* AAC decoder
* Copyright (c) 2005-2006 Oded Shimon ( ods15 ods15 dyndns org )
* Copyright (c) 2006-2007 Maxim Gavrilov ( maxim.gavrilov gmail com )
+ * Copyright (c) 2008-2013 Alex Converse <alex.converse@gmail.com>
*
* AAC LATM decoder
* Copyright (c) 2008-2010 Paul Kendall <paul@kcbbs.gen.nz>
#include "internal.h"
#include "get_bits.h"
#include "fft.h"
-#include "fmtconvert.h"
+#include "imdct15.h"
#include "lpc.h"
#include "kbdwin.h"
#include "sinewin.h"
#include "aac.h"
#include "aactab.h"
#include "aacdectab.h"
+#include "adts_header.h"
#include "cbrt_tablegen.h"
#include "sbr.h"
#include "aacsbr.h"
#include "mpeg4audio.h"
-#include "aacadtsdec.h"
#include "libavutil/intfloat.h"
#include <assert.h>
#include <errno.h>
#include <math.h>
+#include <stdint.h>
#include <string.h>
#if ARCH_ARM
# include "arm/aac.h"
#endif
+#include "libavutil/thread.h"
+
static VLC vlc_scalefactors;
static VLC vlc_spectral[11];
enum ChannelPosition che_pos,
int type, int id, int *channels)
{
- if (*channels >= MAX_CHANNELS)
- return AVERROR_INVALIDDATA;
if (che_pos) {
if (!ac->che[type][id]) {
if (!(ac->che[type][id] = av_mallocz(sizeof(ChannelElement))))
ff_aac_sbr_ctx_init(ac, &ac->che[type][id]->sbr);
}
if (type != TYPE_CCE) {
+ if (*channels >= MAX_CHANNELS - 2)
+ return AVERROR_INVALIDDATA;
ac->output_element[(*channels)++] = &ac->che[type][id]->ch[0];
if (type == TYPE_CPE ||
(type == TYPE_SCE && ac->oc[1].m4ac.ps == 1)) {
if (num_back_channels < 0)
return 0;
+ if (num_side_channels == 0 && num_back_channels >= 4) {
+ num_side_channels = 2;
+ num_back_channels -= 2;
+ }
+
i = 0;
if (num_front_channels & 1) {
e2c_vec[i] = (struct elem_to_channel) {
AVCodecContext *avctx = ac->avctx;
int i, channels = 0, ret;
uint64_t layout = 0;
+ uint8_t id_map[TYPE_END][MAX_ELEM_ID] = {{ 0 }};
+ uint8_t type_counts[TYPE_END] = { 0 };
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;
}
-
+ for (i = 0; i < tags; i++) {
+ int type = layout_map[i][0];
+ int id = layout_map[i][1];
+ id_map[type][id] = type_counts[type]++;
+ }
// 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)
for (i = 0; i < tags; i++) {
int type = layout_map[i][0];
int id = layout_map[i][1];
+ int iid = id_map[type][id];
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);
+ ret = che_configure(ac, position, type, iid, &channels);
if (ret < 0)
return ret;
+ ac->tag_che_map[type][id] = ac->che[type][iid];
}
if (ac->oc[1].m4ac.ps == 1 && channels == 2) {
if (layout == AV_CH_FRONT_CENTER) {
}
}
- memcpy(ac->tag_che_map, ac->che, 4 * MAX_ELEM_ID * sizeof(ac->che[0][0]));
avctx->channel_layout = ac->oc[1].channel_layout = layout;
avctx->channels = ac->oc[1].channels = channels;
ac->oc[1].status = oc_type;
int *tags,
int channel_config)
{
- if (channel_config < 1 || channel_config > 7) {
+ if (channel_config < 1 || (channel_config > 7 && channel_config < 11) ||
+ channel_config > 12) {
