latmenc.c 7.3 KB
Newer Older
Kieran Kunhya's avatar
Kieran Kunhya committed
1 2 3 4
/*
 * LATM/LOAS muxer
 * Copyright (c) 2011 Kieran Kunhya <kieran@kunhya.com>
 *
5
 * This file is part of FFmpeg.
Kieran Kunhya's avatar
Kieran Kunhya committed
6
 *
7
 * FFmpeg is free software; you can redistribute it and/or
Kieran Kunhya's avatar
Kieran Kunhya committed
8 9 10 11
 * 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.
 *
12
 * FFmpeg is distributed in the hope that it will be useful,
Kieran Kunhya's avatar
Kieran Kunhya committed
13 14 15 16 17
 * 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
18
 * License along with FFmpeg; if not, write to the Free Software
Kieran Kunhya's avatar
Kieran Kunhya committed
19 20 21 22 23 24 25 26 27
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

#include "libavcodec/get_bits.h"
#include "libavcodec/put_bits.h"
#include "libavcodec/avcodec.h"
#include "libavcodec/mpeg4audio.h"
#include "libavutil/opt.h"
#include "avformat.h"
28
#include "rawenc.h"
Kieran Kunhya's avatar
Kieran Kunhya committed
29

30 31
#define MAX_EXTRADATA_SIZE 1024

Kieran Kunhya's avatar
Kieran Kunhya committed
32
typedef struct {
33
    AVClass *av_class;
Kieran Kunhya's avatar
Kieran Kunhya committed
34 35 36 37 38
    int off;
    int channel_conf;
    int object_type;
    int counter;
    int mod;
39
    uint8_t buffer[0x1fff + MAX_EXTRADATA_SIZE + 1024];
Kieran Kunhya's avatar
Kieran Kunhya committed
40 41 42 43
} LATMContext;

static const AVOption options[] = {
    {"smc-interval", "StreamMuxConfig interval.",
44
     offsetof(LATMContext, mod), AV_OPT_TYPE_INT, {.i64 = 0x0014}, 0x0001, 0xffff, AV_OPT_FLAG_ENCODING_PARAM},
Kieran Kunhya's avatar
Kieran Kunhya committed
45 46 47 48 49 50 51 52 53 54 55 56 57 58
    {NULL},
};

static const AVClass latm_muxer_class = {
    .class_name = "LATM/LOAS muxer",
    .item_name  = av_default_item_name,
    .option     = options,
    .version    = LIBAVUTIL_VERSION_INT,
};

static int latm_decode_extradata(LATMContext *ctx, uint8_t *buf, int size)
{
    MPEG4AudioConfig m4ac;

59 60 61 62
    if (size > MAX_EXTRADATA_SIZE) {
        av_log(ctx, AV_LOG_ERROR, "Extradata is larger than currently supported.\n");
        return AVERROR_INVALIDDATA;
    }
63
    ctx->off = avpriv_mpeg4audio_get_config(&m4ac, buf, size * 8, 1);
Kieran Kunhya's avatar
Kieran Kunhya committed
64 65 66
    if (ctx->off < 0)
        return ctx->off;

67 68 69 70 71
    if (ctx->object_type == AOT_ALS && (ctx->off & 7)) {
        // as long as avpriv_mpeg4audio_get_config works correctly this is impossible
        av_log(ctx, AV_LOG_ERROR, "BUG: ALS offset is not byte-aligned\n");
        return AVERROR_INVALIDDATA;
    }
Kieran Kunhya's avatar
Kieran Kunhya committed
72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88
    /* FIXME: are any formats not allowed in LATM? */

    if (m4ac.object_type > AOT_SBR && m4ac.object_type != AOT_ALS) {
        av_log(ctx, AV_LOG_ERROR, "Muxing MPEG-4 AOT %d in LATM is not supported\n", m4ac.object_type);
        return AVERROR_INVALIDDATA;
    }
    ctx->channel_conf = m4ac.chan_config;
    ctx->object_type  = m4ac.object_type;

    return 0;
}

static int latm_write_header(AVFormatContext *s)
{
    LATMContext *ctx = s->priv_data;
    AVCodecContext *avctx = s->streams[0]->codec;

