FFmpeg/libavcodec/opus_celt.c
Lynne 6b22e28f4c opusdsp: adjust and optimize C function to match assembly
The C and asm versions behaved differently _outside_ of the codec.

The C version returned pre-multiplied 'state' for the next execution
to use right away, while the assembly version outputted non-multiplied
'state' for the next execution to multiply to save instructions.
Since the initial state when initialized or seeking is always 0,
and since C and asm versions were never mixed, there was no issue.

However, comparing outputs directly in checkasm doesn't work without
dividing the initial state by CELT_EMPH_COEFF and multiplying the
returned state by CELT_EMPH_COEFF for the assembly function.

Since its actually faster to do this in C as well, copy the behavior the
asm versions use. As a reminder, the initial state 0 is divided by
CELT_EMPH_COEFF on seek and init (just in case in the future this is
changed, its technically more correct to init with CELT_EMPH_COEFF than 0,
however when seeking this will result in more audiable pops, unlike with 0
where the output gets in sync over a few samples).
2019-09-11 03:28:22 +01:00

581 lines
18 KiB
C

/*
* Copyright (c) 2012 Andrew D'Addesio
* Copyright (c) 2013-2014 Mozilla Corporation
* Copyright (c) 2016 Rostislav Pehlivanov <atomnuker@gmail.com>
*
* 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
* Opus CELT decoder
*/
#include "opus_celt.h"
#include "opustab.h"
#include "opus_pvq.h"
/* Use the 2D z-transform to apply prediction in both the time domain (alpha)
* and the frequency domain (beta) */
static void celt_decode_coarse_energy(CeltFrame *f, OpusRangeCoder *rc)
{
int i, j;
float prev[2] = { 0 };
float alpha = ff_celt_alpha_coef[f->size];
float beta = ff_celt_beta_coef[f->size];
const uint8_t *model = ff_celt_coarse_energy_dist[f->size][0];
/* intra frame */
if (opus_rc_tell(rc) + 3 <= f->framebits && ff_opus_rc_dec_log(rc, 3)) {
alpha = 0.0f;
beta = 1.0f - (4915.0f/32768.0f);
model = ff_celt_coarse_energy_dist[f->size][1];
}
for (i = 0; i < CELT_MAX_BANDS; i++) {
for (j = 0; j < f->channels; j++) {
CeltBlock *block = &f->block[j];
float value;
int available;
if (i < f->start_band || i >= f->end_band) {
block->energy[i] = 0.0;
continue;
}
available = f->framebits - opus_rc_tell(rc);
if (available >= 15) {
/* decode using a Laplace distribution */
int k = FFMIN(i, 20) << 1;
value = ff_opus_rc_dec_laplace(rc, model[k] << 7, model[k+1] << 6);
} else if (available >= 2) {
int x = ff_opus_rc_dec_cdf(rc, ff_celt_model_energy_small);
value = (x>>1) ^ -(x&1);
} else if (available >= 1) {
value = -(float)ff_opus_rc_dec_log(rc, 1);
} else value = -1;
block->energy[i] = FFMAX(-9.0f, block->energy[i]) * alpha + prev[j] + value;
prev[j] += beta * value;
}
}
}
static void celt_decode_fine_energy(CeltFrame *f, OpusRangeCoder *rc)
{
int i;
for (i = f->start_band; i < f->end_band; i++) {
int j;
if (!f->fine_bits[i])
continue;
for (j = 0; j < f->channels; j++) {
CeltBlock *block = &f->block[j];
int q2;
float offset;
q2 = ff_opus_rc_get_raw(rc, f->fine_bits[i]);
offset = (q2 + 0.5f) * (1 << (14 - f->fine_bits[i])) / 16384.0f - 0.5f;
block->energy[i] += offset;
}
}
}
static void celt_decode_final_energy(CeltFrame *f, OpusRangeCoder *rc)
{
int priority, i, j;
int bits_left = f->framebits - opus_rc_tell(rc);
for (priority = 0; priority < 2; priority++) {
for (i = f->start_band; i < f->end_band && bits_left >= f->channels; i++) {
if (f->fine_priority[i] != priority || f->fine_bits[i] >= CELT_MAX_FINE_BITS)
continue;
for (j = 0; j < f->channels; j++) {
int q2;
float offset;
q2 = ff_opus_rc_get_raw(rc, 1);
offset = (q2 - 0.5f) * (1 << (14 - f->fine_bits[i] - 1)) / 16384.0f;
