Newer
Older
/*
* AAC decoder
* Copyright (c) 2005-2006 Oded Shimon ( ods15 ods15 dyndns org )
* Copyright (c) 2006-2007 Maxim Gavrilov ( maxim.gavrilov gmail com )
*
* AAC LATM decoder
* Copyright (c) 2008-2010 Paul Kendall <paul@kcbbs.gen.nz>
* Copyright (c) 2010 Janne Grunau <janne-libav@jannau.net>
* 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
*/
/**
* AAC decoder
* @author Oded Shimon ( ods15 ods15 dyndns org )
* @author Maxim Gavrilov ( maxim.gavrilov gmail com )
*/
/*
* supported tools
*
* Support? Name
* N (code in SoC repo) gain control
* Y block switching
* Y window shapes - standard
* N window shapes - Low Delay
* Y filterbank - standard
* N (code in SoC repo) filterbank - Scalable Sample Rate
* Y Temporal Noise Shaping
* Y intensity stereo
* Y channel coupling
* Y frequency domain prediction
* Y Perceptual Noise Substitution
* Y Mid/Side stereo
* N Scalable Inverse AAC Quantization
* N Frequency Selective Switch
* N upsampling filter
* Y quantization & coding - AAC
* N quantization & coding - TwinVQ
* N quantization & coding - BSAC
* N AAC Error Resilience tools
* N Error Resilience payload syntax
* N Error Protection tool
* N CELP
* N Silence Compression
* N HVXC
* N HVXC 4kbits/s VR
* N Structured Audio tools
* N Structured Audio Sample Bank Format
* N MIDI
* N Harmonic and Individual Lines plus Noise
* N Text-To-Speech Interface
* Y (not in this code) Layer-1
* Y (not in this code) Layer-2
* Y (not in this code) Layer-3
* N SinuSoidal Coding (Transient, Sinusoid, Noise)
* N Direct Stream Transfer
*
* Note: - HE AAC v1 comprises LC AAC with Spectral Band Replication.
* - HE AAC v2 comprises LC AAC with Spectral Band Replication and
Parametric Stereo.
*/
#include "internal.h"
#include "fmtconvert.h"
Vitor Sessak
committed
#include "lpc.h"
#include "aac.h"
#include "aactab.h"
Robert Swain
committed
#include "aacdectab.h"
#include "cbrt_tablegen.h"
#include "aacadtsdec.h"
#include "libavutil/intfloat.h"
#include <assert.h>
#include <errno.h>
#include <math.h>
#include <string.h>
#if ARCH_ARM
# include "arm/aac.h"
#endif
static VLC vlc_scalefactors;
static VLC vlc_spectral[11];
#define overread_err "Input buffer exhausted before END element found\n"
Alex Converse
committed
static int count_channels(uint8_t (*layout)[3], int tags)
{
Alex Converse
committed
int i, sum = 0;
for (i = 0; i < tags; i++) {
int syn_ele = layout[i][0];
int pos = layout[i][2];
sum += (1 + (syn_ele == TYPE_CPE)) *
(pos != AAC_CHANNEL_OFF && pos != AAC_CHANNEL_CC);
}
return sum;
}
Robert Swain
committed
/**
* Check for the channel element in the current channel position configuration.
* If it exists, make sure the appropriate element is allocated and map the
* channel order to match the internal FFmpeg channel layout.