89
    if (avctx->codec_id == AV_CODEC_ID_AAC_LATM)
90 91
        return 0;

Kieran Kunhya's avatar
Kieran Kunhya committed
92 93 94 95 96 97 98
    if (avctx->extradata_size > 0 &&
        latm_decode_extradata(ctx, avctx->extradata, avctx->extradata_size) < 0)
        return AVERROR_INVALIDDATA;

    return 0;
}

99
static void latm_write_frame_header(AVFormatContext *s, PutBitContext *bs)
Kieran Kunhya's avatar
Kieran Kunhya committed
100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117
{
    LATMContext *ctx = s->priv_data;
    AVCodecContext *avctx = s->streams[0]->codec;
    int header_size;

    /* AudioMuxElement */
    put_bits(bs, 1, !!ctx->counter);

    if (!ctx->counter) {
        /* StreamMuxConfig */
        put_bits(bs, 1, 0); /* audioMuxVersion */
        put_bits(bs, 1, 1); /* allStreamsSameTimeFraming */
        put_bits(bs, 6, 0); /* numSubFrames */
        put_bits(bs, 4, 0); /* numProgram */
        put_bits(bs, 3, 0); /* numLayer */

        /* AudioSpecificConfig */
        if (ctx->object_type == AOT_ALS) {
118 119
            header_size = avctx->extradata_size-(ctx->off >> 3);
            avpriv_copy_bits(bs, &avctx->extradata[ctx->off >> 3], header_size);
Kieran Kunhya's avatar
Kieran Kunhya committed
120
        } else {
121 122
            // + 3 assumes not scalable and dependsOnCoreCoder == 0,
            // see decode_ga_specific_config in libavcodec/aacdec.c
123
            avpriv_copy_bits(bs, avctx->extradata, ctx->off + 3);
Kieran Kunhya's avatar
Kieran Kunhya committed
124 125

            if (!ctx->channel_conf) {
126
                GetBitContext gb;
127
                init_get_bits8(&gb, avctx->extradata, avctx->extradata_size);
128
                skip_bits_long(&gb, ctx->off + 3);
129
                avpriv_copy_pce_data(bs, &gb);
Kieran Kunhya's avatar
Kieran Kunhya committed
130 131 132 133
            }
        }

        put_bits(bs, 3, 0); /* frameLengthType */
134
        put_bits(bs, 8, 0xff); /* latmBufferFullness */
Kieran Kunhya's avatar
Kieran Kunhya committed
135 136 137 138 139 140 141 142 143 144 145

        put_bits(bs, 1, 0); /* otherDataPresent */
        put_bits(bs, 1, 0); /* crcCheckPresent */
    }

    ctx->counter++;
    ctx->counter %= ctx->mod;
}

static int latm_write_packet(AVFormatContext *s, AVPacket *pkt)
{
146
    LATMContext *ctx = s->priv_data;
Kieran Kunhya's avatar
Kieran Kunhya committed
147 148 149 150 151
    AVIOContext *pb = s->pb;
    PutBitContext bs;
    int i, len;
    uint8_t loas_header[] = "\x56\xe0\x00";

152
    if (s->streams[0]->codec->codec_id == AV_CODEC_ID_AAC_LATM)
153 154
        return ff_raw_write_packet(s, pkt);