f->block[j].energy[i] += offset;
bits_left--;
}
}
}
}
static void celt_decode_tf_changes(CeltFrame *f, OpusRangeCoder *rc)
{
int i, diff = 0, tf_select = 0, tf_changed = 0, tf_select_bit;
int consumed, bits = f->transient ? 2 : 4;
consumed = opus_rc_tell(rc);
tf_select_bit = (f->size != 0 && consumed+bits+1 <= f->framebits);
for (i = f->start_band; i < f->end_band; i++) {
if (consumed+bits+tf_select_bit <= f->framebits) {
diff ^= ff_opus_rc_dec_log(rc, bits);
consumed = opus_rc_tell(rc);
tf_changed |= diff;
}
f->tf_change[i] = diff;
bits = f->transient ? 4 : 5;
}
if (tf_select_bit && ff_celt_tf_select[f->size][f->transient][0][tf_changed] !=
ff_celt_tf_select[f->size][f->transient][1][tf_changed])
tf_select = ff_opus_rc_dec_log(rc, 1);
for (i = f->start_band; i < f->end_band; i++) {
f->tf_change[i] = ff_celt_tf_select[f->size][f->transient][tf_select][f->tf_change[i]];
}
}
static void celt_denormalize(CeltFrame *f, CeltBlock *block, float *data)
{
int i, j;
for (i = f->start_band; i < f->end_band; i++) {
float *dst = data + (ff_celt_freq_bands[i] << f->size);
float log_norm = block->energy[i] + ff_celt_mean_energy[i];
float norm = exp2f(FFMIN(log_norm, 32.0f));
for (j = 0; j < ff_celt_freq_range[i] << f->size; j++)
dst[j] *= norm;
}
}
static void celt_postfilter_apply_transition(CeltBlock *block, float *data)
{
const int T0 = block->pf_period_old;
const int T1 = block->pf_period;
float g00, g01, g02;
float g10, g11, g12;
float x0, x1, x2, x3, x4;
int i;
if (block->pf_gains[0] == 0.0 &&
block->pf_gains_old[0] == 0.0)
return;
g00 = block->pf_gains_old[0];
g01 = block->pf_gains_old[1];
g02 = block->pf_gains_old[2];
g10 = block->pf_gains[0];
g11 = block->pf_gains[1];
g12 = block->pf_gains[2];
x1 = data[-T1 + 1];
x2 = data[-T1];
x3 = data[-T1 - 1];
x4 = data[-T1 - 2];
for (i = 0; i < CELT_OVERLAP; i++) {
float w = ff_celt_window2[i];
x0 = data[i - T1 + 2];
data[i] += (1.0 - w) * g00 * data[i - T0] +
(1.0 - w) * g01 * (data[i - T0 - 1] + data[i - T0 + 1]) +
(1.0 - w) * g02 * (data[i - T0 - 2] + data[i - T0 + 2]) +
w * g10 * x2 +
w * g11 * (x1 + x3) +
w * g12 * (x0 + x4);
x4 = x3;
x3 = x2;
x2 = x1;
x1 = x0;
}
}
static void celt_postfilter(CeltFrame *f, CeltBlock *block)
{
int len = f->blocksize * f->blocks;
const int filter_len = len - 2 * CELT_OVERLAP;
celt_postfilter_apply_transition(block, block->buf + 1024);
block->pf_period_old = block->pf_period;
memcpy(block->pf_gains_old, block->pf_gains, sizeof(block->pf_gains));
block->pf_period = block->pf_period_new;
memcpy(block->pf_gains, block->pf_gains_new, sizeof(block->pf_gains));
if (len > CELT_OVERLAP) {
celt_postfilter_apply_transition(block, block->buf + 1024 + CELT_OVERLAP);
if (block->pf_gains[0] > FLT_EPSILON && filter_len > 0)
f->opusdsp.postfilter(block->buf + 1024 + 2 * CELT_OVERLAP,
block->pf_period, block->pf_gains,
filter_len);
block->pf_period_old = block->pf_period;
memcpy(block->pf_gains_old, block->pf_gains, sizeof(block->pf_gains));
}
memmove(block->buf, block->buf + len, (1024 + CELT_OVERLAP / 2) * sizeof(float));
}
static int parse_postfilter(CeltFrame *f, OpusRangeCoder *rc, int consumed)
{
int i;
memset(f->block[0].pf_gains_new, 0, sizeof(f->block[0].pf_gains_new));
memset(f->block[1].pf_gains_new, 0, sizeof(f->block[1].pf_gains_new));
if (f->start_band == 0 && consumed + 16 <= f->framebits) {
int has_postfilter = ff_opus_rc_dec_log(rc, 1);
if (has_postfilter) {
float gain;
int tapset, octave, period;
octave = ff_opus_rc_dec_uint(rc, 6);
period = (16 << octave) + ff_opus_rc_get_raw(rc, 4 + octave) - 1;
gain = 0.09375f * (ff_opus_rc_get_raw(rc, 3) + 1);
tapset = (opus_rc_tell(rc) + 2 <= f->framebits) ?