av_log(avctx, AV_LOG_ERROR,
"invalid default channel configuration (%d)\n",
channel_config);
/* For indexed channel configurations map the channels solely based
* on position. */
switch (ac->oc[1].m4ac.chan_config) {
+ case 12:
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 11:
+ if (ac->tags_mapped == 2 &&
+ ac->oc[1].m4ac.chan_config == 11 &&
+ type == TYPE_SCE) {
+ ac->tags_mapped++;
+ return ac->tag_che_map[TYPE_SCE][elem_id] = ac->che[TYPE_SCE][1];
+ }
case 6:
/* Some streams incorrectly code 5.1 audio as
* SCE[0] CPE[0] CPE[1] SCE[1]
case AAC_CHANNEL_LFE:
syn_ele = TYPE_LFE;
break;
+ default:
+ // AAC_CHANNEL_OFF has no channel map
+ return;
}
layout_map[0][0] = syn_ele;
layout_map[0][1] = get_bits(gb, 4);
avpriv_request_sample(avctx, "960/120 MDCT window");
return AVERROR_PATCHWELCOME;
}
+ m4ac->frame_length_short = 0;
if (get_bits1(gb)) // dependsOnCoreCoder
skip_bits(gb, 14); // coreCoderDelay
case AOT_ER_AAC_LD:
res_flags = get_bits(gb, 3);
if (res_flags) {
- av_log(avctx, AV_LOG_ERROR,
- "AAC data resilience not supported (flags %x)\n",
- res_flags);
+ avpriv_report_missing_feature(avctx,
+ "AAC data resilience (flags %x)",
+ res_flags);
return AVERROR_PATCHWELCOME;
}
break;
case AOT_ER_AAC_LD:
ep_config = get_bits(gb, 2);
if (ep_config) {
- av_log(avctx, AV_LOG_ERROR,
- "epConfig %d is not supported.\n",
- ep_config);
+ avpriv_report_missing_feature(avctx,
+ "epConfig %d", ep_config);
return AVERROR_PATCHWELCOME;
}
}
return 0;
}
+static int decode_eld_specific_config(AACContext *ac, AVCodecContext *avctx,
+ GetBitContext *gb,
+ MPEG4AudioConfig *m4ac,
+ int channel_config)
+{
+ int ret, ep_config, res_flags;
+ uint8_t layout_map[MAX_ELEM_ID*4][3];
+ int tags = 0;
+ const int ELDEXT_TERM = 0;
+
+ m4ac->ps = 0;
+ m4ac->sbr = 0;
+
+ m4ac->frame_length_short = get_bits1(gb);
+ res_flags = get_bits(gb, 3);
+ if (res_flags) {
+ avpriv_report_missing_feature(avctx,
+ "AAC data resilience (flags %x)",
+ res_flags);
+ return AVERROR_PATCHWELCOME;
+ }
+
+ if (get_bits1(gb)) { // ldSbrPresentFlag
+ avpriv_report_missing_feature(avctx,
+ "Low Delay SBR");
+ return AVERROR_PATCHWELCOME;
+ }
+
+ while (get_bits(gb, 4) != ELDEXT_TERM) {
+ int len = get_bits(gb, 4);
+ if (len == 15)
+ len += get_bits(gb, 8);
+ if (len == 15 + 255)
+ len += get_bits(gb, 16);
+ if (get_bits_left(gb) < len * 8 + 4) {
+ av_log(avctx, AV_LOG_ERROR, overread_err);
+ return AVERROR_INVALIDDATA;
+ }
+ skip_bits_long(gb, 8 * len);
+ }
+
+ if ((ret = set_default_channel_config(avctx, layout_map,
+ &tags, channel_config)))
+ return ret;
+
+ if (ac && (ret = output_configure(ac, layout_map, tags, OC_GLOBAL_HDR, 0)))
+ return ret;
+
+ ep_config = get_bits(gb, 2);
+ if (ep_config) {
+ avpriv_report_missing_feature(avctx,
+ "epConfig %d", ep_config);
+ return AVERROR_PATCHWELCOME;
+ }
+ return 0;
+}
+
/**
* Decode audio specific configuration; reference: table 1.13.