Kieran Kunhya's avatar
Kieran Kunhya committed
155 156 157 158
    if (pkt->size > 2 && pkt->data[0] == 0xff && (pkt->data[1] >> 4) == 0xf) {
        av_log(s, AV_LOG_ERROR, "ADTS header detected - ADTS will not be incorrectly muxed into LATM\n");
        return AVERROR_INVALIDDATA;
    }
159 160 161 162 163 164 165 166 167

    if (!s->streams[0]->codec->extradata) {
        if(pkt->size > 2 && pkt->data[0] == 0x56 && (pkt->data[1] >> 4) == 0xe &&
            (AV_RB16(pkt->data + 1) & 0x1FFF) + 3 == pkt->size)
            return ff_raw_write_packet(s, pkt);
        else
            return AVERROR_INVALIDDATA;
    }

168 169
    if (pkt->size > 0x1fff)
        goto too_large;
Kieran Kunhya's avatar
Kieran Kunhya committed
170

171
    init_put_bits(&bs, ctx->buffer, pkt->size+1024+MAX_EXTRADATA_SIZE);
Kieran Kunhya's avatar
Kieran Kunhya committed
172 173 174 175 176 177 178 179 180 181 182

    latm_write_frame_header(s, &bs);

    /* PayloadLengthInfo() */
    for (i = 0; i <= pkt->size-255; i+=255)
        put_bits(&bs, 8, 255);

    put_bits(&bs, 8, pkt->size-i);

    /* The LATM payload is written unaligned */

183
    /* PayloadMux() */
184 185 186 187 188 189 190 191 192 193
    if (pkt->size && (pkt->data[0] & 0xe1) == 0x81) {
        // Convert byte-aligned DSE to non-aligned.
        // Due to the input format encoding we know that
        // it is naturally byte-aligned in the input stream,
        // so there are no padding bits to account for.
        // To avoid having to add padding bits and rearrange
        // the whole stream we just remove the byte-align flag.
        // This allows us to remux our FATE AAC samples into latm
        // files that are still playable with minimal effort.
        put_bits(&bs, 8, pkt->data[0] & 0xfe);
194 195 196
        avpriv_copy_bits(&bs, pkt->data + 1, 8*pkt->size - 8);
    } else
        avpriv_copy_bits(&bs, pkt->data, 8*pkt->size);
Kieran Kunhya's avatar
Kieran Kunhya committed
197

198
    avpriv_align_put_bits(&bs);
Kieran Kunhya's avatar
Kieran Kunhya committed
199 200 201 202
    flush_put_bits(&bs);

    len = put_bits_count(&bs) >> 3;

203 204 205
    if (len > 0x1fff)
        goto too_large;

Kieran Kunhya's avatar
Kieran Kunhya committed
206 207 208 209
    loas_header[1] |= (len >> 8) & 0x1f;
    loas_header[2] |= len & 0xff;

    avio_write(pb, loas_header, 3);
210
    avio_write(pb, ctx->buffer, len);
Kieran Kunhya's avatar
Kieran Kunhya committed
211 212

    return 0;
213 214 215 216

too_large:
    av_log(s, AV_LOG_ERROR, "LATM packet size larger than maximum size 0x1fff\n");
    return AVERROR_INVALIDDATA;
Kieran Kunhya's avatar
Kieran Kunhya committed
217 218 219 220 221 222
}

AVOutputFormat ff_latm_muxer = {
    .name           = "latm",
    .long_name      = NULL_IF_CONFIG_SMALL("LOAS/LATM"),
    .mime_type      = "audio/MP4A-LATM",
223
    .extensions     = "latm,loas",
Kieran Kunhya's avatar
Kieran Kunhya committed
224
    .priv_data_size = sizeof(LATMContext),
225 226
    .audio_codec    = AV_CODEC_ID_AAC,
    .video_codec    = AV_CODEC_ID_NONE,
Kieran Kunhya's avatar
Kieran Kunhya committed
227 228 229
    .write_header   = latm_write_header,
    .write_packet   = latm_write_packet,
    .priv_class     = &latm_muxer_class,
230
    .flags          = AVFMT_NOTIMESTAMPS,
Kieran Kunhya's avatar
Kieran Kunhya committed
231
};