ff_opus_rc_dec_cdf(rc, ff_celt_model_tapset) : 0;
for (i = 0; i < 2; i++) {
CeltBlock *block = &f->block[i];
block->pf_period_new = FFMAX(period, CELT_POSTFILTER_MINPERIOD);
block->pf_gains_new[0] = gain * ff_celt_postfilter_taps[tapset][0];
block->pf_gains_new[1] = gain * ff_celt_postfilter_taps[tapset][1];
block->pf_gains_new[2] = gain * ff_celt_postfilter_taps[tapset][2];
}
}
consumed = opus_rc_tell(rc);
}
return consumed;
}
static void process_anticollapse(CeltFrame *f, CeltBlock *block, float *X)
{
int i, j, k;
for (i = f->start_band; i < f->end_band; i++) {
int renormalize = 0;
float *xptr;
float prev[2];
float Ediff, r;
float thresh, sqrt_1;
int depth;
/* depth in 1/8 bits */
depth = (1 + f->pulses[i]) / (ff_celt_freq_range[i] << f->size);
thresh = exp2f(-1.0 - 0.125f * depth);
sqrt_1 = 1.0f / sqrtf(ff_celt_freq_range[i] << f->size);
xptr = X + (ff_celt_freq_bands[i] << f->size);
prev[0] = block->prev_energy[0][i];
prev[1] = block->prev_energy[1][i];
if (f->channels == 1) {
CeltBlock *block1 = &f->block[1];
prev[0] = FFMAX(prev[0], block1->prev_energy[0][i]);
prev[1] = FFMAX(prev[1], block1->prev_energy[1][i]);
}
Ediff = block->energy[i] - FFMIN(prev[0], prev[1]);
Ediff = FFMAX(0, Ediff);
/* r needs to be multiplied by 2 or 2*sqrt(2) depending on LM because
short blocks don't have the same energy as long */
r = exp2f(1 - Ediff);
if (f->size == 3)
r *= M_SQRT2;
r = FFMIN(thresh, r) * sqrt_1;
for (k = 0; k < 1 << f->size; k++) {
/* Detect collapse */
if (!(block->collapse_masks[i] & 1 << k)) {
/* Fill with noise */
for (j = 0; j < ff_celt_freq_range[i]; j++)
xptr[(j << f->size) + k] = (celt_rng(f) & 0x8000) ? r : -r;
renormalize = 1;
}
}
/* We just added some energy, so we need to renormalize */
if (renormalize)
celt_renormalize_vector(xptr, ff_celt_freq_range[i] << f->size, 1.0f);
}
}
int ff_celt_decode_frame(CeltFrame *f, OpusRangeCoder *rc,
float **output, int channels, int frame_size,
int start_band, int end_band)
{
int i, j, downmix = 0;
int consumed; // bits of entropy consumed thus far for this frame
MDCT15Context *imdct;
if (channels != 1 && channels != 2) {
av_log(f->avctx, AV_LOG_ERROR, "Invalid number of coded channels: %d\n",
channels);
return AVERROR_INVALIDDATA;
}
if (start_band < 0 || start_band > end_band || end_band > CELT_MAX_BANDS) {
av_log(f->avctx, AV_LOG_ERROR, "Invalid start/end band: %d %d\n",
start_band, end_band);
return AVERROR_INVALIDDATA;
}
f->silence = 0;
f->transient = 0;
f->anticollapse = 0;
f->flushed = 0;
f->channels = channels;
f->start_band = start_band;
f->end_band = end_band;
f->framebits = rc->rb.bytes * 8;
f->size = av_log2(frame_size / CELT_SHORT_BLOCKSIZE);
if (f->size > CELT_MAX_LOG_BLOCKS ||
frame_size != CELT_SHORT_BLOCKSIZE * (1 << f->size)) {