*
* @param che_pos current channel position configuration
* @param type channel element type
* @param id channel element id
* @param channels count of the number of channels in the configuration
*
* @return Returns error status. 0 - OK, !0 - error
*/
static av_cold int che_configure(AACContext *ac,
Robert Swain
committed
{
if (!ac->che[type][id]) {
if (!(ac->che[type][id] = av_mallocz(sizeof(ChannelElement))))
return AVERROR(ENOMEM);
ff_aac_sbr_ctx_init(ac, &ac->che[type][id]->sbr);
}
Robert Swain
committed
if (type != TYPE_CCE) {
if (*channels >= MAX_CHANNELS - (type == TYPE_CPE || (type == TYPE_SCE && ac->oc[1].m4ac.ps == 1))) {
av_log(ac->avctx, AV_LOG_ERROR, "Too many channels\n");
return AVERROR_INVALIDDATA;
}
ac->output_element[(*channels)++] = &ac->che[type][id]->ch[0];
(type == TYPE_SCE && ac->oc[1].m4ac.ps == 1)) {
ac->output_element[(*channels)++] = &ac->che[type][id]->ch[1];
Robert Swain
committed
}
}
} else {
if (ac->che[type][id])
ff_aac_sbr_ctx_close(&ac->che[type][id]->sbr);
Robert Swain
committed
av_freep(&ac->che[type][id]);
Robert Swain
committed
return 0;
}
static int frame_configure_elements(AVCodecContext *avctx)
{
AACContext *ac = avctx->priv_data;
int type, id, ch, ret;
/* set channel pointers to internal buffers by default */
for (type = 0; type < 4; type++) {
for (id = 0; id < MAX_ELEM_ID; id++) {
ChannelElement *che = ac->che[type][id];
if (che) {
che->ch[0].ret = che->ch[0].ret_buf;
che->ch[1].ret = che->ch[1].ret_buf;
}
}
}
/* get output buffer */
ac->frame.nb_samples = 2048;
if ((ret = ff_get_buffer(avctx, &ac->frame)) < 0) {
av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n");
return ret;
}
/* map output channel pointers to AVFrame data */
for (ch = 0; ch < avctx->channels; ch++) {
if (ac->output_element[ch])
ac->output_element[ch]->ret = (float *)ac->frame.extended_data[ch];
}
return 0;
}
Alex Converse
committed
struct elem_to_channel {
uint64_t av_position;
uint8_t syn_ele;
uint8_t elem_id;
uint8_t aac_position;
};
static int assign_pair(struct elem_to_channel e2c_vec[MAX_ELEM_ID],
uint8_t (*layout_map)[3], int offset, uint64_t left,
Alex Converse
committed
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
uint64_t right, int pos)
{
if (layout_map[offset][0] == TYPE_CPE) {
e2c_vec[offset] = (struct elem_to_channel) {
.av_position = left | right, .syn_ele = TYPE_CPE,
.elem_id = layout_map[offset ][1], .aac_position = pos };
return 1;
} else {
e2c_vec[offset] = (struct elem_to_channel) {
.av_position = left, .syn_ele = TYPE_SCE,
.elem_id = layout_map[offset ][1], .aac_position = pos };
e2c_vec[offset + 1] = (struct elem_to_channel) {
.av_position = right, .syn_ele = TYPE_SCE,
.elem_id = layout_map[offset + 1][1], .aac_position = pos };
return 2;
}
}
static int count_paired_channels(uint8_t (*layout_map)[3], int tags, int pos, int *current) {
int num_pos_channels = 0;
int first_cpe = 0;
int sce_parity = 0;
int i;
for (i = *current; i < tags; i++) {
if (layout_map[i][2] != pos)
break;
if (layout_map[i][0] == TYPE_CPE) {
if (sce_parity) {
if (pos == AAC_CHANNEL_FRONT && !first_cpe) {
Alex Converse
committed
sce_parity = 0;
} else {
return -1;
}
}
num_pos_channels += 2;
first_cpe = 1;
} else {
num_pos_channels++;
sce_parity ^= 1;
}
}
if (sce_parity &&
((pos == AAC_CHANNEL_FRONT && first_cpe) || pos == AAC_CHANNEL_SIDE))
return -1;
*current = i;
return num_pos_channels;
}
static uint64_t sniff_channel_order(uint8_t (*layout_map)[3], int tags)
{
int i, n, total_non_cc_elements;
struct elem_to_channel e2c_vec[4*MAX_ELEM_ID] = {{ 0 }};
Alex Converse
committed
int num_front_channels, num_side_channels, num_back_channels;
uint64_t layout;
if (FF_ARRAY_ELEMS(e2c_vec) < tags)
Alex Converse
committed
i = 0;
num_front_channels =
