ffmpeg/libavcodec/ac3enc_template.c

363 lines
13 KiB
C
Raw Normal View History

/*
* AC-3 encoder float/fixed template
* Copyright (c) 2000 Fabrice Bellard
* Copyright (c) 2006-2011 Justin Ruggles <justin.ruggles@gmail.com>
* Copyright (c) 2006-2010 Prakash Punnoor <prakash@punnoor.de>
*
* 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
* AC-3 encoder float/fixed template
*/
#include "config_components.h"
#include <stdint.h>
#include "libavutil/attributes.h"
#include "libavutil/mem_internal.h"
#include "audiodsp.h"
#include "ac3enc.h"
#include "eac3enc.h"
#if AC3ENC_FLOAT
#define RENAME(element) element ## _float
#else
#define RENAME(element) element ## _fixed
#endif
/*
* Apply the MDCT to input samples to generate frequency coefficients.
* This applies the KBD window and normalizes the input to reduce precision
* loss due to fixed-point calculations.
*/
static void apply_mdct(AC3EncodeContext *s, uint8_t * const *samples)
{
int blk, ch;
for (ch = 0; ch < s->channels; ch++) {
const SampleType *input_samples0 = (const SampleType*)s->planar_samples[ch];
/* Reorder channels from native order to AC-3 order. */
const SampleType *input_samples1 = (const SampleType*)samples[s->channel_map[ch]];
for (blk = 0; blk < s->num_blocks; blk++) {
AC3Block *block = &s->blocks[blk];
SampleType *windowed_samples = s->RENAME(windowed_samples);
s->fdsp->vector_fmul(windowed_samples, input_samples0,
s->RENAME(mdct_window), AC3_BLOCK_SIZE);
s->fdsp->vector_fmul_reverse(windowed_samples + AC3_BLOCK_SIZE,
input_samples1,
s->RENAME(mdct_window), AC3_BLOCK_SIZE);
2022-10-26 13:12:58 +02:00
s->tx_fn(s->tx, block->mdct_coef[ch+1],
windowed_samples, sizeof(*windowed_samples));
input_samples0 = input_samples1;
input_samples1 += AC3_BLOCK_SIZE;
}
/* Store last 256 samples of current frame */
memcpy(s->planar_samples[ch], input_samples0,
AC3_BLOCK_SIZE * sizeof(*input_samples0));
}
}
/*
* Calculate coupling channel and coupling coordinates.
*/
static void apply_channel_coupling(AC3EncodeContext *s)
{
LOCAL_ALIGNED_32(CoefType, cpl_coords, [AC3_MAX_BLOCKS], [AC3_MAX_CHANNELS][16]);
#if AC3ENC_FLOAT
LOCAL_ALIGNED_32(int32_t, fixed_cpl_coords, [AC3_MAX_BLOCKS], [AC3_MAX_CHANNELS][16]);
#else
int32_t (*fixed_cpl_coords)[AC3_MAX_CHANNELS][16] = cpl_coords;
#endif
int av_uninit(blk), ch, bnd, i, j;
CoefSumType energy[AC3_MAX_BLOCKS][AC3_MAX_CHANNELS][16] = {{{0}}};
int cpl_start, num_cpl_coefs;
memset(cpl_coords, 0, AC3_MAX_BLOCKS * sizeof(*cpl_coords));
#if AC3ENC_FLOAT
memset(fixed_cpl_coords, 0, AC3_MAX_BLOCKS * sizeof(*cpl_coords));
#endif
/* align start to 16-byte boundary. align length to multiple of 32.
