importer/amr_imp.c

   1 /*****************************************************************************
   2  * amr_imp.c
   3  *****************************************************************************
   4  * Copyright (C) 2010-2017 L-SMASH project
   5  *
   6  * Authors: Takashi Hirata <silverfilain@gmail.com>
   7  * Contributors: Yusuke Nakamura <muken.the.vfrmaniac@gmail.com>
   8  *
   9  * Permission to use, copy, modify, and/or distribute this software for any
  10  * purpose with or without fee is hereby granted, provided that the above
  11  * copyright notice and this permission notice appear in all copies.
  12  *
  13  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  14  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  15  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  16  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  17  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  18  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  19  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  20  *****************************************************************************/
  21
  22 /* This file is available under an ISC license. */
  23
  24 #include "common/internal.h" /* must be placed first */
  25
  26 #include <string.h>
  27
  28 #define LSMASH_IMPORTER_INTERNAL
  29 #include "importer.h"
  30
  31 /***************************************************************************
  32     AMR-NB/WB storage format importer
  33     3GPP TS 26.101 V11.0.0 (2012-9)
  34     3GPP TS 26.201 V11.0.0 (2012-9)
  35     3GPP TS 26.244 V12.3.0 (2014-03)
  36     http://www.ietf.org/rfc/rfc3267.txt (Obsoleted)
  37     http://www.ietf.org/rfc/rfc4867.txt
  38 ***************************************************************************/
  39 typedef struct
  40 {
  41     int             wb; /* 0: AMR-NB, 1: AMR-WB */
  42     uint32_t        samples_in_frame;
  43     uint32_t        au_number;
  44 } amr_importer_t;
  45
  46 static void remove_amr_importer
  47 (
  48     amr_importer_t *amr_imp
  49 )
  50 {
  51     lsmash_free( amr_imp );
  52 }
  53
  54 static amr_importer_t *create_amr_importer
  55 (
  56     importer_t *importer
  57 )
  58 {
  59     return (amr_importer_t *)lsmash_malloc_zero( sizeof(amr_importer_t) );
  60 }
  61
  62 static void amr_cleanup
  63 (
  64     importer_t *importer
  65 )
  66 {
  67     debug_if( importer && importer->info )
  68         remove_amr_importer( importer->info );
  69 }
  70
  71 static int amr_get_accessunit
  72 (
  73     importer_t       *importer,
  74     uint32_t          track_number,
  75     lsmash_sample_t **p_sample
  76 )
  77 {
  78     if( !importer->info )
  79         return LSMASH_ERR_NAMELESS;
  80     if( track_number != 1 )
  81         return LSMASH_ERR_FUNCTION_PARAM;
  82     amr_importer_t *amr_imp = (amr_importer_t *)importer->info;
  83     lsmash_bs_t    *bs      = importer->bs;
  84     if( importer->status == IMPORTER_EOF || lsmash_bs_is_end( bs, 0 ) )
  85     {
  86         /* EOF */
  87         importer->status = IMPORTER_EOF;
  88         return IMPORTER_EOF;
  89     }
  90     /* Each speech frame consists of one speech frame header and one speech data.
  91      * At the end of each speech data, octet alignment if needed.
  92      *   Speech frame header
  93      *      0 1 2 3 4 5 6 7
  94      *     +-+-------+-+-+-+
  95      *     |P|  FT   |Q|P|P|
  96      *     +-+-------+-+-+-+
  97      *    FT: Frame type index
  98      *    Q : Frame quality indicator
  99      *    P : Must be set to 0
 100      * FT= 9, 10 and 11 for AMR-NB shall not be used in the file format.
 101      * FT=12, 13 and 14 for AMR-NB are not defined yet in the file format.
 102      * FT=10, 11, 12 and 13 for AMR-WB are not defined yet in the file format.
 103      * FT determines the size of the speech frame starting with it.
 104      */
 105     uint8_t FT = (lsmash_bs_show_byte( bs, 0 ) >> 3) & 0x0F;
 106     const int frame_size[2][16] =
 107     {
 108         { 13, 14, 16, 18, 20, 21, 27, 32,  6, -1, -1, -1, 0, 0, 0, 1 },
 109         { 18, 24, 33, 37, 41, 47, 51, 59, 61,  6,  0,  0, 0, 0, 1, 1 }
 110     };
 111     int read_size = frame_size[ amr_imp->wb ][FT];
 112     if( read_size <= 0 )
 113     {
 114         lsmash_log( importer, LSMASH_LOG_ERROR, "an %s speech frame is detected.\n", read_size < 0 ? "invalid" : "unknown" );
 115         importer->status = IMPORTER_ERROR;
 116         return read_size < 0 ? LSMASH_ERR_INVALID_DATA : LSMASH_ERR_NAMELESS;
 117     }
 118     lsmash_sample_t *sample = lsmash_create_sample( read_size );
 119     if( !sample )
 120         return LSMASH_ERR_MEMORY_ALLOC;
 121     *p_sample = sample;
 122     if( lsmash_bs_get_bytes_ex( bs, read_size, sample->data ) != read_size )
 123     {
 124         lsmash_log( importer, LSMASH_LOG_WARNING, "the stream is truncated at the end.\n" );
 125         importer->status = IMPORTER_EOF;
 126         return LSMASH_ERR_INVALID_DATA;
 127     }
 128     sample->length        = read_size;
 129     sample->dts           = amr_imp->au_number ++ * amr_imp->samples_in_frame;
 130     sample->cts           = sample->dts;
 131     sample->prop.ra_flags = ISOM_SAMPLE_RANDOM_ACCESS_FLAG_SYNC;
 132     return 0;
 133 }
 134
 135 static int amr_check_magic_number
 136 (
 137     lsmash_bs_t *bs
 138 )
 139 {
 140 #define AMR_STORAGE_MAGIC_LENGTH  6
 141 #define AMR_AMRWB_EX_MAGIC_LENGTH 3
 142     /* Check the magic number for single-channel AMR-NB/AMR-WB files.