count_paired_channels(layout_map, tags, AAC_CHANNEL_FRONT, &i);
if (num_front_channels < 0)
return 0;
num_side_channels =
count_paired_channels(layout_map, tags, AAC_CHANNEL_SIDE, &i);
if (num_side_channels < 0)
return 0;
num_back_channels =
count_paired_channels(layout_map, tags, AAC_CHANNEL_BACK, &i);
if (num_back_channels < 0)
return 0;
i = 0;
if (num_front_channels & 1) {
e2c_vec[i] = (struct elem_to_channel) {
.av_position = AV_CH_FRONT_CENTER, .syn_ele = TYPE_SCE,
.elem_id = layout_map[i][1], .aac_position = AAC_CHANNEL_FRONT };
i++;
num_front_channels--;
}
if (num_front_channels >= 4) {
i += assign_pair(e2c_vec, layout_map, i,
Alex Converse
committed
AV_CH_FRONT_LEFT_OF_CENTER,
AV_CH_FRONT_RIGHT_OF_CENTER,
AAC_CHANNEL_FRONT);
num_front_channels -= 2;
}
if (num_front_channels >= 2) {
i += assign_pair(e2c_vec, layout_map, i,
Alex Converse
committed
AV_CH_FRONT_LEFT,
AV_CH_FRONT_RIGHT,
AAC_CHANNEL_FRONT);
num_front_channels -= 2;
}
while (num_front_channels >= 2) {
i += assign_pair(e2c_vec, layout_map, i,
Alex Converse
committed
UINT64_MAX,
UINT64_MAX,
AAC_CHANNEL_FRONT);
num_front_channels -= 2;
}
if (num_side_channels >= 2) {
i += assign_pair(e2c_vec, layout_map, i,
Alex Converse
committed
AV_CH_SIDE_LEFT,
AV_CH_SIDE_RIGHT,
AAC_CHANNEL_FRONT);
num_side_channels -= 2;
}
while (num_side_channels >= 2) {
i += assign_pair(e2c_vec, layout_map, i,
Alex Converse
committed
UINT64_MAX,
UINT64_MAX,
AAC_CHANNEL_SIDE);
num_side_channels -= 2;
}
while (num_back_channels >= 4) {
i += assign_pair(e2c_vec, layout_map, i,
Alex Converse
committed
UINT64_MAX,
UINT64_MAX,
AAC_CHANNEL_BACK);
num_back_channels -= 2;
}
if (num_back_channels >= 2) {
i += assign_pair(e2c_vec, layout_map, i,
Alex Converse
committed
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
AV_CH_BACK_LEFT,
AV_CH_BACK_RIGHT,
AAC_CHANNEL_BACK);
num_back_channels -= 2;
}
if (num_back_channels) {
e2c_vec[i] = (struct elem_to_channel) {
.av_position = AV_CH_BACK_CENTER, .syn_ele = TYPE_SCE,
.elem_id = layout_map[i][1], .aac_position = AAC_CHANNEL_BACK };
i++;
num_back_channels--;
}
if (i < tags && layout_map[i][2] == AAC_CHANNEL_LFE) {
e2c_vec[i] = (struct elem_to_channel) {
.av_position = AV_CH_LOW_FREQUENCY, .syn_ele = TYPE_LFE,
.elem_id = layout_map[i][1], .aac_position = AAC_CHANNEL_LFE };
i++;
}
while (i < tags && layout_map[i][2] == AAC_CHANNEL_LFE) {
e2c_vec[i] = (struct elem_to_channel) {
.av_position = UINT64_MAX, .syn_ele = TYPE_LFE,
.elem_id = layout_map[i][1], .aac_position = AAC_CHANNEL_LFE };
i++;
}
// Must choose a stable sort
total_non_cc_elements = n = i;
do {
int next_n = 0;
for (i = 1; i < n; i++) {
if (e2c_vec[i-1].av_position > e2c_vec[i].av_position) {
FFSWAP(struct elem_to_channel, e2c_vec[i-1], e2c_vec[i]);
next_n = i;
}
}
n = next_n;
} while (n > 0);
layout = 0;
for (i = 0; i < total_non_cc_elements; i++) {
layout_map[i][0] = e2c_vec[i].syn_ele;
layout_map[i][1] = e2c_vec[i].elem_id;
layout_map[i][2] = e2c_vec[i].aac_position;
if (e2c_vec[i].av_position != UINT64_MAX) {
layout |= e2c_vec[i].av_position;
}
}
return layout;
}
/**
* Save current output configuration if and only if it has been locked.
*/
static void push_output_configuration(AACContext *ac) {
if (ac->oc[1].status == OC_LOCKED) {
ac->oc[0] = ac->oc[1];
}
ac->oc[1].status = OC_NONE;
}
/**
* Restore the previous output configuration if and only if the current
* configuration is unlocked.
*/
static void pop_output_configuration(AACContext *ac) {
Alex Converse
committed
if (ac->oc[1].status != OC_LOCKED && ac->oc[0].status != OC_NONE) {
ac->oc[1] = ac->oc[0];
ac->avctx->channels = ac->oc[1].channels;
ac->avctx->channel_layout = ac->oc[1].channel_layout;
/**
* Configure output channel order based on the current program configuration element.