*
GetBitContext gb;
int i, ret;
- av_dlog(avctx, "extradata size %d\n", avctx->extradata_size);
+ ff_dlog(avctx, "extradata size %d\n", avctx->extradata_size);
for (i = 0; i < avctx->extradata_size; i++)
- av_dlog(avctx, "%02x ", avctx->extradata[i]);
- av_dlog(avctx, "\n");
+ ff_dlog(avctx, "%02x ", avctx->extradata[i]);
+ ff_dlog(avctx, "\n");
if ((ret = init_get_bits(&gb, data, bit_size)) < 0)
return ret;
m4ac, m4ac->chan_config)) < 0)
return ret;
break;
+ case AOT_ER_AAC_ELD:
+ if ((ret = decode_eld_specific_config(ac, avctx, &gb,
+ m4ac, m4ac->chan_config)) < 0)
+ return ret;
+ break;
default:
- av_log(avctx, AV_LOG_ERROR,
- "Audio object type %s%d is not supported.\n",
- m4ac->sbr == 1 ? "SBR+" : "",
- m4ac->object_type);
+ avpriv_report_missing_feature(avctx,
+ "Audio object type %s%d",
+ m4ac->sbr == 1 ? "SBR+" : "",
+ m4ac->object_type);
return AVERROR(ENOSYS);
}
- av_dlog(avctx,
+ ff_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,
sizeof(ff_aac_spectral_codes[num][0]), \
size);
+static av_cold void aac_static_table_init(void)
+{
+ 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_aac_sbr_init();
+
+ ff_aac_tableinit();
+
+ 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);
+
+
+ // 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( 9);
+ ff_init_ff_sine_windows( 7);
+
+ cbrt_tableinit();
+}
+
+static AVOnce aac_init = AV_ONCE_INIT;
+
static av_cold int aac_decode_init(AVCodecContext *avctx)
{
AACContext *ac = avctx->priv_data;
int ret;
+ ret = ff_thread_once(&aac_init, &aac_static_table_init);
+ if (ret != 0)
+ return AVERROR_UNKNOWN;
+
ac->avctx = avctx;
ac->oc[1].m4ac.sample_rate = avctx->sample_rate;
}
}
- 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_aac_sbr_init();
-
- ff_fmt_convert_init(&ac->fmt_conv, avctx);
- avpriv_float_dsp_init(&ac->fdsp, avctx->flags & CODEC_FLAG_BITEXACT);
+ avpriv_float_dsp_init(&ac->fdsp, avctx->flags & AV_CODEC_FLAG_BITEXACT);
ac->random_state = 0x1f2e3d4c;
- ff_aac_tableinit();
-
- 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_ld, 10, 1, 1.0 / (32768.0 * 512.0));
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_long_512, 4.0, 512);
- ff_kbd_window_init(ff_aac_kbd_short_128, 6.0, 128);
- ff_init_ff_sine_windows(10);
- ff_init_ff_sine_windows( 9);
- ff_init_ff_sine_windows( 7);
-
- cbrt_tableinit();
+ ret = ff_imdct15_init(&ac->mdct480, 5);
+ if (ret < 0)
+ return ret;
return 0;
}
static int decode_ics_info(AACContext *ac, IndividualChannelStream *ics,
GetBitContext *gb)
{
- if (get_bits1(gb)) {
- av_log(ac->avctx, AV_LOG_ERROR, "Reserved bit set.\n");
- return AVERROR_INVALIDDATA;
- }
- ics->window_sequence[1] = ics->window_sequence[0];
- ics->window_sequence[0] = get_bits(gb, 2);
- if (ac->oc[1].m4ac.object_type == AOT_ER_AAC_LD &&
- ics->window_sequence[0] != ONLY_LONG_SEQUENCE) {
- av_log(ac->avctx, AV_LOG_ERROR,
- "AAC LD is only defined for ONLY_LONG_SEQUENCE but "
- "window sequence %d found.\n", ics->window_sequence[0]);
- ics->window_sequence[0] = ONLY_LONG_SEQUENCE;
- return AVERROR_INVALIDDATA;
+ const MPEG4AudioConfig *const m4ac = &ac->oc[1].m4ac;
+ const int aot = m4ac->object_type;
+ const int sampling_index = m4ac->sampling_index;
+ if (aot != AOT_ER_AAC_ELD) {
+ if (get_bits1(gb)) {
+ av_log(ac->avctx, AV_LOG_ERROR, "Reserved bit set.\n");
+ if (ac->avctx->err_recognition & AV_EF_BITSTREAM)
+ return AVERROR_INVALIDDATA;
+ }
+ ics->window_sequence[1] = ics->window_sequence[0];
+ ics->window_sequence[0] = get_bits(gb, 2);
+ if (aot == AOT_ER_AAC_LD &&
+ ics->window_sequence[0] != ONLY_LONG_SEQUENCE) {
+ av_log(ac->avctx, AV_LOG_ERROR,
+ "AAC LD is only defined for ONLY_LONG_SEQUENCE but "
+ "window sequence %d found.\n", ics->window_sequence[0]);
+ ics->window_sequence[0] = ONLY_LONG_SEQUENCE;
+ return AVERROR_INVALIDDATA;
+ }
+ ics->use_kb_window[1] = ics->use_kb_window[0];
+ ics->use_kb_window[0] = get_bits1(gb);
}
