This patch does 4 things, all of which interact and thus it woudln't be possible to commit them separately without causing either quality regressions or assertion failures. Fate comparison targets don't all reflect improvements in quality, yet listening tests show substantially improved quality and stability. 1. Increase SF range utilization. The spec requires SF delta values to be constrained within the range -60..60. The previous code was applying that range to the whole SF array and not only the deltas of consecutive values, because doing so requires smarter code: zeroing or otherwise skipping a band may invalidate lots of SF choices. This patch implements that logic to allow the coders to utilize the full dynamic range of scalefactors, increasing quality quite considerably, and fixing delta-SF-related assertion failures, since now the limitation is enforced rather than asserted. 2. PNS tweaks The previous modification makes big improvements in twoloop's efficiency, and every time that happens PNS logic needs to be tweaked accordingly to avoid it from stepping all over twoloop's decisions. This patch includes modifications of the sort. 3. Account for lowpass cutoff during PSY analysis The closer PSY's allocation is to final allocation the better the quality is, and given these modifications, twoloop is now very efficient at avoiding holes. Thus, to compute accurate thresholds, PSY needs to account for the lowpass applied implicitly during twoloop (by zeroing high bands). This patch makes twoloop set the cutoff in psymodel's context the first time it runs, and makes PSY account for it during threshold computation, making PE and threshold computations closer to the final allocation and thus achieving better subjective quality. 4. Tweaks to RC lambda tracking loop in relation to PNS Without this tweak some corner cases cause quality regressions. Basically, lambda needs to react faster to overall bitrate efficiency changes since now PNS can be quite successful in enforcing maximum bitrates, when PSY allocates too many bits to the lower bands, suppressing the signals RC logic uses to lower lambda in those cases and causing aggressive PNS. This tweak makes PNS much less aggressive, though it can still use some further tweaks. Also update MIPS specializations and adjust fuzz Also in lavc/mips/aacpsy_mips.h: remove trailing whitespace
2015-12-01 06:28:36 +00:00
/** Scout out next nonzero bands */
ff_init_nextband_map(sce0, nextband0);
ff_init_nextband_map(sce1, nextband1);
prev_mid = sce0->sf_idx[0];
prev_side = sce1->sf_idx[0];
for (w = 0; w < sce0->ics.num_windows; w += sce0->ics.group_len[w]) {
start = 0;
for (g = 0; g < sce0->ics.num_swb; g++) {
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
float bmax = bval2bmax(g * 17.0f / sce0->ics.num_swb) / 0.0045f;
if (!cpe->is_mask[w*16+g])
cpe->ms_mask[w*16+g] = 0;
if (!sce0->zeroes[w*16+g] && !sce1->zeroes[w*16+g] && !cpe->is_mask[w*16+g]) {
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
float Mmax = 0.0f, Smax = 0.0f;
/* Must compute mid/side SF and book for the whole window group */
for (w2 = 0; w2 < sce0->ics.group_len[w]; w2++) {
for (i = 0; i < sce0->ics.swb_sizes[g]; i++) {
M[i] = (sce0->coeffs[start+(w+w2)*128+i]
+ sce1->coeffs[start+(w+w2)*128+i]) * 0.5;
S[i] = M[i]
- sce1->coeffs[start+(w+w2)*128+i];
}
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
abs_pow34_v(M34, M, sce0->ics.swb_sizes[g]);
abs_pow34_v(S34, S, sce0->ics.swb_sizes[g]);
for (i = 0; i < sce0->ics.swb_sizes[g]; i++ ) {
Mmax = FFMAX(Mmax, M34[i]);
Smax = FFMAX(Smax, S34[i]);
}
}
for (sid_sf_boost = 0; sid_sf_boost < 4; sid_sf_boost++) {
float dist1 = 0.0f, dist2 = 0.0f;
int B0 = 0, B1 = 0;
int minidx;
int mididx, sididx;
int midcb, sidcb;
minidx = FFMIN(sce0->sf_idx[w*16+g], sce1->sf_idx[w*16+g]);