*
* @return Returns error status. 0 - OK, !0 - error
*/
static int output_configure(AACContext *ac,
uint8_t layout_map[MAX_ELEM_ID*4][3], int tags,
enum OCStatus oc_type, int get_new_frame)
Alex Converse
committed
int i, channels = 0, ret;
if (ac->oc[1].layout_map != layout_map) {
memcpy(ac->oc[1].layout_map, layout_map, tags * sizeof(layout_map[0]));
ac->oc[1].layout_map_tags = tags;
Alex Converse
committed
}
// Try to sniff a reasonable channel order, otherwise output the
// channels in the order the PCE declared them.
if (avctx->request_channel_layout != AV_CH_LAYOUT_NATIVE)
layout = sniff_channel_order(layout_map, tags);
for (i = 0; i < tags; i++) {
int type = layout_map[i][0];
int id = layout_map[i][1];
int position = layout_map[i][2];
// Allocate or free elements depending on if they are in the
// current program configuration.
ret = che_configure(ac, position, type, id, &channels);
if (ret < 0)
return ret;
if (ac->oc[1].m4ac.ps == 1 && channels == 2) {
if (layout == AV_CH_FRONT_CENTER) {
layout = AV_CH_FRONT_LEFT|AV_CH_FRONT_RIGHT;
} else {
layout = 0;
}
}
memcpy(ac->tag_che_map, ac->che, 4 * MAX_ELEM_ID * sizeof(ac->che[0][0]));
if (layout) avctx->channel_layout = layout;
ac->oc[1].channel_layout = layout;
avctx->channels = ac->oc[1].channels = channels;
ac->oc[1].status = oc_type;
Alex Converse
committed
if (get_new_frame) {
if ((ret = frame_configure_elements(ac->avctx)) < 0)
return ret;
}
static void flush(AVCodecContext *avctx)
{
AACContext *ac= avctx->priv_data;
int type, i, j;
for (type = 3; type >= 0; type--) {
for (i = 0; i < MAX_ELEM_ID; i++) {
ChannelElement *che = ac->che[type][i];
if (che) {
for (j = 0; j <= 1; j++) {
memset(che->ch[j].saved, 0, sizeof(che->ch[j].saved));
}
}
}
}
}
Alex Converse
committed
/**
* Set up channel positions based on a default channel configuration
* as specified in table 1.17.
*
* @return Returns error status. 0 - OK, !0 - error
*/
static int set_default_channel_config(AVCodecContext *avctx,
Alex Converse
committed
uint8_t (*layout_map)[3],
int *tags,
int channel_config)
{
if (channel_config < 1 || channel_config > 7) {
av_log(avctx, AV_LOG_ERROR, "invalid default channel configuration (%d)\n",
channel_config);
return -1;
}
*tags = tags_per_config[channel_config];
memcpy(layout_map, aac_channel_layout_map[channel_config-1], *tags * sizeof(*layout_map));
return 0;
}
static ChannelElement *get_che(AACContext *ac, int type, int elem_id)
{
// For PCE based channel configurations map the channels solely based on tags.
if (!ac->oc[1].m4ac.chan_config) {
Alex Converse
committed
return ac->tag_che_map[type][elem_id];
}
Michael Niedermayer
committed
// Allow single CPE stereo files to be signalled with mono configuration.
if (!ac->tags_mapped && type == TYPE_CPE && ac->oc[1].m4ac.chan_config == 1) {
Michael Niedermayer
committed
uint8_t layout_map[MAX_ELEM_ID*4][3];
int layout_map_tags;
push_output_configuration(ac);
Michael Niedermayer
committed
av_log(ac->avctx, AV_LOG_DEBUG, "mono with CPE\n");
Michael Niedermayer
committed
if (set_default_channel_config(ac->avctx, layout_map, &layout_map_tags,
2) < 0)
return NULL;
if (output_configure(ac, layout_map, layout_map_tags,
OC_TRIAL_FRAME, 1) < 0)
Michael Niedermayer
committed
return NULL;
ac->oc[1].m4ac.chan_config = 2;
Alex Converse
committed
ac->oc[1].m4ac.ps = 0;
if (!ac->tags_mapped && type == TYPE_SCE && ac->oc[1].m4ac.chan_config == 2) {
uint8_t layout_map[MAX_ELEM_ID*4][3];
int layout_map_tags;
push_output_configuration(ac);
av_log(ac->avctx, AV_LOG_DEBUG, "stereo with SCE\n");
if (set_default_channel_config(ac->avctx, layout_map, &layout_map_tags,
1) < 0)
return NULL;
if (output_configure(ac, layout_map, layout_map_tags,
OC_TRIAL_FRAME, 1) < 0)
return NULL;
ac->oc[1].m4ac.chan_config = 1;
Alex Converse
committed
if (ac->oc[1].m4ac.sbr)
ac->oc[1].m4ac.ps = -1;
Michael Niedermayer
committed
}
Alex Converse
committed
// For indexed channel configurations map the channels solely based on position.