note: coupling start bin % 4 will always be 1 */
cpl_start = s->start_freq[CPL_CH] - 1;
num_cpl_coefs = FFALIGN(s->num_cpl_subbands * 12 + 1, 32);
cpl_start = FFMIN(256, cpl_start + num_cpl_coefs) - num_cpl_coefs;
/* calculate coupling channel from fbw channels */
for (blk = 0; blk < s->num_blocks; blk++) {
AC3Block *block = &s->blocks[blk];
CoefType *cpl_coef = &block->mdct_coef[CPL_CH][cpl_start];
if (!block->cpl_in_use)
continue;
memset(cpl_coef, 0, num_cpl_coefs * sizeof(*cpl_coef));
for (ch = 1; ch <= s->fbw_channels; ch++) {
CoefType *ch_coef = &block->mdct_coef[ch][cpl_start];
if (!block->channel_in_cpl[ch])
continue;
for (i = 0; i < num_cpl_coefs; i++)
cpl_coef[i] += ch_coef[i];
}
/* coefficients must be clipped in order to be encoded */
clip_coefficients(&s->adsp, cpl_coef, num_cpl_coefs);
}
/* calculate energy in each band in coupling channel and each fbw channel */
/* TODO: possibly use SIMD to speed up energy calculation */
bnd = 0;
i = s->start_freq[CPL_CH];
while (i < s->cpl_end_freq) {
int band_size = s->cpl_band_sizes[bnd];
for (ch = CPL_CH; ch <= s->fbw_channels; ch++) {
for (blk = 0; blk < s->num_blocks; blk++) {
AC3Block *block = &s->blocks[blk];
if (!block->cpl_in_use || (ch > CPL_CH && !block->channel_in_cpl[ch]))
continue;
for (j = 0; j < band_size; j++) {
CoefType v = block->mdct_coef[ch][i+j];
MAC_COEF(energy[blk][ch][bnd], v, v);
}
}
}
i += band_size;
bnd++;
}
/* calculate coupling coordinates for all blocks for all channels */
for (blk = 0; blk < s->num_blocks; blk++) {
AC3Block *block = &s->blocks[blk];
if (!block->cpl_in_use)
continue;
for (ch = 1; ch <= s->fbw_channels; ch++) {
if (!block->channel_in_cpl[ch])
continue;
for (bnd = 0; bnd < s->num_cpl_bands; bnd++) {
cpl_coords[blk][ch][bnd] = calc_cpl_coord(energy[blk][ch][bnd],
energy[blk][CPL_CH][bnd]);
}
}
}
/* determine which blocks to send new coupling coordinates for */
for (blk = 0; blk < s->num_blocks; blk++) {
AC3Block *block = &s->blocks[blk];
AC3Block *block0 = blk ? &s->blocks[blk-1] : NULL;
memset(block->new_cpl_coords, 0, sizeof(block->new_cpl_coords));
if (block->cpl_in_use) {
/* send new coordinates if this is the first block, if previous
* block did not use coupling but this block does, the channels
* using coupling has changed from the previous block, or the
* coordinate difference from the last block for any channel is
* greater than a threshold value. */
if (blk == 0 || !block0->cpl_in_use) {
for (ch = 1; ch <= s->fbw_channels; ch++)
block->new_cpl_coords[ch] = 1;
} else {
for (ch = 1; ch <= s->fbw_channels; ch++) {
if (!block->channel_in_cpl[ch])
continue;
if (!block0->channel_in_cpl[ch]) {
block->new_cpl_coords[ch] = 1;
} else {
CoefSumType coord_diff = 0;
for (bnd = 0; bnd < s->num_cpl_bands; bnd++) {
coord_diff += FFABS(cpl_coords[blk-1][ch][bnd] -
cpl_coords[blk ][ch][bnd]);
}
coord_diff /= s->num_cpl_bands;
if (coord_diff > NEW_CPL_COORD_THRESHOLD)
block->new_cpl_coords[ch] = 1;
}
}
}
}
}
av_assert1(s->fbw_channels > 0);
/* calculate final coupling coordinates, taking into account reusing of
coordinates in successive blocks */
for (bnd = 0; bnd < s->num_cpl_bands; bnd++) {
blk = 0;
while (blk < s->num_blocks) {
int av_uninit(blk1);
AC3Block *block = &s->blocks[blk];
if (!block->cpl_in_use) {
blk++;
continue;
}
for (ch = 1; ch <= s->fbw_channels; ch++) {
CoefSumType energy_ch, energy_cpl;
if (!block->channel_in_cpl[ch])
continue;
energy_cpl = energy[blk][CPL_CH][bnd];
energy_ch = energy[blk][ch][bnd];
blk1 = blk+1;
while (blk1 < s->num_blocks && !s->blocks[blk1].new_cpl_coords[ch]) {
if (s->blocks[blk1].cpl_in_use) {
energy_cpl += energy[blk1][CPL_CH][bnd];
energy_ch += energy[blk1][ch][bnd];
}
blk1++;
}
cpl_coords[blk][ch][bnd] = calc_cpl_coord(energy_ch, energy_cpl);
}
blk = blk1;
}
}
/* calculate exponents/mantissas for coupling coordinates */
for (blk = 0; blk < s->num_blocks; blk++) {
AC3Block *block = &s->blocks[blk];
if (!block->cpl_in_use)
continue;
#if AC3ENC_FLOAT
s->ac3dsp.float_to_fixed24(fixed_cpl_coords[blk][1],