AAC encoder: improve SF range utilization This patch does 4 things, all of which interact and thus it woudln't be possible to commit them separately without causing either quality regressions or assertion failures. Fate comparison targets don't all reflect improvements in quality, yet listening tests show substantially improved quality and stability. 1. Increase SF range utilization. The spec requires SF delta values to be constrained within the range -60..60. The previous code was applying that range to the whole SF array and not only the deltas of consecutive values, because doing so requires smarter code: zeroing or otherwise skipping a band may invalidate lots of SF choices. This patch implements that logic to allow the coders to utilize the full dynamic range of scalefactors, increasing quality quite considerably, and fixing delta-SF-related assertion failures, since now the limitation is enforced rather than asserted. 2. PNS tweaks The previous modification makes big improvements in twoloop's efficiency, and every time that happens PNS logic needs to be tweaked accordingly to avoid it from stepping all over twoloop's decisions. This patch includes modifications of the sort. 3. Account for lowpass cutoff during PSY analysis The closer PSY's allocation is to final allocation the better the quality is, and given these modifications, twoloop is now very efficient at avoiding holes. Thus, to compute accurate thresholds, PSY needs to account for the lowpass applied implicitly during twoloop (by zeroing high bands). This patch makes twoloop set the cutoff in psymodel's context the first time it runs, and makes PSY account for it during threshold computation, making PE and threshold computations closer to the final allocation and thus achieving better subjective quality. 4. Tweaks to RC lambda tracking loop in relation to PNS Without this tweak some corner cases cause quality regressions. Basically, lambda needs to react faster to overall bitrate efficiency changes since now PNS can be quite successful in enforcing maximum bitrates, when PSY allocates too many bits to the lower bands, suppressing the signals RC logic uses to lower lambda in those cases and causing aggressive PNS. This tweak makes PNS much less aggressive, though it can still use some further tweaks. Also update MIPS specializations and adjust fuzz Also in lavc/mips/aacpsy_mips.h: remove trailing whitespace
2015-12-01 06:28:36 +00:00
mididx = av_clip(minidx, 0, SCALE_MAX_POS - SCALE_DIV_512);
sididx = av_clip(minidx - sid_sf_boost * 3, 0, SCALE_MAX_POS - SCALE_DIV_512);
if (sce0->band_type[w*16+g] != NOISE_BT && sce1->band_type[w*16+g] != NOISE_BT
AAC encoder: improve SF range utilization This patch does 4 things, all of which interact and thus it woudln't be possible to commit them separately without causing either quality regressions or assertion failures. Fate comparison targets don't all reflect improvements in quality, yet listening tests show substantially improved quality and stability. 1. Increase SF range utilization. The spec requires SF delta values to be constrained within the range -60..60. The previous code was applying that range to the whole SF array and not only the deltas of consecutive values, because doing so requires smarter code: zeroing or otherwise skipping a band may invalidate lots of SF choices. This patch implements that logic to allow the coders to utilize the full dynamic range of scalefactors, increasing quality quite considerably, and fixing delta-SF-related assertion failures, since now the limitation is enforced rather than asserted. 2. PNS tweaks The previous modification makes big improvements in twoloop's efficiency, and every time that happens PNS logic needs to be tweaked accordingly to avoid it from stepping all over twoloop's decisions. This patch includes modifications of the sort. 3. Account for lowpass cutoff during PSY analysis The closer PSY's allocation is to final allocation the better the quality is, and given these modifications, twoloop is now very efficient at avoiding holes. Thus, to compute accurate thresholds, PSY needs to account for the lowpass applied implicitly during twoloop (by zeroing high bands). This patch makes twoloop set the cutoff in psymodel's context the first time it runs, and makes PSY account for it during threshold computation, making PE and threshold computations closer to the final allocation and thus achieving better subjective quality. 4. Tweaks to RC lambda tracking loop in relation to PNS Without this tweak some corner cases cause quality regressions. Basically, lambda needs to react faster to overall bitrate efficiency changes since now PNS can be quite successful in enforcing maximum bitrates, when PSY allocates too many bits to the lower bands, suppressing the signals RC logic uses to lower lambda in those cases and causing aggressive PNS. This tweak makes PNS much less aggressive, though it can still use some further tweaks. Also update MIPS specializations and adjust fuzz Also in lavc/mips/aacpsy_mips.h: remove trailing whitespace