- ics->use_kb_window[1] = ics->use_kb_window[0];
- ics->use_kb_window[0] = get_bits1(gb);
ics->num_window_groups = 1;
ics->group_len[0] = 1;
if (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) {
}
}
ics->num_windows = 8;
- ics->swb_offset = ff_swb_offset_128[ac->oc[1].m4ac.sampling_index];
- ics->num_swb = ff_aac_num_swb_128[ac->oc[1].m4ac.sampling_index];
- ics->tns_max_bands = ff_tns_max_bands_128[ac->oc[1].m4ac.sampling_index];
+ ics->swb_offset = ff_swb_offset_128[sampling_index];
+ ics->num_swb = ff_aac_num_swb_128[sampling_index];
+ ics->tns_max_bands = ff_tns_max_bands_128[sampling_index];
ics->predictor_present = 0;
} else {
- ics->max_sfb = get_bits(gb, 6);
- ics->num_windows = 1;
- if (ac->oc[1].m4ac.object_type == AOT_ER_AAC_LD) {
- ics->swb_offset = ff_swb_offset_512[ac->oc[1].m4ac.sampling_index];
- ics->num_swb = ff_aac_num_swb_512[ac->oc[1].m4ac.sampling_index];
+ ics->max_sfb = get_bits(gb, 6);
+ ics->num_windows = 1;
+ if (aot == AOT_ER_AAC_LD || aot == AOT_ER_AAC_ELD) {
+ if (m4ac->frame_length_short) {
+ ics->swb_offset = ff_swb_offset_480[sampling_index];
+ ics->num_swb = ff_aac_num_swb_480[sampling_index];
+ ics->tns_max_bands = ff_tns_max_bands_480[sampling_index];
+ } else {
+ ics->swb_offset = ff_swb_offset_512[sampling_index];
+ ics->num_swb = ff_aac_num_swb_512[sampling_index];
+ ics->tns_max_bands = ff_tns_max_bands_512[sampling_index];
+ }
if (!ics->num_swb || !ics->swb_offset)
return AVERROR_BUG;
} else {
- ics->swb_offset = ff_swb_offset_1024[ac->oc[1].m4ac.sampling_index];
- ics->num_swb = ff_aac_num_swb_1024[ac->oc[1].m4ac.sampling_index];
+ ics->swb_offset = ff_swb_offset_1024[sampling_index];
+ ics->num_swb = ff_aac_num_swb_1024[sampling_index];
+ ics->tns_max_bands = ff_tns_max_bands_1024[sampling_index];
+ }
+ if (aot != AOT_ER_AAC_ELD) {
+ ics->predictor_present = get_bits1(gb);
+ ics->predictor_reset_group = 0;
}
- ics->tns_max_bands = ff_tns_max_bands_1024[ac->oc[1].m4ac.sampling_index];
- ics->predictor_present = get_bits1(gb);
- ics->predictor_reset_group = 0;
if (ics->predictor_present) {
- if (ac->oc[1].m4ac.object_type == AOT_AAC_MAIN) {
+ if (aot == AOT_AAC_MAIN) {
if (decode_prediction(ac, ics, gb)) {
return AVERROR_INVALIDDATA;
}
- } else if (ac->oc[1].m4ac.object_type == AOT_AAC_LC ||
- ac->oc[1].m4ac.object_type == AOT_ER_AAC_LC) {
+ } else if (aot == AOT_AAC_LC ||
+ aot == AOT_ER_AAC_LC) {
av_log(ac->avctx, AV_LOG_ERROR,
"Prediction is not allowed in AAC-LC.\n");
return AVERROR_INVALIDDATA;
} else {
- if (ac->oc[1].m4ac.object_type == AOT_ER_AAC_LD) {
- av_log(ac->avctx, AV_LOG_ERROR,
- "LTP in ER AAC LD not yet implemented.\n");
+ if (aot == AOT_ER_AAC_LD) {
+ avpriv_report_missing_feature(ac->avctx, "LTP in ER AAC LD");
return AVERROR_PATCHWELCOME;
}
if ((ics->ltp.present = get_bits(gb, 1)))
int ms_present)
{
int idx;
+ int max_idx = cpe->ch[0].ics.num_window_groups * cpe->ch[0].ics.max_sfb;
if (ms_present == 1) {
- for (idx = 0;
- idx < cpe->ch[0].ics.num_window_groups * cpe->ch[0].ics.max_sfb;
- idx++)
+ for (idx = 0; idx < max_idx; idx++)
cpe->ms_mask[idx] = get_bits1(gb);
} else if (ms_present == 2) {
- memset(cpe->ms_mask, 1, cpe->ch[0].ics.num_window_groups * cpe->ch[0].ics.max_sfb * sizeof(cpe->ms_mask[0]));
+ memset(cpe->ms_mask, 1, max_idx * sizeof(cpe->ms_mask[0]));
}
}
TemporalNoiseShaping *tns = &sce->tns;
IndividualChannelStream *ics = &sce->ics;
float *out = sce->coeffs;
- int global_gain, er_syntax, pulse_present = 0;
+ int global_gain, eld_syntax, er_syntax, pulse_present = 0;
int ret;
+ eld_syntax = ac->oc[1].m4ac.object_type == AOT_ER_AAC_ELD;
+ er_syntax = ac->oc[1].m4ac.object_type == AOT_ER_AAC_LC ||
+ ac->oc[1].m4ac.object_type == AOT_ER_AAC_LTP ||
+ ac->oc[1].m4ac.object_type == AOT_ER_AAC_LD ||
+ ac->oc[1].m4ac.object_type == AOT_ER_AAC_ELD;
+
/* This assignment is to silence a GCC warning about the variable being used
* uninitialized when in fact it always is.