cpl_coords[blk][1],
s->fbw_channels * 16);
#endif
s->ac3dsp.extract_exponents(block->cpl_coord_exp[1],
fixed_cpl_coords[blk][1],
s->fbw_channels * 16);
for (ch = 1; ch <= s->fbw_channels; ch++) {
int bnd, min_exp, max_exp, master_exp;
if (!block->new_cpl_coords[ch])
continue;
/* determine master exponent */
min_exp = max_exp = block->cpl_coord_exp[ch][0];
for (bnd = 1; bnd < s->num_cpl_bands; bnd++) {
int exp = block->cpl_coord_exp[ch][bnd];
min_exp = FFMIN(exp, min_exp);
max_exp = FFMAX(exp, max_exp);
}
master_exp = ((max_exp - 15) + 2) / 3;
master_exp = FFMAX(master_exp, 0);
while (min_exp < master_exp * 3)
master_exp--;
for (bnd = 0; bnd < s->num_cpl_bands; bnd++) {
block->cpl_coord_exp[ch][bnd] = av_clip(block->cpl_coord_exp[ch][bnd] -
master_exp * 3, 0, 15);
}
block->cpl_master_exp[ch] = master_exp;
/* quantize mantissas */
for (bnd = 0; bnd < s->num_cpl_bands; bnd++) {
int cpl_exp = block->cpl_coord_exp[ch][bnd];
int cpl_mant = (fixed_cpl_coords[blk][ch][bnd] << (5 + cpl_exp + master_exp * 3)) >> 24;
if (cpl_exp == 15)
cpl_mant >>= 1;
else
cpl_mant -= 16;
block->cpl_coord_mant[ch][bnd] = cpl_mant;
}
}
}
if (AC3ENC_FLOAT && CONFIG_EAC3_ENCODER && s->eac3)
ff_eac3_set_cpl_states(s);
}
/*
* Determine rematrixing flags for each block and band.
*/
static void compute_rematrixing_strategy(AC3EncodeContext *s)
{
int nb_coefs;
int blk, bnd;
AC3Block *block, *block0 = NULL;
if (s->channel_mode != AC3_CHMODE_STEREO)
return;
for (blk = 0; blk < s->num_blocks; blk++) {
block = &s->blocks[blk];
block->new_rematrixing_strategy = !blk;
block->num_rematrixing_bands = 4;
if (block->cpl_in_use) {
block->num_rematrixing_bands -= (s->start_freq[CPL_CH] <= 61);
block->num_rematrixing_bands -= (s->start_freq[CPL_CH] == 37);
if (blk && block->num_rematrixing_bands != block0->num_rematrixing_bands)
block->new_rematrixing_strategy = 1;
}
nb_coefs = FFMIN(block->end_freq[1], block->end_freq[2]);
if (!s->rematrixing_enabled) {
block0 = block;
continue;
}
for (bnd = 0; bnd < block->num_rematrixing_bands; bnd++) {
/* calculate sum of squared coeffs for one band in one block */
int start = ff_ac3_rematrix_band_tab[bnd];
int end = FFMIN(nb_coefs, ff_ac3_rematrix_band_tab[bnd+1]);
CoefSumType sum[4];
sum_square_butterfly(s, sum, block->mdct_coef[1] + start,
block->mdct_coef[2] + start, end - start);
/* compare sums to determine if rematrixing will be used for this band */
if (FFMIN(sum[2], sum[3]) < FFMIN(sum[0], sum[1]))
block->rematrixing_flags[bnd] = 1;
else
block->rematrixing_flags[bnd] = 0;
/* determine if new rematrixing flags will be sent */
if (blk &&
block->rematrixing_flags[bnd] != block0->rematrixing_flags[bnd]) {
block->new_rematrixing_strategy = 1;
}
}
block0 = block;
}
}
static void encode_frame(AC3EncodeContext *s, uint8_t * const *samples)
{
apply_mdct(s, samples);
s->cpl_on = s->cpl_enabled;
ff_ac3_compute_coupling_strategy(s);
if (s->cpl_on)
apply_channel_coupling(s);
compute_rematrixing_strategy(s);
ac3enc_fixed: convert to 32-bit sample format The AC3 encoder used to be a separate library called "Aften", which got merged into libavcodec (literally, SVN commits and all). The merge preserved as much features from the library as possible. The code had two versions - a fixed point version and a floating point version. FFmpeg had floating point DSP code used by other codecs, the AC3 decoder including, so the floating-point DSP was simply replaced with FFmpeg's own functions. However, FFmpeg had no fixed-point audio code at that point. So the encoder brought along its own fixed-point DSP functions, including a fixed-point MDCT. The fixed-point MDCT itself is trivially just a float MDCT with a different type and each multiply being a fixed-point multiply. So over time, it got refactored, and the FFT used for all other codecs was templated. Due to design decisions at the time, the fixed-point version of the encoder operates at 16-bits of precision. Although convenient, this, even at the time, was inadequate and inefficient. The encoder is noisy, does not produce