switch (ac->oc[1].m4ac.chan_config) {
Alex Converse
committed
case 7:
if (ac->tags_mapped == 3 && type == TYPE_CPE) {
ac->tags_mapped++;
return ac->tag_che_map[TYPE_CPE][elem_id] = ac->che[TYPE_CPE][2];
}
case 6:
/* Some streams incorrectly code 5.1 audio as SCE[0] CPE[0] CPE[1] SCE[1]
instead of SCE[0] CPE[0] CPE[1] LFE[0]. If we seem to have
encountered such a stream, transfer the LFE[0] element to the SCE[1]'s mapping */
if (ac->tags_mapped == tags_per_config[ac->oc[1].m4ac.chan_config] - 1 && (type == TYPE_LFE || type == TYPE_SCE)) {
Alex Converse
committed
ac->tags_mapped++;
return ac->tag_che_map[type][elem_id] = ac->che[TYPE_LFE][0];
}
case 5:
if (ac->tags_mapped == 2 && type == TYPE_CPE) {
ac->tags_mapped++;
return ac->tag_che_map[TYPE_CPE][elem_id] = ac->che[TYPE_CPE][1];
}
case 4:
if (ac->tags_mapped == 2 && ac->oc[1].m4ac.chan_config == 4 && type == TYPE_SCE) {
Alex Converse
committed
ac->tags_mapped++;
return ac->tag_che_map[TYPE_SCE][elem_id] = ac->che[TYPE_SCE][1];
}
case 3:
case 2:
if (ac->tags_mapped == (ac->oc[1].m4ac.chan_config != 2) && type == TYPE_CPE) {
Alex Converse
committed
ac->tags_mapped++;
return ac->tag_che_map[TYPE_CPE][elem_id] = ac->che[TYPE_CPE][0];
} else if (ac->oc[1].m4ac.chan_config == 2) {
Alex Converse
committed
return NULL;
}
case 1:
if (!ac->tags_mapped && type == TYPE_SCE) {
ac->tags_mapped++;
return ac->tag_che_map[TYPE_SCE][elem_id] = ac->che[TYPE_SCE][0];
}
default:
return NULL;
}
}
/**
* Decode an array of 4 bit element IDs, optionally interleaved with a stereo/mono switching bit.
*
* @param type speaker type/position for these channels
*/
Alex Converse
committed
static void decode_channel_map(uint8_t layout_map[][3],
enum ChannelPosition type,
GetBitContext *gb, int n)
{
while (n--) {
Alex Converse
committed
enum RawDataBlockType syn_ele;
switch (type) {
case AAC_CHANNEL_FRONT:
case AAC_CHANNEL_BACK:
case AAC_CHANNEL_SIDE:
syn_ele = get_bits1(gb);
break;
case AAC_CHANNEL_CC:
skip_bits1(gb);
syn_ele = TYPE_CCE;
break;
case AAC_CHANNEL_LFE:
syn_ele = TYPE_LFE;
break;
default:
av_assert0(0);
Alex Converse
committed
}
layout_map[0][0] = syn_ele;
layout_map[0][1] = get_bits(gb, 4);
layout_map[0][2] = type;
layout_map++;
}
}
/**
* Decode program configuration element; reference: table 4.2.