2015-12-01 06:28:36 +00:00
&& ( !ff_sfdelta_can_replace(sce0, nextband0, prev_mid, mididx, w*16+g)
|| !ff_sfdelta_can_replace(sce1, nextband1, prev_side, sididx, w*16+g))) {
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
/* scalefactor range violation, bad stuff, will decrease quality unacceptably */
continue;
}
AAC encoder: improve SF range utilization This patch does 4 things, all of which interact and thus it woudln't be possible to commit them separately without causing either quality regressions or assertion failures. Fate comparison targets don't all reflect improvements in quality, yet listening tests show substantially improved quality and stability. 1. Increase SF range utilization. The spec requires SF delta values to be constrained within the range -60..60. The previous code was applying that range to the whole SF array and not only the deltas of consecutive values, because doing so requires smarter code: zeroing or otherwise skipping a band may invalidate lots of SF choices. This patch implements that logic to allow the coders to utilize the full dynamic range of scalefactors, increasing quality quite considerably, and fixing delta-SF-related assertion failures, since now the limitation is enforced rather than asserted. 2. PNS tweaks The previous modification makes big improvements in twoloop's efficiency, and every time that happens PNS logic needs to be tweaked accordingly to avoid it from stepping all over twoloop's decisions. This patch includes modifications of the sort. 3. Account for lowpass cutoff during PSY analysis The closer PSY's allocation is to final allocation the better the quality is, and given these modifications, twoloop is now very efficient at avoiding holes. Thus, to compute accurate thresholds, PSY needs to account for the lowpass applied implicitly during twoloop (by zeroing high bands). This patch makes twoloop set the cutoff in psymodel's context the first time it runs, and makes PSY account for it during threshold computation, making PE and threshold computations closer to the final allocation and thus achieving better subjective quality. 4. Tweaks to RC lambda tracking loop in relation to PNS Without this tweak some corner cases cause quality regressions. Basically, lambda needs to react faster to overall bitrate efficiency changes since now PNS can be quite successful in enforcing maximum bitrates, when PSY allocates too many bits to the lower bands, suppressing the signals RC logic uses to lower lambda in those cases and causing aggressive PNS. This tweak makes PNS much less aggressive, though it can still use some further tweaks. Also update MIPS specializations and adjust fuzz Also in lavc/mips/aacpsy_mips.h: remove trailing whitespace
2015-12-01 06:28:36 +00:00
midcb = find_min_book(Mmax, mididx);
sidcb = find_min_book(Smax, sididx);
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
/* No CB can be zero */
midcb = FFMAX(1,midcb);
sidcb = FFMAX(1,sidcb);
for (w2 = 0; w2 < sce0->ics.group_len[w]; w2++) {
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);
int b1,b2,b3,b4;
for (i = 0; i < sce0->ics.swb_sizes[g]; i++) {
M[i] = (sce0->coeffs[start+(w+w2)*128+i]
+ sce1->coeffs[start+(w+w2)*128+i]) * 0.5;
S[i] = M[i]
- sce1->coeffs[start+(w+w2)*128+i];
}
abs_pow34_v(L34, sce0->coeffs+start+(w+w2)*128, sce0->ics.swb_sizes[g]);
abs_pow34_v(R34, sce1->coeffs+start+(w+w2)*128, sce0->ics.swb_sizes[g]);
abs_pow34_v(M34, M, sce0->ics.swb_sizes[g]);
abs_pow34_v(S34, S, sce0->ics.swb_sizes[g]);
dist1 += quantize_band_cost(s, &sce0->coeffs[start + (w+w2)*128],
L34,
sce0->ics.swb_sizes[g],
sce0->sf_idx[w*16+g],
sce0->band_type[w*16+g],
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
lambda / band0->threshold, INFINITY, &b1, NULL, 0);
dist1 += quantize_band_cost(s, &sce1->coeffs[start + (w+w2)*128],
R34,
sce1->ics.swb_sizes[g],
sce1->sf_idx[w*16+g],
sce1->band_type[w*16+g],
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
lambda / band1->threshold, INFINITY, &b2, NULL, 0);