*/
return ret;
pulse_present = 0;
- er_syntax = ac->oc[1].m4ac.object_type == AOT_ER_AAC_LC ||
- ac->oc[1].m4ac.object_type == AOT_ER_AAC_LTP ||
- ac->oc[1].m4ac.object_type == AOT_ER_AAC_LD;
if (!scale_flag) {
- if ((pulse_present = get_bits1(gb))) {
+ if (!eld_syntax && (pulse_present = get_bits1(gb))) {
if (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) {
av_log(ac->avctx, AV_LOG_ERROR,
"Pulse tool not allowed in eight short sequence.\n");
if (tns->present && !er_syntax)
if (decode_tns(ac, tns, gb, ics) < 0)
return AVERROR_INVALIDDATA;
- if (get_bits1(gb)) {
+ if (!eld_syntax && get_bits1(gb)) {
avpriv_request_sample(ac->avctx, "SSR");
return AVERROR_PATCHWELCOME;
}
- // I see no textual basis in the spec for this occuring after SSR gain
- // control, but this is what both reference and real implmentations do
+ // I see no textual basis in the spec for this occurring after SSR gain
+ // control, but this is what both reference and real implementations do
if (tns->present && er_syntax)
if (decode_tns(ac, tns, gb, ics) < 0)
return AVERROR_INVALIDDATA;
static int decode_cpe(AACContext *ac, GetBitContext *gb, ChannelElement *cpe)
{
int i, ret, common_window, ms_present = 0;
+ int eld_syntax = ac->oc[1].m4ac.object_type == AOT_ER_AAC_ELD;
- common_window = get_bits1(gb);
+ common_window = eld_syntax || get_bits1(gb);
if (common_window) {
if (decode_ics_info(ac, &cpe->ch[0].ics, gb))
return AVERROR_INVALIDDATA;
} else if (ac->oc[1].m4ac.ps == -1 && ac->oc[1].status < OC_LOCKED && ac->avctx->channels == 1) {
ac->oc[1].m4ac.sbr = 1;
ac->oc[1].m4ac.ps = 1;
+ ac->avctx->profile = FF_PROFILE_AAC_HE_V2;
output_configure(ac, ac->oc[1].layout_map, ac->oc[1].layout_map_tags,
ac->oc[1].status, 1);
} else {
ac->oc[1].m4ac.sbr = 1;
+ ac->avctx->profile = FF_PROFILE_AAC_HE;
}
res = ff_decode_sbr_extension(ac, &che->sbr, gb, crc_flag, cnt, elem_type);
break;
float *in = sce->coeffs;
float *out = sce->ret;
float *saved = sce->saved;
- const float *lwindow_prev = ics->use_kb_window[1] ? ff_aac_kbd_long_512 : ff_sine_512;
float *buf = ac->buf_mdct;
// imdct
ac->mdct.imdct_half(&ac->mdct_ld, buf, in);
// window overlapping
- ac->fdsp.vector_fmul_window(out, saved, buf, lwindow_prev, 256);
+ if (ics->use_kb_window[1]) {
+ // AAC LD uses a low overlap sine window instead of a KBD window
+ memcpy(out, saved, 192 * sizeof(float));
+ ac->fdsp.vector_fmul_window(out + 192, saved + 192, buf, ff_sine_128, 64);
+ memcpy( out + 320, buf + 64, 192 * sizeof(float));
+ } else {
+ ac->fdsp.vector_fmul_window(out, saved, buf, ff_sine_512, 256);
+ }
// buffer update
memcpy(saved, buf + 256, 256 * sizeof(float));
}
+static void imdct_and_windowing_eld(AACContext *ac, SingleChannelElement *sce)
+{
+ float *in = sce->coeffs;
+ float *out = sce->ret;
+ float *saved = sce->saved;
+ float *buf = ac->buf_mdct;
+ int i;
+ const int n = ac->oc[1].m4ac.frame_length_short ? 480 : 512;
+ const int n2 = n >> 1;
+ const int n4 = n >> 2;
+ const float *const window = n == 480 ? ff_aac_eld_window_480 :
+ ff_aac_eld_window_512;
+
+ // Inverse transform, mapped to the conventional IMDCT by
+ // Chivukula, R.K.; Reznik, Y.A.; Devarajan, V.,
+ // "Efficient algorithms for MPEG-4 AAC-ELD, AAC-LD and AAC-LC filterbanks,"
+ // Audio, Language and Image Processing, 2008. ICALIP 2008. International Conference on
+ // URL: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=4590245&isnumber=4589950
+ for (i = 0; i < n2; i+=2) {
+ float temp;
+ temp = in[i ]; in[i ] = -in[n - 1 - i]; in[n - 1 - i] = temp;
+ temp = -in[i + 1]; in[i + 1] = in[n - 2 - i]; in[n - 2 - i] = temp;
+ }
+ if (n == 480)
+ ac->mdct480->imdct_half(ac->mdct480, buf, in, 1, -1.f/(16*1024*960));
+ else
+ ac->mdct.imdct_half(&ac->mdct_ld, buf, in);
+ for (i = 0; i < n; i+=2) {
+ buf[i] = -buf[i];
+ }
+ // Like with the regular IMDCT at this point we still have the middle half
+ // of a transform but with even symmetry on the left and odd symmetry on
+ // the right
+
+ // window overlapping
+ // The spec says to use samples [0..511] but the reference decoder uses
+ // samples [128..639].
