virtualenv/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-11-26 19:01:44 +02:00
Code Issues Releases Activity
e6954fd087
FFmpeg/libavcodec/aacenc_ltp.c

237 lines
8.3 KiB
C
Raw Normal View History

aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
/*
aacenc_ltp: correct header description comment
2015-10-17 14:28:55 +02:00
* AAC encoder long term prediction extension
aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
* Copyright (C) 2015 Rostislav Pehlivanov
*
* 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
aacenc_ltp: correct header description comment
2015-10-17 14:28:55 +02:00
* AAC encoder long term prediction extension
aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
* @author Rostislav Pehlivanov ( atomnuker gmail com )
*/
#include "aacenc_ltp.h"
#include "aacenc_quantization.h"
#include "aacenc_utils.h"
/**
* Encode LTP data.
*/
void ff_aac_encode_ltp_info(AACEncContext *s, SingleChannelElement *sce,
int common_window)
{
int i;
IndividualChannelStream *ics = &sce->ics;
if (s->profile != FF_PROFILE_AAC_LTP || !ics->predictor_present)
return;
if (common_window)
put_bits(&s->pb, 1, 0);
put_bits(&s->pb, 1, ics->ltp.present);
if (!ics->ltp.present)
return;
put_bits(&s->pb, 11, ics->ltp.lag);
put_bits(&s->pb, 3, ics->ltp.coef_idx);
for (i = 0; i < FFMIN(ics->max_sfb, MAX_LTP_LONG_SFB); i++)
put_bits(&s->pb, 1, ics->ltp.used[i]);
}
void ff_aac_ltp_insert_new_frame(AACEncContext *s)
{
int i, ch, tag, chans, cur_channel, start_ch = 0;
ChannelElement *cpe;
SingleChannelElement *sce;
for (i = 0; i < s->chan_map[0]; i++) {
cpe = &s->cpe[i];
tag = s->chan_map[i+1];
chans = tag == TYPE_CPE ? 2 : 1;
for (ch = 0; ch < chans; ch++) {
sce = &cpe->ch[ch];
cur_channel = start_ch + ch;
/* New sample + overlap */
memcpy(&sce->ltp_state[0], &sce->ltp_state[1024], 1024*sizeof(sce->ltp_state[0]));
memcpy(&sce->ltp_state[1024], &s->planar_samples[cur_channel][2048], 1024*sizeof(sce->ltp_state[0]));
memcpy(&sce->ltp_state[2048], &sce->ret_buf[0], 1024*sizeof(sce->ltp_state[0]));
aacenc_ltp: adjust and speed up autocorrelation calculations There were some errors in the calculation as well as an entire unnecessary loop to find the gain coefficient. Merge the two loops. Thanks to @ubitux for the suggestions and testing.
2015-10-17 23:50:55 +02:00
sce->ics.ltp.lag = 0;
aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
}
start_ch += chans;
}
}
aac_ltp: split, reorder and improve prediction algorithm This commit attempts to mirror what the decoder does more closely in addition to fixing some shortcomings.
2015-11-26 19:40:04 +02:00
static void get_lag(float *buf, const float *new, LongTermPrediction *ltp)
aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
{
avcodec/aacenc_ltp: fix use of uninitialized values Fixes some valgrind warnings. Reviewed-by: Rostislav Pehlivanov <atomnuker@gmail.com> Signed-off-by: James Almer <jamrial@gmail.com>
2017-04-13 00:38:17 +02:00
int i, j, lag = 0, max_corr = 0;
float max_ratio = 0.0f;
aacenc_ltp: adjust and speed up autocorrelation calculations There were some errors in the calculation as well as an entire unnecessary loop to find the gain coefficient. Merge the two loops. Thanks to @ubitux for the suggestions and testing.
2015-10-17 23:50:55 +02:00
for (i = 0; i < 2048; i++) {
aac_ltp: split, reorder and improve prediction algorithm This commit attempts to mirror what the decoder does more closely in addition to fixing some shortcomings.
2015-11-26 19:40:04 +02:00
float corr, s0 = 0.0f, s1 = 0.0f;
aacenc_ltp: adjust and speed up autocorrelation calculations There were some errors in the calculation as well as an entire unnecessary loop to find the gain coefficient. Merge the two loops. Thanks to @ubitux for the suggestions and testing.
2015-10-17 23:50:55 +02:00
const int start = FFMAX(0, i - 1024);
for (j = start; j < 2048; j++) {
const int idx = j - i + 1024;
aac_ltp: split, reorder and improve prediction algorithm This commit attempts to mirror what the decoder does more closely in addition to fixing some shortcomings.
2015-11-26 19:40:04 +02:00
s0 += new[j]*buf[idx];
s1 += buf[idx]*buf[idx];
aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
}
corr = s1 > 0.0f ? s0/sqrt(s1) : 0.0f;
if (corr > max_corr) {
max_corr = corr;
lag = i;
aacenc_ltp: adjust and speed up autocorrelation calculations There were some errors in the calculation as well as an entire unnecessary loop to find the gain coefficient. Merge the two loops. Thanks to @ubitux for the suggestions and testing.