dist2 += quantize_band_cost(s, M,
M34,
sce0->ics.swb_sizes[g],
mididx,
midcb,
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
lambda / minthr, INFINITY, &b3, NULL, 0);
dist2 += quantize_band_cost(s, S,
S34,
sce1->ics.swb_sizes[g],
sididx,
sidcb,
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
mslambda / (minthr * bmax), INFINITY, &b4, NULL, 0);
B0 += b1+b2;
B1 += b3+b4;
dist1 -= b1+b2;
dist2 -= b3+b4;
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
}
cpe->ms_mask[w*16+g] = dist2 <= dist1 && B1 < B0;
if (cpe->ms_mask[w*16+g]) {
if (sce0->band_type[w*16+g] != NOISE_BT && sce1->band_type[w*16+g] != NOISE_BT) {
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
sce0->sf_idx[w*16+g] = mididx;
sce1->sf_idx[w*16+g] = sididx;
sce0->band_type[w*16+g] = midcb;
sce1->band_type[w*16+g] = sidcb;
} else if ((sce0->band_type[w*16+g] != NOISE_BT) ^ (sce1->band_type[w*16+g] != NOISE_BT)) {
/* ms_mask unneeded, and it confuses some decoders */
cpe->ms_mask[w*16+g] = 0;
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
}
break;
} else if (B1 > B0) {
/* More boost won't fix this */
break;
}
}
}
AAC encoder: improve SF range utilization This patch does 4 things, all of which interact and thus it woudln't be possible to commit them separately without causing either quality regressions or assertion failures. Fate comparison targets don't all reflect improvements in quality, yet listening tests show substantially improved quality and stability. 1. Increase SF range utilization. The spec requires SF delta values to be constrained within the range -60..60. The previous code was applying that range to the whole SF array and not only the deltas of consecutive values, because doing so requires smarter code: zeroing or otherwise skipping a band may invalidate lots of SF choices. This patch implements that logic to allow the coders to utilize the full dynamic range of scalefactors, increasing quality quite considerably, and fixing delta-SF-related assertion failures, since now the limitation is enforced rather than asserted. 2. PNS tweaks The previous modification makes big improvements in twoloop's efficiency, and every time that happens PNS logic needs to be tweaked accordingly to avoid it from stepping all over twoloop's decisions. This patch includes modifications of the sort. 3. Account for lowpass cutoff during PSY analysis The closer PSY's allocation is to final allocation the better the quality is, and given these modifications, twoloop is now very efficient at avoiding holes. Thus, to compute accurate thresholds, PSY needs to account for the lowpass applied implicitly during twoloop (by zeroing high bands). This patch makes twoloop set the cutoff in psymodel's context the first time it runs, and makes PSY account for it during threshold computation, making PE and threshold computations closer to the final allocation and thus achieving better subjective quality. 4. Tweaks to RC lambda tracking loop in relation to PNS Without this tweak some corner cases cause quality regressions. Basically, lambda needs to react faster to overall bitrate efficiency changes since now PNS can be quite successful in enforcing maximum bitrates, when PSY allocates too many bits to the lower bands, suppressing the signals RC logic uses to lower lambda in those cases and causing aggressive PNS. This tweak makes PNS much less aggressive, though it can still use some further tweaks. Also update MIPS specializations and adjust fuzz Also in lavc/mips/aacpsy_mips.h: remove trailing whitespace
2015-12-01 06:28:36 +00:00
if (!sce0->zeroes[w*16+g] && sce0->band_type[w*16+g] < RESERVED_BT)
prev_mid = sce0->sf_idx[w*16+g];
if (!sce1->zeroes[w*16+g] && !cpe->is_mask[w*16+g] && sce1->band_type[w*16+g] < RESERVED_BT)
prev_side = sce1->sf_idx[w*16+g];
start += sce0->ics.swb_sizes[g];
}
}
}
AACCoefficientsEncoder ff_aac_coders[AAC_CODER_NB] = {
[AAC_CODER_ANMR] = {
search_for_quantizers_anmr,
encode_window_bands_info,
quantize_and_encode_band,
ff_aac_encode_tns_info,
ff_aac_encode_ltp_info,
ff_aac_encode_main_pred,
ff_aac_adjust_common_pred,
ff_aac_adjust_common_ltp,
ff_aac_apply_main_pred,