*
* @return Returns error status. 0 - OK, !0 - error
*/
static int decode_pce(AVCodecContext *avctx, MPEG4AudioConfig *m4ac,
Alex Converse
committed
uint8_t (*layout_map)[3],
int num_front, num_side, num_back, num_lfe, num_assoc_data, num_cc, sampling_index;
Alex Converse
committed
int tags;
skip_bits(gb, 2); // object_type
sampling_index = get_bits(gb, 4);
if (m4ac->sampling_index != sampling_index)
av_log(avctx, AV_LOG_WARNING, "Sample rate index in program config element does not match the sample rate index configured by the container.\n");
Alex Converse
committed
num_front = get_bits(gb, 4);
num_side = get_bits(gb, 4);
num_back = get_bits(gb, 4);
num_lfe = get_bits(gb, 2);
num_assoc_data = get_bits(gb, 3);
num_cc = get_bits(gb, 4);
Robert Swain
committed
if (get_bits1(gb))
skip_bits(gb, 4); // mono_mixdown_tag
if (get_bits1(gb))
skip_bits(gb, 4); // stereo_mixdown_tag
Robert Swain
committed
if (get_bits1(gb))
skip_bits(gb, 3); // mixdown_coeff_index and pseudo_surround
if (get_bits_left(gb) < 4 * (num_front + num_side + num_back + num_lfe + num_assoc_data + num_cc)) {
av_log(avctx, AV_LOG_ERROR, "decode_pce: " overread_err);
Alex Converse
committed
decode_channel_map(layout_map , AAC_CHANNEL_FRONT, gb, num_front);
tags = num_front;
decode_channel_map(layout_map + tags, AAC_CHANNEL_SIDE, gb, num_side);
tags += num_side;
decode_channel_map(layout_map + tags, AAC_CHANNEL_BACK, gb, num_back);
tags += num_back;
decode_channel_map(layout_map + tags, AAC_CHANNEL_LFE, gb, num_lfe);
tags += num_lfe;
skip_bits_long(gb, 4 * num_assoc_data);
Alex Converse
committed
decode_channel_map(layout_map + tags, AAC_CHANNEL_CC, gb, num_cc);
tags += num_cc;
align_get_bits(gb);
/* comment field, first byte is length */
comment_len = get_bits(gb, 8) * 8;
if (get_bits_left(gb) < comment_len) {
av_log(avctx, AV_LOG_ERROR, "decode_pce: " overread_err);
return -1;
}
skip_bits_long(gb, comment_len);
Alex Converse
committed
return tags;
Robert Swain
committed
}
/**
* Decode GA "General Audio" specific configuration; reference: table 4.1.
*
* @param ac pointer to AACContext, may be null
* @param avctx pointer to AVCCodecContext, used for logging
*
* @return Returns error status. 0 - OK, !0 - error
*/
static int decode_ga_specific_config(AACContext *ac, AVCodecContext *avctx,
GetBitContext *gb,
MPEG4AudioConfig *m4ac,
Alex Converse
committed
uint8_t layout_map[MAX_ELEM_ID*4][3];
int tags = 0;
av_log_missing_feature(avctx, "960/120 MDCT window", 1);
return AVERROR_PATCHWELCOME;
}
if (get_bits1(gb)) // dependsOnCoreCoder
skip_bits(gb, 14); // coreCoderDelay
extension_flag = get_bits1(gb);
if (m4ac->object_type == AOT_AAC_SCALABLE ||
m4ac->object_type == AOT_ER_AAC_SCALABLE)
skip_bits(gb, 3); // layerNr
if (channel_config == 0) {
skip_bits(gb, 4); // element_instance_tag
Alex Converse
committed
tags = decode_pce(avctx, m4ac, layout_map, gb);
if (tags < 0)
return tags;
Alex Converse
committed
if ((ret = set_default_channel_config(avctx, layout_map, &tags, channel_config)))
Alex Converse
committed
if (count_channels(layout_map, tags) > 1) {
m4ac->ps = 0;
} else if (m4ac->sbr == 1 && m4ac->ps == -1)
m4ac->ps = 1;
if (ac && (ret = output_configure(ac, layout_map, tags, OC_GLOBAL_HDR, 0)))
return ret;
if (extension_flag) {
switch (m4ac->object_type) {
case AOT_ER_BSAC:
skip_bits(gb, 5); // numOfSubFrame
skip_bits(gb, 11); // layer_length
break;
case AOT_ER_AAC_LC:
case AOT_ER_AAC_LTP:
case AOT_ER_AAC_SCALABLE:
case AOT_ER_AAC_LD:
skip_bits(gb, 3); /* aacSectionDataResilienceFlag
* aacScalefactorDataResilienceFlag
* aacSpectralDataResilienceFlag
*/
}
skip_bits1(gb); // extensionFlag3 (TBD in version 3)
}
return 0;
}
/**
* Decode audio specific configuration; reference: table 1.13.