2015-10-17 23:50:55 +02:00
max_ratio = corr/(2048-start);
aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
}
}
aac_ltp: split, reorder and improve prediction algorithm This commit attempts to mirror what the decoder does more closely in addition to fixing some shortcomings.
2015-11-26 19:40:04 +02:00
ltp->lag = FFMAX(av_clip_uintp2(lag, 11), 0);
ltp->coef_idx = quant_array_idx(max_ratio, ltp_coef, 8);
ltp->coef = ltp_coef[ltp->coef_idx];
}
aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
aac_ltp: split, reorder and improve prediction algorithm This commit attempts to mirror what the decoder does more closely in addition to fixing some shortcomings.
2015-11-26 19:40:04 +02:00
static void generate_samples(float *buf, LongTermPrediction *ltp)
{
int i, samples_num = 2048;
if (!ltp->lag) {
ltp->present = 0;
aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
return;
aac_ltp: split, reorder and improve prediction algorithm This commit attempts to mirror what the decoder does more closely in addition to fixing some shortcomings.
2015-11-26 19:40:04 +02:00
} else if (ltp->lag < 1024) {
samples_num = ltp->lag + 1024;
}
for (i = 0; i < samples_num; i++)
buf[i] = ltp->coef*buf[i + 2048 - ltp->lag];
memset(&buf[i], 0, (2048 - i)*sizeof(float));
}
/**
* Process LTP parameters
* @see Patent WO2006070265A1
*/
void ff_aac_update_ltp(AACEncContext *s, SingleChannelElement *sce)
{
float *pred_signal = &sce->ltp_state[0];
const float *samples = &s->planar_samples[s->cur_channel][1024];
aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
aac_ltp: split, reorder and improve prediction algorithm This commit attempts to mirror what the decoder does more closely in addition to fixing some shortcomings.
2015-11-26 19:40:04 +02:00
if (s->profile != FF_PROFILE_AAC_LTP)
return;
aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
aac_ltp: split, reorder and improve prediction algorithm This commit attempts to mirror what the decoder does more closely in addition to fixing some shortcomings.
2015-11-26 19:40:04 +02:00
/* Calculate lag */
get_lag(pred_signal, samples, &sce->ics.ltp);
generate_samples(pred_signal, &sce->ics.ltp);
aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
}
void ff_aac_adjust_common_ltp(AACEncContext *s, ChannelElement *cpe)
{
int sfb, count = 0;
SingleChannelElement *sce0 = &cpe->ch[0];
SingleChannelElement *sce1 = &cpe->ch[1];
if (!cpe->common_window ||
sce0->ics.window_sequence[0] == EIGHT_SHORT_SEQUENCE ||
aac_ltp: actually signal LTP as off during EIGHT_SHORT windows This hugely reduces the echo which was introduced with the previous commit (though likely because previously everything was broken). Makes LTP actually worthwhile now. Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
2015-11-26 20:20:42 +02:00
sce1->ics.window_sequence[0] == EIGHT_SHORT_SEQUENCE) {
sce0->ics.ltp.present = 0;
aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
return;
aac_ltp: actually signal LTP as off during EIGHT_SHORT windows This hugely reduces the echo which was introduced with the previous commit (though likely because previously everything was broken). Makes LTP actually worthwhile now. Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
2015-11-26 20:20:42 +02:00
}
aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
for (sfb = 0; sfb < FFMIN(sce0->ics.max_sfb, MAX_LTP_LONG_SFB); sfb++) {
int sum = sce0->ics.ltp.used[sfb] + sce1->ics.ltp.used[sfb];
if (sum != 2) {
sce0->ics.ltp.used[sfb] = 0;
lavc/aacenc_ltp: remove unnecessary condition check. Condition 'sum==2' is always true, so remove the check logic to make the code clean. Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
2019-05-10 09:49:45 +02:00
} else {
aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
count++;
}
}
sce0->ics.ltp.present = !!count;
sce0->ics.predictor_present = !!count;
}
/**
* Mark LTP sfb's
*/
void ff_aac_search_for_ltp(AACEncContext *s, SingleChannelElement *sce,
int common_window)
{
int w, g, w2, i, start = 0, count = 0;
int saved_bits = -(15 + FFMIN(sce->ics.max_sfb, MAX_LTP_LONG_SFB));
float *C34 = &s->scoefs[128*0], *PCD = &s->scoefs[128*1];
float *PCD34 = &s->scoefs[128*2];
const int max_ltp = FFMIN(sce->ics.max_sfb, MAX_LTP_LONG_SFB);
aacenc_ltp: adjust and speed up autocorrelation calculations There were some errors in the calculation as well as an entire unnecessary loop to find the gain coefficient. Merge the two loops. Thanks to @ubitux for the suggestions and testing.
2015-10-17 23:50:55 +02:00
if (sce->ics.window_sequence[0] == EIGHT_SHORT_SEQUENCE) {
if (sce->ics.ltp.lag) {
aacenc_ltp: fix out of bounds memory access Discovered by Coverity.