aacenc_tns: rework the way coefficients are calculated This commit abandons the way the specifications state to quantize the coefficients, makes use of the new LPC float functions and is much better. The original way of converting non-normalized float samples to int32_t which out LPC system expects was wrong and it was wrong to assume the coefficients that are generated are also valid. It was essentially a full garbage-in, garbage-out system and it definitely shows when looking at spectrals and listening. The high frequencies were very overattenuated. The new LPC function performs the analysis directly. The specifications state to quantize the coefficients into four bit index values using an asin() function which of course had to have ugly ternary operators because the function turns negative if the coefficients are negative which when encoding causes invalid bitstream to get generated. This deviates from this by using the direct TNS tables, which are fairly small since you only have 4 bits at most for index values. The LPC values are directly quantized against the tables and are then used to perform filtering after the requantization, which simply fetches the array values. The end result is that TNS works much better now and doesn't attenuate anything but the actual signal, e.g. TNS removes quantization errors and does it's job correctly now. It might be enabled by default soon since it doesn't hurt and helps reduce nastyness at low bitrates. Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
2015-08-29 05:47:31 +00:00
ff_aac_apply_tns,
ff_aac_update_ltp,
ff_aac_ltp_insert_new_frame,
set_special_band_scalefactors,
search_for_pns,
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
mark_pns,
ff_aac_search_for_tns,
ff_aac_search_for_ltp,
search_for_ms,
ff_aac_search_for_is,
ff_aac_search_for_pred,
},
[AAC_CODER_TWOLOOP] = {
search_for_quantizers_twoloop,
codebook_trellis_rate,
quantize_and_encode_band,
ff_aac_encode_tns_info,
ff_aac_encode_ltp_info,
ff_aac_encode_main_pred,
ff_aac_adjust_common_pred,
ff_aac_adjust_common_ltp,
ff_aac_apply_main_pred,
aacenc_tns: rework the way coefficients are calculated This commit abandons the way the specifications state to quantize the coefficients, makes use of the new LPC float functions and is much better. The original way of converting non-normalized float samples to int32_t which out LPC system expects was wrong and it was wrong to assume the coefficients that are generated are also valid. It was essentially a full garbage-in, garbage-out system and it definitely shows when looking at spectrals and listening. The high frequencies were very overattenuated. The new LPC function performs the analysis directly. The specifications state to quantize the coefficients into four bit index values using an asin() function which of course had to have ugly ternary operators because the function turns negative if the coefficients are negative which when encoding causes invalid bitstream to get generated. This deviates from this by using the direct TNS tables, which are fairly small since you only have 4 bits at most for index values. The LPC values are directly quantized against the tables and are then used to perform filtering after the requantization, which simply fetches the array values. The end result is that TNS works much better now and doesn't attenuate anything but the actual signal, e.g. TNS removes quantization errors and does it's job correctly now. It might be enabled by default soon since it doesn't hurt and helps reduce nastyness at low bitrates. Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
2015-08-29 05:47:31 +00:00
ff_aac_apply_tns,
ff_aac_update_ltp,
ff_aac_ltp_insert_new_frame,
set_special_band_scalefactors,
search_for_pns,
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
mark_pns,
ff_aac_search_for_tns,
ff_aac_search_for_ltp,
search_for_ms,
ff_aac_search_for_is,
ff_aac_search_for_pred,
},
[AAC_CODER_FAST] = {
search_for_quantizers_fast,
encode_window_bands_info,
quantize_and_encode_band,
ff_aac_encode_tns_info,
ff_aac_encode_ltp_info,
ff_aac_encode_main_pred,
ff_aac_adjust_common_pred,
ff_aac_adjust_common_ltp,
ff_aac_apply_main_pred,
