FFmpeg
latmenc.c
Go to the documentation of this file.
1 /*
2  * LATM/LOAS muxer
3  * Copyright (c) 2011 Kieran Kunhya <kieran@kunhya.com>
4  *
5  * This file is part of FFmpeg.
6  *
7  * FFmpeg is free software; you can redistribute it and/or
8  * modify it under the terms of the GNU Lesser General Public
9  * License as published by the Free Software Foundation; either
10  * version 2.1 of the License, or (at your option) any later version.
11  *
12  * FFmpeg is distributed in the hope that it will be useful,
13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15  * Lesser General Public License for more details.
16  *
17  * You should have received a copy of the GNU Lesser General Public
18  * License along with FFmpeg; if not, write to the Free Software
19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20  */
21 
22 #include "libavcodec/get_bits.h"
23 #include "libavcodec/put_bits.h"
24 #include "libavcodec/avcodec.h"
25 #include "libavcodec/mpeg4audio.h"
26 #include "libavutil/opt.h"
27 #include "avformat.h"
28 #include "internal.h"
29 #include "rawenc.h"
30 
31 #define MAX_EXTRADATA_SIZE 1024
32 
33 typedef struct LATMContext {
35  int off;
38  int counter;
39  int mod;
40  uint8_t buffer[0x1fff + MAX_EXTRADATA_SIZE + 1024];
41 } LATMContext;
42 
43 static const AVOption options[] = {
44  {"smc-interval", "StreamMuxConfig interval.",
45  offsetof(LATMContext, mod), AV_OPT_TYPE_INT, {.i64 = 0x0014}, 0x0001, 0xffff, AV_OPT_FLAG_ENCODING_PARAM},
46  {NULL},
47 };
48 
49 static const AVClass latm_muxer_class = {
50  .class_name = "LATM/LOAS muxer",
51  .item_name = av_default_item_name,
52  .option = options,
53  .version = LIBAVUTIL_VERSION_INT,
54 };
55 
57 {
59  MPEG4AudioConfig m4ac;
60 
61  if (size > MAX_EXTRADATA_SIZE) {
62  av_log(s, AV_LOG_ERROR, "Extradata is larger than currently supported.\n");
63  return AVERROR_INVALIDDATA;
64  }
65  ctx->off = avpriv_mpeg4audio_get_config2(&m4ac, buf, size, 1, s);
66  if (ctx->off < 0)
67  return ctx->off;
68 
69  if (ctx->object_type == AOT_ALS && (ctx->off & 7)) {
70  // as long as avpriv_mpeg4audio_get_config works correctly this is impossible
71  av_log(s, AV_LOG_ERROR, "BUG: ALS offset is not byte-aligned\n");
72  return AVERROR_INVALIDDATA;
73  }
74  /* FIXME: are any formats not allowed in LATM? */
75 
76  if (m4ac.object_type > AOT_SBR && m4ac.object_type != AOT_ALS) {
77  av_log(s, AV_LOG_ERROR, "Muxing MPEG-4 AOT %d in LATM is not supported\n", m4ac.object_type);
78  return AVERROR_INVALIDDATA;
79  }
80  ctx->channel_conf = m4ac.chan_config;
81  ctx->object_type = m4ac.object_type;
82 
83  return 0;
84 }
85 
87 {
88  AVCodecParameters *par = s->streams[0]->codecpar;
89 
90  if (par->codec_id == AV_CODEC_ID_AAC_LATM)
91  return 0;
92  if (par->codec_id != AV_CODEC_ID_AAC && par->codec_id != AV_CODEC_ID_MP4ALS) {
93  av_log(s, AV_LOG_ERROR, "Only AAC, LATM and ALS are supported\n");
94  return AVERROR(EINVAL);
95  }
96 
97  if (par->extradata_size > 0 &&
99  return AVERROR_INVALIDDATA;
100 
101  return 0;
102 }
103 
105 {
106  LATMContext *ctx = s->priv_data;
107  AVCodecParameters *par = s->streams[0]->codecpar;
108  int header_size;
109 
110  /* AudioMuxElement */