2015-12-04 01:12:37 +02:00
memset(&sce->ltp_state[0], 0, 3072*sizeof(sce->ltp_state[0]));
aacenc_ltp: adjust and speed up autocorrelation calculations There were some errors in the calculation as well as an entire unnecessary loop to find the gain coefficient. Merge the two loops. Thanks to @ubitux for the suggestions and testing.
2015-10-17 23:50:55 +02:00
memset(&sce->ics.ltp, 0, sizeof(LongTermPrediction));
}
return;
}
aacenc_ltp: disable LTP with high lambda values Makes no sense to enable for high bitrates, the coder does well enough. Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
2015-12-08 15:24:48 +02:00
if (!sce->ics.ltp.lag || s->lambda > 120.0f)
aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
return;
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
start = 0;
for (g = 0; g < sce->ics.num_swb; g++) {
int bits1 = 0, bits2 = 0;
float dist1 = 0.0f, dist2 = 0.0f;
if (w*16+g > max_ltp) {
start += sce->ics.swb_sizes[g];
continue;
}
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
int bits_tmp1, bits_tmp2;
FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g];
for (i = 0; i < sce->ics.swb_sizes[g]; i++)
PCD[i] = sce->coeffs[start+(w+w2)*128+i] - sce->lcoeffs[start+(w+w2)*128+i];
aacenc: add SIMD optimizations for abs_pow34 and quantization Performance improvements: quant_bands: with: 681 decicycles in quant_bands, 8388453 runs, 155 skips without: 1190 decicycles in quant_bands, 8388386 runs, 222 skips Around 42% for the function Twoloop coder: abs_pow34: with/without: 7.82s/8.17s Around 4% for the entire encoder Both: with/without: 7.15s/8.17s Around 12% for the entire encoder Fast coder: abs_pow34: with/without: 3.40s/3.77s Around 10% for the entire encoder Both: with/without: 3.02s/3.77s Around 20% faster for the entire encoder Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com> Tested-by: Michael Niedermayer <michael@niedermayer.cc> Reviewed-by: James Almer <jamrial@gmail.com>
2016-10-08 16:59:14 +02:00
s->abs_pow34(C34, &sce->coeffs[start+(w+w2)*128], sce->ics.swb_sizes[g]);
s->abs_pow34(PCD34, PCD, sce->ics.swb_sizes[g]);
aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
dist1 += quantize_band_cost(s, &sce->coeffs[start+(w+w2)*128], C34, sce->ics.swb_sizes[g],
sce->sf_idx[(w+w2)*16+g], sce->band_type[(w+w2)*16+g],
avcodec/aacenc_quantization: Remove always-zero function parameter rtz is only ever nonzero for quantize_and_encode_band(). Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-08-01 09:45:10 +02:00
s->lambda/band->threshold, INFINITY, &bits_tmp1, NULL);
aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
dist2 += quantize_band_cost(s, PCD, PCD34, sce->ics.swb_sizes[g],
sce->sf_idx[(w+w2)*16+g],
sce->band_type[(w+w2)*16+g],
avcodec/aacenc_quantization: Remove always-zero function parameter rtz is only ever nonzero for quantize_and_encode_band(). Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-08-01 09:45:10 +02:00
s->lambda/band->threshold, INFINITY, &bits_tmp2, NULL);
aacenc: add support for encoding files using Long Term Prediction Long Term Prediction allows for prediction of spectral coefficients via the previously decoded time-dependent samples. This feature works well with harmonic content 2 or more frames long, like speech, human or non-human, piano music or any constant tones at very low bitrates. It should be noted that the current coder is highly efficient and the rate control system is unable to encode files at extremely low bitrates (less than 14kbps seems to be impossible) so this extension isn't capable of optimum operation. Dramatic difference is observable with some types of audio and speech but for the most part the audiable differences are subtle. The spectrum looks better however so the encoder is able to harvest the additional bits that this feature provies, should the user choose to enable it. So it's best to enable this feature only if encoding at the absolutely lowest bitrate that the encoder is capable of.
2015-10-17 03:22:51 +02:00
bits1 += bits_tmp1;
bits2 += bits_tmp2;
}
if (dist2 < dist1 && bits2 < bits1) {
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++)
for (i = 0; i < sce->ics.swb_sizes[g]; i++)
sce->coeffs[start+(w+w2)*128+i] -= sce->lcoeffs[start+(w+w2)*128+i];
sce->ics.ltp.used[w*16+g] = 1;
saved_bits += bits1 - bits2;
count++;
}
start += sce->ics.swb_sizes[g];
}
}
sce->ics.ltp.present = !!count && (saved_bits >= 0);
sce->ics.predictor_present = !!sce->ics.ltp.present;
/* Reset any marked sfbs */
if (!sce->ics.ltp.present && !!count) {
for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
start = 0;
for (g = 0; g < sce->ics.num_swb; g++) {
if (sce->ics.ltp.used[w*16+g]) {
for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) {
for (i = 0; i < sce->ics.swb_sizes[g]; i++) {
sce->coeffs[start+(w+w2)*128+i] += sce->lcoeffs[start+(w+w2)*128+i];
}
}
}
start += sce->ics.swb_sizes[g];
}
}
}
}
Reference in New Issue Copy Permalink