+ for (i = n4; i < n2; i ++) {
+ out[i - n4] = buf[n2 - 1 - i] * window[i - n4] +
+ saved[ i + n2] * window[i + n - n4] +
+ -saved[ n + n2 - 1 - i] * window[i + 2*n - n4] +
+ -saved[2*n + n2 + i] * window[i + 3*n - n4];
+ }
+ for (i = 0; i < n2; i ++) {
+ out[n4 + i] = buf[i] * window[i + n2 - n4] +
+ -saved[ n - 1 - i] * window[i + n2 + n - n4] +
+ -saved[ n + i] * window[i + n2 + 2*n - n4] +
+ saved[2*n + n - 1 - i] * window[i + n2 + 3*n - n4];
+ }
+ for (i = 0; i < n4; i ++) {
+ out[n2 + n4 + i] = buf[ i + n2] * window[i + n - n4] +
+ -saved[ n2 - 1 - i] * window[i + 2*n - n4] +
+ -saved[ n + n2 + i] * window[i + 3*n - n4];
+ }
+
+ // buffer update
+ memmove(saved + n, saved, 2 * n * sizeof(float));
+ memcpy( saved, buf, n * sizeof(float));
+}
+
/**
* Apply dependent channel coupling (applied before IMDCT).
*
const float gain = cce->coup.gain[index][idx];
for (group = 0; group < ics->group_len[g]; group++) {
for (k = offsets[i]; k < offsets[i + 1]; k++) {
- // XXX dsputil-ize
+ // FIXME: SIMDify
dest[group * 128 + k] += gain * src[group * 128 + k];
}
}
{
int i, type;
void (*imdct_and_window)(AACContext *ac, SingleChannelElement *sce);
- if (ac->oc[1].m4ac.object_type == AOT_ER_AAC_LD)
+ switch (ac->oc[1].m4ac.object_type) {
+ case AOT_ER_AAC_LD:
imdct_and_window = imdct_and_windowing_ld;
- else
+ break;
+ case AOT_ER_AAC_ELD:
+ imdct_and_window = imdct_and_windowing_eld;
+ break;
+ default:
imdct_and_window = imdct_and_windowing;
+ }
for (type = 3; type >= 0; type--) {
for (i = 0; i < MAX_ELEM_ID; i++) {
ChannelElement *che = ac->che[type][i];
uint8_t layout_map[MAX_ELEM_ID*4][3];
int layout_map_tags, ret;
- size = avpriv_aac_parse_header(gb, &hdr_info);
+ size = ff_adts_header_parse(gb, &hdr_info);
if (size > 0) {
if (hdr_info.num_aac_frames != 1) {
avpriv_report_missing_feature(ac->avctx,
ac->oc[1].m4ac.sample_rate = hdr_info.sample_rate;
ac->oc[1].m4ac.sampling_index = hdr_info.sampling_index;
ac->oc[1].m4ac.object_type = hdr_info.object_type;
+ ac->oc[1].m4ac.frame_length_short = 0;
if (ac->oc[0].status != OC_LOCKED ||
ac->oc[0].m4ac.chan_config != hdr_info.chan_config ||
ac->oc[0].m4ac.sample_rate != hdr_info.sample_rate) {
int *got_frame_ptr, GetBitContext *gb)
{
AACContext *ac = avctx->priv_data;
+ const MPEG4AudioConfig *const m4ac = &ac->oc[1].m4ac;
ChannelElement *che;
int err, i;
- int samples = 1024;
- int chan_config = ac->oc[1].m4ac.chan_config;
+ int samples = m4ac->frame_length_short ? 960 : 1024;
+ int chan_config = m4ac->chan_config;
+ int aot = m4ac->object_type;
- if (ac->oc[1].m4ac.object_type == AOT_ER_AAC_LD)
+ if (aot == AOT_ER_AAC_LD || aot == AOT_ER_AAC_ELD)
samples >>= 1;
ac->frame = data;
if ((err = frame_configure_elements(avctx)) < 0)
return err;
+ // The FF_PROFILE_AAC_* defines are all object_type - 1
+ // This may lead to an undefined profile being signaled
+ ac->avctx->profile = aot - 1;
+
ac->tags_mapped = 0;
- if (chan_config < 0 || chan_config >= 8) {