111  put_bits(bs, 1, !!ctx->counter);
112 
113  if (!ctx->counter) {
114  /* StreamMuxConfig */
115  put_bits(bs, 1, 0); /* audioMuxVersion */
116  put_bits(bs, 1, 1); /* allStreamsSameTimeFraming */
117  put_bits(bs, 6, 0); /* numSubFrames */
118  put_bits(bs, 4, 0); /* numProgram */
119  put_bits(bs, 3, 0); /* numLayer */
120 
121  /* AudioSpecificConfig */
122  if (ctx->object_type == AOT_ALS) {
123  header_size = par->extradata_size-(ctx->off >> 3);
124  avpriv_copy_bits(bs, &par->extradata[ctx->off >> 3], header_size);
125  } else {
126  // + 3 assumes not scalable and dependsOnCoreCoder == 0,
127  // see decode_ga_specific_config in libavcodec/aacdec.c
128  avpriv_copy_bits(bs, par->extradata, ctx->off + 3);
129 
130  if (!ctx->channel_conf) {
131  GetBitContext gb;
132  int ret = init_get_bits8(&gb, par->extradata, par->extradata_size);
133  av_assert0(ret >= 0); // extradata size has been checked already, so this should not fail
134  skip_bits_long(&gb, ctx->off + 3);
135  ff_copy_pce_data(bs, &gb);
136  }
137  }
138 
139  put_bits(bs, 3, 0); /* frameLengthType */
140  put_bits(bs, 8, 0xff); /* latmBufferFullness */
141 
142  put_bits(bs, 1, 0); /* otherDataPresent */
143  put_bits(bs, 1, 0); /* crcCheckPresent */
144  }
145 
146  ctx->counter++;
147  ctx->counter %= ctx->mod;
148 }
149 
151 {
152  LATMContext *ctx = s->priv_data;
153  AVCodecParameters *par = s->streams[0]->codecpar;
154  AVIOContext *pb = s->pb;
155  PutBitContext bs;
156  int i, len;
157  uint8_t loas_header[] = "\x56\xe0\x00";
158 
159  if (par->codec_id == AV_CODEC_ID_AAC_LATM)
160  return ff_raw_write_packet(s, pkt);
161 
162  if (!par->extradata) {
163  if(pkt->size > 2 && pkt->data[0] == 0x56 && (pkt->data[1] >> 4) == 0xe &&
164  (AV_RB16(pkt->data + 1) & 0x1FFF) + 3 == pkt->size)
165  return ff_raw_write_packet(s, pkt);
166  else {
167  uint8_t *side_data;
168  int side_data_size = 0, ret;
169 
171  &side_data_size);
172  if (side_data_size) {
173  if (latm_decode_extradata(s, side_data, side_data_size) < 0)
174  return AVERROR_INVALIDDATA;
175  ret = ff_alloc_extradata(par, side_data_size);
176  if (ret < 0)
177  return ret;
178  memcpy(par->extradata, side_data, side_data_size);
179  } else
180  return AVERROR_INVALIDDATA;
181  }
182  }
183 
184  if (pkt->size > 0x1fff)
185  goto too_large;
186 
187  init_put_bits(&bs, ctx->buffer, pkt->size+1024+MAX_EXTRADATA_SIZE);
188 
189  latm_write_frame_header(s, &bs);
190 
191  /* PayloadLengthInfo() */
192  for (i = 0; i <= pkt->size-255; i+=255)
193  put_bits(&bs, 8, 255);
194 
195  put_bits(&bs, 8, pkt->size-i);
196 
197  /* The LATM payload is written unaligned */
198 
199  /* PayloadMux() */
200  if (pkt->size && (pkt->data[0] & 0xe1) == 0x81) {
201  // Convert byte-aligned DSE to non-aligned.
202  // Due to the input format encoding we know that
203  // it is naturally byte-aligned in the input stream,
204  // so there are no padding bits to account for.
205  // To avoid having to add padding bits and rearrange
206  // the whole stream we just remove the byte-align flag.
207  // This allows us to remux our FATE AAC samples into latm
208  // files that are still playable with minimal effort.