av_log(f->avctx, AV_LOG_ERROR, "Invalid CELT frame size: %d\n",
frame_size);
return AVERROR_INVALIDDATA;
}
if (!f->output_channels)
f->output_channels = channels;
for (i = 0; i < f->channels; i++) {
memset(f->block[i].coeffs, 0, sizeof(f->block[i].coeffs));
memset(f->block[i].collapse_masks, 0, sizeof(f->block[i].collapse_masks));
}
consumed = opus_rc_tell(rc);
/* obtain silence flag */
if (consumed >= f->framebits)
f->silence = 1;
else if (consumed == 1)
f->silence = ff_opus_rc_dec_log(rc, 15);
if (f->silence) {
consumed = f->framebits;
rc->total_bits += f->framebits - opus_rc_tell(rc);
}
/* obtain post-filter options */
consumed = parse_postfilter(f, rc, consumed);
/* obtain transient flag */
if (f->size != 0 && consumed+3 <= f->framebits)
f->transient = ff_opus_rc_dec_log(rc, 3);
f->blocks = f->transient ? 1 << f->size : 1;
f->blocksize = frame_size / f->blocks;
imdct = f->imdct[f->transient ? 0 : f->size];
if (channels == 1) {
for (i = 0; i < CELT_MAX_BANDS; i++)
f->block[0].energy[i] = FFMAX(f->block[0].energy[i], f->block[1].energy[i]);
}
celt_decode_coarse_energy(f, rc);
celt_decode_tf_changes (f, rc);
ff_celt_bitalloc (f, rc, 0);
celt_decode_fine_energy (f, rc);
ff_celt_quant_bands (f, rc);
if (f->anticollapse_needed)
f->anticollapse = ff_opus_rc_get_raw(rc, 1);
celt_decode_final_energy(f, rc);
/* apply anti-collapse processing and denormalization to
* each coded channel */
for (i = 0; i < f->channels; i++) {
CeltBlock *block = &f->block[i];
if (f->anticollapse)
process_anticollapse(f, block, f->block[i].coeffs);
celt_denormalize(f, block, f->block[i].coeffs);
}
/* stereo -> mono downmix */
if (f->output_channels < f->channels) {
f->dsp->vector_fmac_scalar(f->block[0].coeffs, f->block[1].coeffs, 1.0, FFALIGN(frame_size, 16));
downmix = 1;
} else if (f->output_channels > f->channels)
memcpy(f->block[1].coeffs, f->block[0].coeffs, frame_size * sizeof(float));
if (f->silence) {
for (i = 0; i < 2; i++) {
CeltBlock *block = &f->block[i];
for (j = 0; j < FF_ARRAY_ELEMS(block->energy); j++)
block->energy[j] = CELT_ENERGY_SILENCE;
}
memset(f->block[0].coeffs, 0, sizeof(f->block[0].coeffs));
memset(f->block[1].coeffs, 0, sizeof(f->block[1].coeffs));
}
/* transform and output for each output channel */
for (i = 0; i < f->output_channels; i++) {
CeltBlock *block = &f->block[i];
/* iMDCT and overlap-add */
for (j = 0; j < f->blocks; j++) {
float *dst = block->buf + 1024 + j * f->blocksize;
imdct->imdct_half(imdct, dst + CELT_OVERLAP / 2, f->block[i].coeffs + j,
f->blocks);
f->dsp->vector_fmul_window(dst, dst, dst + CELT_OVERLAP / 2,
ff_celt_window, CELT_OVERLAP / 2);
}
if (downmix)
f->dsp->vector_fmul_scalar(&block->buf[1024], &block->buf[1024], 0.5f, frame_size);
/* postfilter */
celt_postfilter(f, block);
/* deemphasis */