+ if (chan_config < 0 || (chan_config >= 8 && chan_config < 11) || chan_config >= 13) {
avpriv_request_sample(avctx, "Unknown ER channel configuration %d",
- ac->oc[1].m4ac.chan_config);
+ chan_config);
return AVERROR_INVALIDDATA;
}
for (i = 0; i < tags_per_config[chan_config]; i++) {
elem_type, elem_id);
return AVERROR_INVALIDDATA;
}
- skip_bits(gb, 4);
+ if (aot != AOT_ER_AAC_ELD)
+ skip_bits(gb, 4);
switch (elem_type) {
case TYPE_SCE:
err = decode_ics(ac, &che->ch[0], gb, 0, 0);
spectral_to_sample(ac);
ac->frame->nb_samples = samples;
+ ac->frame->sample_rate = avctx->sample_rate;
*got_frame_ptr = 1;
skip_bits_long(gb, get_bits_left(gb));
}
}
- if ((err = frame_configure_elements(avctx)) < 0)
- goto fail;
+ if (avctx->channels)
+ if ((err = frame_configure_elements(avctx)) < 0)
+ goto fail;
+
+ // The FF_PROFILE_AAC_* defines are all object_type - 1
+ // This may lead to an undefined profile being signaled
+ ac->avctx->profile = ac->oc[1].m4ac.object_type - 1;
ac->tags_mapped = 0;
// parse
while ((elem_type = get_bits(gb, 3)) != TYPE_END) {
elem_id = get_bits(gb, 4);
+ if (!avctx->channels && elem_type != TYPE_PCE) {
+ err = AVERROR_INVALIDDATA;
+ goto fail;
+ }
+
if (elem_type < TYPE_DSE) {
if (!(che=get_che(ac, elem_type, elem_id))) {
av_log(ac->avctx, AV_LOG_ERROR, "channel element %d.%d is not allocated\n",
}
}
+ if (!avctx->channels) {
+ *got_frame_ptr = 0;
+ return 0;
+ }
+
spectral_to_sample(ac);
multiplier = (ac->oc[1].m4ac.sbr == 1) ? ac->oc[1].m4ac.ext_sample_rate > ac->oc[1].m4ac.sample_rate : 0;
samples <<= multiplier;
- if (samples)
- ac->frame->nb_samples = samples;
- *got_frame_ptr = !!samples;
-
if (ac->oc[1].status && audio_found) {
avctx->sample_rate = ac->oc[1].m4ac.sample_rate << multiplier;
avctx->frame_size = samples;
ac->oc[1].status = OC_LOCKED;
}
+ if (samples) {
+ ac->frame->nb_samples = samples;
+ ac->frame->sample_rate = avctx->sample_rate;
+ }
+ *got_frame_ptr = !!samples;
+
return 0;
fail:
pop_output_configuration(ac);
if (new_extradata) {
av_free(avctx->extradata);
avctx->extradata = av_mallocz(new_extradata_size +
- FF_INPUT_BUFFER_PADDING_SIZE);
+ AV_INPUT_BUFFER_PADDING_SIZE);
if (!avctx->extradata)
return AVERROR(ENOMEM);
avctx->extradata_size = new_extradata_size;
case AOT_ER_AAC_LC:
case AOT_ER_AAC_LTP:
case AOT_ER_AAC_LD:
+ case AOT_ER_AAC_ELD:
err = aac_decode_er_frame(avctx, data, got_frame_ptr, &gb);
break;
default:
ff_mdct_end(&ac->mdct_small);
ff_mdct_end(&ac->mdct_ld);
ff_mdct_end(&ac->mdct_ltp);
+ ff_imdct15_uninit(&ac->mdct480);
return 0;
}
struct LATMContext {
AACContext aac_ctx; ///< containing AACContext
- int initialized; ///< initilized after a valid extradata was seen
+ int initialized; ///< initialized after a valid extradata was seen
// parser data
int audio_mux_version_A; ///< LATM syntax version
if (bits_consumed < 0)
return AVERROR_INVALIDDATA;
- if (ac->oc[1].m4ac.sample_rate != m4ac.sample_rate ||
+ if (!latmctx->initialized ||
+ ac->oc[1].m4ac.sample_rate != m4ac.sample_rate ||