209  put_bits(&bs, 8, pkt->data[0] & 0xfe);
210  avpriv_copy_bits(&bs, pkt->data + 1, 8*pkt->size - 8);
211  } else
212  avpriv_copy_bits(&bs, pkt->data, 8*pkt->size);
213 
215  flush_put_bits(&bs);
216 
217  len = put_bits_count(&bs) >> 3;
218 
219  if (len > 0x1fff)
220  goto too_large;
221 
222  loas_header[1] |= (len >> 8) & 0x1f;
223  loas_header[2] |= len & 0xff;
224 
225  avio_write(pb, loas_header, 3);
226  avio_write(pb, ctx->buffer, len);
227 
228  return 0;
229 
230 too_large:
231  av_log(s, AV_LOG_ERROR, "LATM packet size larger than maximum size 0x1fff\n");
232  return AVERROR_INVALIDDATA;
233 }
234 
235 static int latm_check_bitstream(struct AVFormatContext *s, const AVPacket *pkt)
236 {
237  int ret = 1;
238  AVStream *st = s->streams[pkt->stream_index];
239 
240  if (st->codecpar->codec_id == AV_CODEC_ID_AAC) {
241  if (pkt->size > 2 && (AV_RB16(pkt->data) & 0xfff0) == 0xfff0)
242  ret = ff_stream_add_bitstream_filter(st, "aac_adtstoasc", NULL);
243  }
244 
245  return ret;
246 }
247 
249  .name = "latm",
250  .long_name = NULL_IF_CONFIG_SMALL("LOAS/LATM"),
251  .mime_type = "audio/MP4A-LATM",
252  .extensions = "latm,loas",
253  .priv_data_size = sizeof(LATMContext),
254  .audio_codec = AV_CODEC_ID_AAC,
255  .video_codec = AV_CODEC_ID_NONE,
258  .priv_class = &latm_muxer_class,
259  .check_bitstream= latm_check_bitstream,
261 };
static void write_packet(OutputFile *of, AVPacket *pkt, OutputStream *ost, int unqueue)
Definition: ffmpeg.c:689
#define NULL
Definition: coverity.c:32
Bytestream IO Context.
Definition: avio.h:161
#define AVERROR_INVALIDDATA
Invalid data found when processing input.
Definition: error.h:59
static const AVOption options[]
Definition: latmenc.c:43
AVOption.
Definition: opt.h:246
static void put_bits(Jpeg2000EncoderContext *s, int val, int n)
put n times val bit
Definition: j2kenc.c:208
#define LIBAVUTIL_VERSION_INT
Definition: version.h:85
static void skip_bits_long(GetBitContext *s, int n)
Skips the specified number of bits.
Definition: get_bits.h:291
enum AVCodecID codec_id
Specific type of the encoded data (the codec used).
Definition: avcodec.h:3968
int size
Definition: avcodec.h:1481
const char * av_default_item_name(void *ptr)
Return the context name.
Definition: log.c:191
void avpriv_copy_bits(PutBitContext *pb, const uint8_t *src, int length)
Copy the content of src to the bitstream.
Definition: bitstream.c:64
void avpriv_align_put_bits(PutBitContext *s)
Pad the bitstream with zeros up to the next byte boundary.
Definition: bitstream.c:48
static const AVClass latm_muxer_class
Definition: latmenc.c:49
static AVPacket pkt
uint64_t_TMPL AV_WL64 unsigned int_TMPL AV_WL32 unsigned int_TMPL AV_WL24 unsigned int_TMPL AV_WL16 uint64_t_TMPL AV_WB64 unsigned int_TMPL AV_WB32 unsigned int_TMPL AV_WB24 unsigned int_TMPL AV_RB16
Definition: bytestream.h:87
This struct describes the properties of an encoded stream.
Definition: avcodec.h:3960
Format I/O context.
Definition: avformat.h:1358
int counter
Definition: latmenc.c:38
const char * class_name
The name of the class; usually it is the same name as the context structure type to which the AVClass...
Definition: log.h:72
#define av_assert0(cond)
assert() equivalent, that is always enabled.
Definition: avassert.h:37
uint8_t
int avpriv_mpeg4audio_get_config2(MPEG4AudioConfig *c, const uint8_t *buf, int size, int sync_extension, void *logctx)
Parse MPEG-4 systems extradata from a raw buffer to retrieve audio configuration. ...
Definition: mpeg4audio.c:177
AVOptions.
AVStream ** streams
A list of all streams in the file.
Definition: avformat.h:1426
AVClass * av_class
Definition: latmenc.c:34
uint8_t * data
Definition: avcodec.h:1480
bitstream reader API header.
ptrdiff_t size
Definition: opengl_enc.c:100
void avio_write(AVIOContext *s, const unsigned char *buf, int size)
Definition: aviobuf.c:218
#define av_log(a,...)
#define AV_OPT_FLAG_ENCODING_PARAM
a generic parameter which can be set by the user for muxing or encoding
Definition: opt.h:276
#define i(width, name, range_min, range_max)
Definition: cbs_h2645.c:259
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:176
uint8_t * av_packet_get_side_data(const AVPacket *pkt, enum AVPacketSideDataType type, int *size)
Get side information from packet.