block->emph_coeff = f->opusdsp.deemphasis(output[i],
&block->buf[1024 - frame_size],
block->emph_coeff, frame_size);
}
if (channels == 1)
memcpy(f->block[1].energy, f->block[0].energy, sizeof(f->block[0].energy));
for (i = 0; i < 2; i++ ) {
CeltBlock *block = &f->block[i];
if (!f->transient) {
memcpy(block->prev_energy[1], block->prev_energy[0], sizeof(block->prev_energy[0]));
memcpy(block->prev_energy[0], block->energy, sizeof(block->prev_energy[0]));
} else {
for (j = 0; j < CELT_MAX_BANDS; j++)
block->prev_energy[0][j] = FFMIN(block->prev_energy[0][j], block->energy[j]);
}
for (j = 0; j < f->start_band; j++) {
block->prev_energy[0][j] = CELT_ENERGY_SILENCE;
block->energy[j] = 0.0;
}
for (j = f->end_band; j < CELT_MAX_BANDS; j++) {
block->prev_energy[0][j] = CELT_ENERGY_SILENCE;
block->energy[j] = 0.0;
}
}
f->seed = rc->range;
return 0;
}
void ff_celt_flush(CeltFrame *f)
{
int i, j;
if (f->flushed)
return;
for (i = 0; i < 2; i++) {
CeltBlock *block = &f->block[i];
for (j = 0; j < CELT_MAX_BANDS; j++)
block->prev_energy[0][j] = block->prev_energy[1][j] = CELT_ENERGY_SILENCE;
memset(block->energy, 0, sizeof(block->energy));
memset(block->buf, 0, sizeof(block->buf));
memset(block->pf_gains, 0, sizeof(block->pf_gains));
memset(block->pf_gains_old, 0, sizeof(block->pf_gains_old));
memset(block->pf_gains_new, 0, sizeof(block->pf_gains_new));
/* libopus uses CELT_EMPH_COEFF on init, but 0 is better since there's
* a lesser discontinuity when seeking.
* The deemphasis functions differ from libopus in that they require
* an initial state divided by the coefficient. */
block->emph_coeff = 0.0f / CELT_EMPH_COEFF;
}
f->seed = 0;
f->flushed = 1;
}
void ff_celt_free(CeltFrame **f)
{
CeltFrame *frm = *f;
int i;
if (!frm)
return;
for (i = 0; i < FF_ARRAY_ELEMS(frm->imdct); i++)
ff_mdct15_uninit(&frm->imdct[i]);
ff_celt_pvq_uninit(&frm->pvq);
av_freep(&frm->dsp);
av_freep(f);
}
int ff_celt_init(AVCodecContext *avctx, CeltFrame **f, int output_channels,
int apply_phase_inv)
{
CeltFrame *frm;
int i, ret;
if (output_channels != 1 && output_channels != 2) {
av_log(avctx, AV_LOG_ERROR, "Invalid number of output channels: %d\n",
output_channels);
return AVERROR(EINVAL);
}
frm = av_mallocz(sizeof(*frm));
if (!frm)
return AVERROR(ENOMEM);
frm->avctx = avctx;
frm->output_channels = output_channels;
frm->apply_phase_inv = apply_phase_inv;
for (i = 0; i < FF_ARRAY_ELEMS(frm->imdct); i++)
if ((ret = ff_mdct15_init(&frm->imdct[i], 1, i + 3, -1.0f/32768)) < 0)
goto fail;
if ((ret = ff_celt_pvq_init(&frm->pvq, 0)) < 0)
goto fail;
frm->dsp = avpriv_float_dsp_alloc(avctx->flags & AV_CODEC_FLAG_BITEXACT);
if (!frm->dsp) {
ret = AVERROR(ENOMEM);
goto fail;
}
ff_opus_dsp_init(&frm->opusdsp);
ff_celt_flush(frm);
*f = frm;
return 0;
fail:
ff_celt_free(&frm);
return ret;
}