ac->oc[1].m4ac.chan_config != m4ac.chan_config) {
av_log(avctx, AV_LOG_INFO, "audio config changed\n");
if (avctx->extradata_size < esize) {
av_free(avctx->extradata);
- avctx->extradata = av_malloc(esize + FF_INPUT_BUFFER_PADDING_SIZE);
+ avctx->extradata = av_malloc(esize + AV_INPUT_BUFFER_PADDING_SIZE);
if (!avctx->extradata)
return AVERROR(ENOMEM);
}
avctx->extradata_size = esize;
memcpy(avctx->extradata, gb->buffer + (config_start_bit/8), esize);
- memset(avctx->extradata+esize, 0, FF_INPUT_BUFFER_PADDING_SIZE);
+ memset(avctx->extradata+esize, 0, AV_INPUT_BUFFER_PADDING_SIZE);
}
skip_bits_long(gb, bits_consumed);
} else if (!latmctx->aac_ctx.avctx->extradata) {
av_log(latmctx->aac_ctx.avctx, AV_LOG_DEBUG,
"no decoder config found\n");
- return AVERROR(EAGAIN);
+ return 1;
}
if (latmctx->audio_mux_version_A == 0) {
int mux_slot_length_bytes = read_payload_length_info(latmctx, gb);
if (muxlength > avpkt->size)
return AVERROR_INVALIDDATA;
- if ((err = read_audio_mux_element(latmctx, &gb)) < 0)
- return err;
+ if ((err = read_audio_mux_element(latmctx, &gb)))
+ return (err < 0) ? err : avpkt->size;
if (!latmctx->initialized) {
if (!avctx->extradata) {
return AVERROR_INVALIDDATA;
}
- if ((err = aac_decode_frame_int(avctx, out, got_frame_ptr, &gb)) < 0)
+ switch (latmctx->aac_ctx.oc[1].m4ac.object_type) {
+ case AOT_ER_AAC_LC:
+ case AOT_ER_AAC_LTP:
+ case AOT_ER_AAC_LD:
+ case AOT_ER_AAC_ELD:
+ err = aac_decode_er_frame(avctx, out, got_frame_ptr, &gb);
+ break;
+ default:
+ err = aac_decode_frame_int(avctx, out, got_frame_ptr, &gb);
+ }
+ if (err < 0)
return err;
return muxlength;
AVCodec ff_aac_decoder = {
.name = "aac",
+ .long_name = NULL_IF_CONFIG_SMALL("AAC (Advanced Audio Coding)"),
.type = AVMEDIA_TYPE_AUDIO,
.id = AV_CODEC_ID_AAC,
.priv_data_size = sizeof(AACContext),
.init = aac_decode_init,
.close = aac_decode_close,
.decode = aac_decode_frame,
- .long_name = NULL_IF_CONFIG_SMALL("AAC (Advanced Audio Coding)"),
.sample_fmts = (const enum AVSampleFormat[]) {
AV_SAMPLE_FMT_FLTP, AV_SAMPLE_FMT_NONE
},
- .capabilities = CODEC_CAP_CHANNEL_CONF | CODEC_CAP_DR1,
+ .capabilities = AV_CODEC_CAP_CHANNEL_CONF | AV_CODEC_CAP_DR1,
+ .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE,
.channel_layouts = aac_channel_layout,
};
*/
AVCodec ff_aac_latm_decoder = {
.name = "aac_latm",
+ .long_name = NULL_IF_CONFIG_SMALL("AAC LATM (Advanced Audio Coding LATM syntax)"),
.type = AVMEDIA_TYPE_AUDIO,
.id = AV_CODEC_ID_AAC_LATM,
.priv_data_size = sizeof(struct LATMContext),
.init = latm_decode_init,
.close = aac_decode_close,
.decode = latm_decode_frame,
- .long_name = NULL_IF_CONFIG_SMALL("AAC LATM (Advanced Audio Coding LATM syntax)"),
.sample_fmts = (const enum AVSampleFormat[]) {
AV_SAMPLE_FMT_FLTP, AV_SAMPLE_FMT_NONE
},
- .capabilities = CODEC_CAP_CHANNEL_CONF | CODEC_CAP_DR1,
+ .capabilities = AV_CODEC_CAP_CHANNEL_CONF | AV_CODEC_CAP_DR1,
+ .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE,
.channel_layouts = aac_channel_layout,
};