*
* @param ac pointer to AACContext, may be null
* @param avctx pointer to AVCCodecContext, used for logging
* @param m4ac pointer to MPEG4AudioConfig, used for parsing
* @param data pointer to buffer holding an audio specific config
* @param bit_size size of audio specific config or data in bits
* @param sync_extension look for an appended sync extension
* @return Returns error status or number of consumed bits. <0 - error
static int decode_audio_specific_config(AACContext *ac,
AVCodecContext *avctx,
MPEG4AudioConfig *m4ac,
const uint8_t *data, int bit_size,
int sync_extension)
GetBitContext gb;
int i;
av_dlog(avctx, "audio specific config size %d\n", bit_size >> 3);
for (i = 0; i < bit_size >> 3; i++)
av_dlog(avctx, "%02x ", data[i]);
init_get_bits(&gb, data, bit_size);
if ((i = avpriv_mpeg4audio_get_config(m4ac, data, bit_size, sync_extension)) < 0)
if (m4ac->sampling_index > 12) {
av_log(avctx, AV_LOG_ERROR, "invalid sampling rate index %d\n", m4ac->sampling_index);
return -1;
}
skip_bits_long(&gb, i);
switch (m4ac->object_type) {
case AOT_AAC_MAIN:
if (decode_ga_specific_config(ac, avctx, &gb, m4ac, m4ac->chan_config))
return -1;
break;
default:
av_log(avctx, AV_LOG_ERROR, "Audio object type %s%d is not supported.\n",
m4ac->sbr == 1? "SBR+" : "", m4ac->object_type);
av_dlog(avctx, "AOT %d chan config %d sampling index %d (%d) SBR %d PS %d\n",
m4ac->object_type, m4ac->chan_config, m4ac->sampling_index,
m4ac->sample_rate, m4ac->sbr, m4ac->ps);
return get_bits_count(&gb);
/**
* linear congruential pseudorandom number generator
*
* @param previous_val pointer to the current state of the generator
*
* @return Returns a 32-bit pseudorandom integer
*/
static av_always_inline int lcg_random(unsigned previous_val)
union { unsigned u; int s; } v = { previous_val * 1664525u + 1013904223 };
return v.s;
static av_always_inline void reset_predict_state(PredictorState *ps)
ps->cor0 = 0.0f;
ps->cor1 = 0.0f;
ps->var0 = 1.0f;
ps->var1 = 1.0f;
}
static void reset_all_predictors(PredictorState *ps)
{
int i;
for (i = 0; i < MAX_PREDICTORS; i++)
reset_predict_state(&ps[i]);
}
static int sample_rate_idx (int rate)
{
if (92017 <= rate) return 0;
else if (75132 <= rate) return 1;
else if (55426 <= rate) return 2;
else if (46009 <= rate) return 3;
else if (37566 <= rate) return 4;
else if (27713 <= rate) return 5;
else if (23004 <= rate) return 6;
else if (18783 <= rate) return 7;
else if (13856 <= rate) return 8;
else if (11502 <= rate) return 9;
else if (9391 <= rate) return 10;
else return 11;
}
static void reset_predictor_group(PredictorState *ps, int group_num)
{
for (i = group_num - 1; i < MAX_PREDICTORS; i += 30)
reset_predict_state(&ps[i]);
}
Alex Converse
committed
#define AAC_INIT_VLC_STATIC(num, size) \
INIT_VLC_STATIC(&vlc_spectral[num], 8, ff_aac_spectral_sizes[num], \
ff_aac_spectral_bits[num], sizeof( ff_aac_spectral_bits[num][0]), sizeof( ff_aac_spectral_bits[num][0]), \
ff_aac_spectral_codes[num], sizeof(ff_aac_spectral_codes[num][0]), sizeof(ff_aac_spectral_codes[num][0]), \
size);
static av_cold int aac_decode_init(AVCodecContext *avctx)
ac->oc[1].m4ac.sample_rate = avctx->sample_rate;
avctx->sample_fmt = AV_SAMPLE_FMT_FLTP;
if (decode_audio_specific_config(ac, ac->avctx, &ac->oc[1].m4ac,