output comparable to the float encoder, and even rings at higher frequencies due to the badly approximated winow function. Enter MIPS (owned by Imagination Technologies at the time). They wanted quick fixed-point decoding on their FPUless cores. So they contributed patches to template the AC3 decoder so it had both a fixed-point and a floating-point version. They also did the same for the AAC decoder. They however, used 32-bit samples. Not 16-bits. And we did not have 32-bit fixed-point DSP functions, including an MDCT. But instead of templating our MDCT to output 3 versions (float, 32-bit fixed and 16-bit fixed), they simply copy-pasted their own MDCT into ours, and completely ifdeffed our own MDCT code out if a 32-bit fixed point MDCT was selected. This is also the status quo nowadays - 2 separate MDCTs, one which produces floating point and 16-bit fixed point versions, and one sort-of integrated which produces 32-bit MDCT. MIPS weren't all that interested in encoding, so they left the encoder as-is, and they didn't care much about the ifdeffery, mess or quality - it's not their problem. So the MDCT/FFT code has always been a thorn in anyone looking to clean up code's eye. Backstory over. Internally AC3 operates on 25-bit fixed-point coefficients. So for the floating point version, the encoder simply runs the float MDCT, and converts the resulting coefficients to 25-bit fixed-point, as AC3 is inherently a fixed-point codec. For the fixed-point version, the input is 16-bit samples, so to maximize precision the frame samples are analyzed and the highest set bit is detected via ac3_max_msb_abs_int16(), and the coefficients are then scaled up via ac3_lshift_int16(), so the input for the FFT is always at least 14 bits, computed in normalize_samples(). After FFT, the coefficients are scaled up to 25 bits. This patch simply changes the encoder to accept 32-bit samples, reusing the already well-optimized 32-bit MDCT code, allowing us to clean up and drop a large part of a very messy code of ours, as well as prepare for the future lavu/tx conversion. The coefficients are simply scaled down to 25 bits during windowing, skipping 2 separate scalings, as the hacks to extend precision are simply no longer necessary. There's no point in running the MDCT always at 32 bits when you're going to drop 6 bits off anyway, the headroom is plenty, and the MDCT rounds properly. This also makes the encoder even slightly more accurate over the float version, as there's no coefficient conversion step necessary. SIZE SAVINGS: ARM32: HARDCODED TABLES: BASE - 10709590 DROP DSP - 10702872 - diff: -6.56KiB DROP MDCT - 10667932 - diff: -34.12KiB - both: -40.68KiB DROP FFT - 10336652 - diff: -323.52KiB - all: -364.20KiB SOFTCODED TABLES: BASE - 9685096 DROP DSP - 9678378 - diff: -6.56KiB DROP MDCT - 9643466 - diff: -34.09KiB - both: -40.65KiB DROP FFT - 9573918 - diff: -67.92KiB - all: -108.57KiB ARM64: HARDCODED TABLES: BASE - 14641112 DROP DSP - 14633806 - diff: -7.13KiB DROP MDCT - 14604812 - diff: -28.31KiB - both: -35.45KiB DROP FFT - 14286826 - diff: -310.53KiB - all: -345.98KiB SOFTCODED TABLES: BASE - 13636238 DROP DSP - 13628932 - diff: -7.13KiB DROP MDCT - 13599866 - diff: -28.38KiB - both: -35.52KiB DROP FFT - 13542080 - diff: -56.43KiB - all: -91.95KiB x86: HARDCODED TABLES: BASE - 12367336 DROP DSP - 12354698 - diff: -12.34KiB DROP MDCT - 12331024 - diff: -23.12KiB - both: -35.46KiB DROP FFT - 12029788 - diff: -294.18KiB - all: -329.64KiB SOFTCODED TABLES: BASE - 11358094 DROP DSP - 11345456 - diff: -12.34KiB DROP MDCT - 11321742 - diff: -23.16KiB - both: -35.50KiB DROP FFT - 11276946 - diff: -43.75KiB - all: -79.25KiB PERFORMANCE (10min random s32le): ARM32 - before - 39.9x - 0m15.046s ARM32 - after - 28.2x - 0m21.525s Speed: -30% ARM64 - before - 36.1x - 0m16.637s ARM64 - after - 36.0x - 0m16.727s Speed: -0.5% x86 - before - 184x - 0m3.277s x86 - after - 190x - 0m3.187s Speed: +3%
2021-01-09 01:51:52 +01:00
#if AC3ENC_FLOAT
scale_coefficients(s);
#endif
}