Definition: avpacket.c:350
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
Definition: internal.h:186
static int put_bits_count(PutBitContext *s)
Definition: put_bits.h:85
int extradata_size
Size of the extradata content in bytes.
Definition: avcodec.h:3986
int channel_conf
Definition: latmenc.c:36
int ff_alloc_extradata(AVCodecParameters *par, int size)
Allocate extradata with additional AV_INPUT_BUFFER_PADDING_SIZE at end which is always set to 0...
Definition: utils.c:3321
const char * name
Definition: avformat.h:505
AVFormatContext * ctx
Definition: movenc.c:48
#define s(width, name)
Definition: cbs_vp9.c:257
int ff_raw_write_packet(AVFormatContext *s, AVPacket *pkt)
Definition: rawenc.c:29
static int latm_decode_extradata(AVFormatContext *s, uint8_t *buf, int size)
Definition: latmenc.c:56
Stream structure.
Definition: avformat.h:881
static int latm_write_packet(AVFormatContext *s, AVPacket *pkt)
Definition: latmenc.c:150
#define AVFMT_NOTIMESTAMPS
Format does not need / have any timestamps.
Definition: avformat.h:467
The AV_PKT_DATA_NEW_EXTRADATA is used to notify the codec or the format that the extradata buffer was...
Definition: avcodec.h:1202
Libavcodec external API header.
AVIOContext * pb
I/O context.
Definition: avformat.h:1400
static int init_get_bits8(GetBitContext *s, const uint8_t *buffer, int byte_size)
Initialize GetBitContext.
Definition: get_bits.h:677
int mod
Definition: latmenc.c:39
#define MAX_EXTRADATA_SIZE
Definition: latmenc.c:31
void * buf
Definition: avisynth_c.h:766
static int latm_write_header(AVFormatContext *s)
Definition: latmenc.c:86
Describe the class of an AVClass context structure.
Definition: log.h:67
int object_type
Definition: latmenc.c:37
static int ff_copy_pce_data(PutBitContext *pb, GetBitContext *gb)
Definition: mpeg4audio.h:146
#define flags(name, subs,...)
Definition: cbs_av1.c:561
Y Spectral Band Replication.
Definition: mpeg4audio.h:94
Main libavformat public API header.
int off
Definition: latmenc.c:35
static void flush_put_bits(PutBitContext *s)
Pad the end of the output stream with zeros.
Definition: put_bits.h:101
static void init_put_bits(PutBitContext *s, uint8_t *buffer, int buffer_size)
Initialize the PutBitContext s.
Definition: put_bits.h:48
AVOutputFormat ff_latm_muxer
Definition: latmenc.c:248
Y Audio LosslesS.
Definition: mpeg4audio.h:122
int len
static void latm_write_frame_header(AVFormatContext *s, PutBitContext *bs)
Definition: latmenc.c:104
int ff_stream_add_bitstream_filter(AVStream *st, const char *name, const char *args)
Add a bitstream filter to a stream.
Definition: utils.c:5556
void * priv_data
Format private data.
Definition: avformat.h:1386
static void write_header(FFV1Context *f)
Definition: ffv1enc.c:349
uint8_t * extradata
Extra binary data needed for initializing the decoder, codec-dependent.
Definition: avcodec.h:3982
AVCodecParameters * codecpar
Codec parameters associated with this stream.
Definition: avformat.h:1028
uint8_t buffer[0x1fff+MAX_EXTRADATA_SIZE+1024]
Definition: latmenc.c:40
int stream_index
Definition: avcodec.h:1482
Filter the word “frame” indicates either a video frame or a group of audio as stored in an AVFrame structure Format for each input and each output the list of supported formats For video that means pixel format For audio that means channel sample they are references to shared objects When the negotiation mechanism computes the intersection of the formats supported at each end of a all references to both lists are replaced with a reference to the intersection And when a single format is eventually chosen for a link amongst the remaining all references to the list are updated That means that if a filter requires that its input and output have the same format amongst a supported all it has to do is use a reference to the same list of formats query_formats can leave some formats unset and return AVERROR(EAGAIN) to cause the negotiation mechanism toagain later.That can be used by filters with complex requirements to use the format negotiated on one link to set the formats supported on another.Frame references ownership and permissions
This structure stores compressed data.
Definition: avcodec.h:1457
static int latm_check_bitstream(struct AVFormatContext *s, const AVPacket *pkt)
Definition: latmenc.c:235
GLuint buffer
Definition: opengl_enc.c:101
bitstream writer API