avctx->extradata,
avctx->extradata_size*8, 1) < 0)
Robert Swain
committed
return -1;
} else {
int sr, i;
Alex Converse
committed
uint8_t layout_map[MAX_ELEM_ID*4][3];
int layout_map_tags;
sr = sample_rate_idx(avctx->sample_rate);
ac->oc[1].m4ac.sampling_index = sr;
ac->oc[1].m4ac.channels = avctx->channels;
ac->oc[1].m4ac.sbr = -1;
ac->oc[1].m4ac.ps = -1;
for (i = 0; i < FF_ARRAY_ELEMS(ff_mpeg4audio_channels); i++)
if (ff_mpeg4audio_channels[i] == avctx->channels)
break;
if (i == FF_ARRAY_ELEMS(ff_mpeg4audio_channels)) {
i = 0;
}
ac->oc[1].m4ac.chan_config = i;
if (ac->oc[1].m4ac.chan_config) {
Alex Converse
committed
int ret = set_default_channel_config(avctx, layout_map,
&layout_map_tags, ac->oc[1].m4ac.chan_config);
if (!ret)
Alex Converse
committed
output_configure(ac, layout_map, layout_map_tags,
Dustin Brody
committed
else if (avctx->err_recognition & AV_EF_EXPLODE)
return AVERROR_INVALIDDATA;
Robert Swain
committed
}
Robert Swain
committed
AAC_INIT_VLC_STATIC( 0, 304);
AAC_INIT_VLC_STATIC( 1, 270);
AAC_INIT_VLC_STATIC( 2, 550);
AAC_INIT_VLC_STATIC( 3, 300);
AAC_INIT_VLC_STATIC( 4, 328);
AAC_INIT_VLC_STATIC( 5, 294);
AAC_INIT_VLC_STATIC( 6, 306);
AAC_INIT_VLC_STATIC( 7, 268);
AAC_INIT_VLC_STATIC( 8, 510);
AAC_INIT_VLC_STATIC( 9, 366);
AAC_INIT_VLC_STATIC(10, 462);
ff_dsputil_init(&ac->dsp, avctx);
ff_fmt_convert_init(&ac->fmt_conv, avctx);
avpriv_float_dsp_init(&ac->fdsp, avctx->flags & CODEC_FLAG_BITEXACT);
ac->random_state = 0x1f2e3d4c;
INIT_VLC_STATIC(&vlc_scalefactors,7,FF_ARRAY_ELEMS(ff_aac_scalefactor_code),
ff_aac_scalefactor_bits, sizeof(ff_aac_scalefactor_bits[0]), sizeof(ff_aac_scalefactor_bits[0]),
ff_aac_scalefactor_code, sizeof(ff_aac_scalefactor_code[0]), sizeof(ff_aac_scalefactor_code[0]),
352);
ff_mdct_init(&ac->mdct, 11, 1, 1.0 / (32768.0 * 1024.0));
ff_mdct_init(&ac->mdct_small, 8, 1, 1.0 / (32768.0 * 128.0));
ff_mdct_init(&ac->mdct_ltp, 11, 0, -2.0 * 32768.0);
// window initialization
ff_kbd_window_init(ff_aac_kbd_long_1024, 4.0, 1024);
ff_kbd_window_init(ff_aac_kbd_short_128, 6.0, 128);
ff_init_ff_sine_windows(10);
ff_init_ff_sine_windows( 7);
cbrt_tableinit();
avcodec_get_frame_defaults(&ac->frame);
avctx->coded_frame = &ac->frame;
/**
* Skip data_stream_element; reference: table 4.10.
*/
static int skip_data_stream_element(AACContext *ac, GetBitContext *gb)
int byte_align = get_bits1(gb);
int count = get_bits(gb, 8);
if (count == 255)
count += get_bits(gb, 8);
if (byte_align)
align_get_bits(gb);
if (get_bits_left(gb) < 8 * count) {
av_log(ac->avctx, AV_LOG_ERROR, "skip_data_stream_element: "overread_err);
skip_bits_long(gb, 8 * count);
static int decode_prediction(AACContext *ac, IndividualChannelStream *ics,
GetBitContext *gb)
{
int sfb;
if (get_bits1(gb)) {
ics->predictor_reset_group = get_bits(gb, 5);
if (ics->predictor_reset_group == 0 || ics->predictor_reset_group > 30) {
av_log(ac->avctx, AV_LOG_ERROR, "Invalid Predictor Reset Group.\n");
return -1;
}
}
for (sfb = 0; sfb < FFMIN(ics->max_sfb, ff_aac_pred_sfb_max[ac->oc[1].m4ac.sampling_index]); sfb++) {
ics->prediction_used[sfb] = get_bits1(gb);
}
return 0;
}
/**
* Decode Long Term Prediction data; reference: table 4.xx.
*/
static void decode_ltp(LongTermPrediction *ltp,