aacenc_tns: rework the way coefficients are calculated This commit abandons the way the specifications state to quantize the coefficients, makes use of the new LPC float functions and is much better. The original way of converting non-normalized float samples to int32_t which out LPC system expects was wrong and it was wrong to assume the coefficients that are generated are also valid. It was essentially a full garbage-in, garbage-out system and it definitely shows when looking at spectrals and listening. The high frequencies were very overattenuated. The new LPC function performs the analysis directly. The specifications state to quantize the coefficients into four bit index values using an asin() function which of course had to have ugly ternary operators because the function turns negative if the coefficients are negative which when encoding causes invalid bitstream to get generated. This deviates from this by using the direct TNS tables, which are fairly small since you only have 4 bits at most for index values. The LPC values are directly quantized against the tables and are then used to perform filtering after the requantization, which simply fetches the array values. The end result is that TNS works much better now and doesn't attenuate anything but the actual signal, e.g. TNS removes quantization errors and does it's job correctly now. It might be enabled by default soon since it doesn't hurt and helps reduce nastyness at low bitrates. Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
2015-08-29 05:47:31 +00:00
ff_aac_apply_tns,
ff_aac_update_ltp,
ff_aac_ltp_insert_new_frame,
set_special_band_scalefactors,
search_for_pns,
AAC encoder: Extensive improvements This finalizes merging of the work in the patches in ticket #2686. Improvements to twoloop and RC logic are extensive. The non-exhaustive list of twoloop improvments includes: - Tweaks to distortion limits on the RD optimization phase of twoloop - Deeper search in twoloop - PNS information marking to let twoloop decide when to use it (turned out having the decision made separately wasn't working) - Tonal band detection and priorization - Better band energy conservation rules - Strict hole avoidance For rate control: - Use psymodel's bit allocation to allow proper use of the bit reservoir. Don't work against the bit reservoir by moving lambda in the opposite direction when psymodel decides to allocate more/less bits to a frame. - Retry the encode if the effective rate lies outside a reasonable margin of psymodel's allocation or the selected ABR. - Log average lambda at the end. Useful info for everyone, but especially for tuning of the various encoder constants that relate to lambda feedback. Psy: - Do not apply lowpass with a FIR filter, instead just let the coder zero bands above the cutoff. The FIR filter induces group delay, and while zeroing bands causes ripple, it's lost in the quantization noise. - Experimental VBR bit allocation code - Tweak automatic lowpass filter threshold to maximize audio bandwidth at all bitrates while still providing acceptable, stable quality. I/S: - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced when the merge was finalized. Measure I/S band energy accounting for phase, and prevent I/S and M/S from being applied both. PNS: - Avoid marking short bands with PNS when they're part of a window group in which there's a large variation of energy from one window to the next. PNS can't preserve those and the effect is extremely noticeable. M/S: - Implement BMLD protection similar to the specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision doesn't conform to section 6.1, a different method had to be implemented, but should provide equivalent protection. - Move the decision logic closer to the method specified in ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically, make sure M/S needs less bits than dual stereo. - Don't apply M/S in bands that are using I/S Now, this of course needed adjustments in the compare targets and fuzz factors of the AAC encoder's fate tests, but if wondering why the targets go up (more distortion), consider the previous coder was using too many bits on LF content (far more than required by psy), and thus those signals will now be more distorted, not less. The extra distortion isn't audible though, I carried extensive ABX testing to make sure. A very similar patch was also extensively tested by Kamendo2 in the context of #2686.
2015-10-11 20:29:50 +00:00
mark_pns,
ff_aac_search_for_tns,
ff_aac_search_for_ltp,
search_for_ms,
ff_aac_search_for_is,
ff_aac_search_for_pred,
},
};