 143      *   For AMR-NB, "#!AMR\n" (or 0x2321414d520a in hexadecimal).
 144      *   For AMR-WB, "#!AMR-WB\n" (or 0x2321414d522d57420a in hexadecimal).
 145      * Note that AMR-NB and AMR-WB data is stored in the 3GPP/3GPP2 file format according to
 146      * the AMR-NB and AMR-WB storage format for single channel header without the AMR magic numbers. */
 147     uint8_t buf[AMR_STORAGE_MAGIC_LENGTH];
 148     if( lsmash_bs_get_bytes_ex( bs, AMR_STORAGE_MAGIC_LENGTH, buf ) != AMR_STORAGE_MAGIC_LENGTH
 149      || memcmp( buf, "#!AMR", AMR_STORAGE_MAGIC_LENGTH - 1 ) )
 150         return LSMASH_ERR_INVALID_DATA;
 151     if( buf[AMR_STORAGE_MAGIC_LENGTH - 1] == '\n' )
 152         /* single-channel AMR-NB file */
 153         return 0;
 154     if( buf[AMR_STORAGE_MAGIC_LENGTH - 1] != '-'
 155      || lsmash_bs_get_bytes_ex( bs, AMR_AMRWB_EX_MAGIC_LENGTH, buf ) != AMR_AMRWB_EX_MAGIC_LENGTH
 156      || memcmp( buf, "WB\n", AMR_AMRWB_EX_MAGIC_LENGTH ) )
 157         return LSMASH_ERR_INVALID_DATA;
 158     /* single-channel AMR-WB file */
 159     return 1;
 160 #undef AMR_STORAGE_MAGIC_LENGTH
 161 #undef AMR_AMRWB_EX_MAGIC_LENGTH
 162 }
 163
 164 static int amr_create_damr
 165 (
 166     lsmash_audio_summary_t *summary,
 167     int                     wb
 168 )
 169 {
 170 #define AMR_DAMR_LENGTH 17
 171     lsmash_bs_t *bs = lsmash_bs_create();
 172     if( !bs )
 173         return LSMASH_ERR_MEMORY_ALLOC;
 174     lsmash_bs_put_be32( bs, AMR_DAMR_LENGTH );
 175     lsmash_bs_put_be32( bs, ISOM_BOX_TYPE_DAMR.fourcc );
 176     /* NOTE: These are specific to each codec vendor, but we're surely not a vendor.
 177      *       Using dummy data. */
 178     lsmash_bs_put_be32( bs, 0x20202020 );           /* vendor */
 179     lsmash_bs_put_byte( bs, 0 );                    /* decoder_version */
 180     /* NOTE: Using safe value for these settings, maybe sub-optimal. */
 181     lsmash_bs_put_be16( bs, wb ? 0xC3FF : 0x81FF ); /* mode_set, represents for all possibly existing and supported frame-types. */
 182     lsmash_bs_put_byte( bs, 1 );                    /* mode_change_period */
 183     lsmash_bs_put_byte( bs, 1 );                    /* frames_per_sample */
 184     lsmash_codec_specific_t *cs = lsmash_malloc_zero( sizeof(lsmash_codec_specific_t) );
 185     if( !cs )
 186     {
 187         lsmash_bs_cleanup( bs );
 188         return LSMASH_ERR_MEMORY_ALLOC;
 189     }
 190     cs->type              = LSMASH_CODEC_SPECIFIC_DATA_TYPE_UNKNOWN;
 191     cs->format            = LSMASH_CODEC_SPECIFIC_FORMAT_UNSTRUCTURED;
 192     cs->destruct          = (lsmash_codec_specific_destructor_t)lsmash_free;
 193     cs->data.unstructured = lsmash_bs_export_data( bs, &cs->size );
 194     cs->size              = AMR_DAMR_LENGTH;
 195     lsmash_bs_cleanup( bs );
 196     if( !cs->data.unstructured
 197      || lsmash_list_add_entry( &summary->opaque->list, cs ) < 0 )
 198     {
 199         lsmash_destroy_codec_specific_data( cs );
 200         return LSMASH_ERR_MEMORY_ALLOC;
 201     }
 202     return 0;
 203 #undef AMR_DAMR_LENGTH
 204 }
 205
 206 static lsmash_audio_summary_t *amr_create_summary
 207 (
 208     importer_t *importer,
 209     int         wb
 210 )
 211 {
 212     /* Establish an audio summary for AMR-NB or AMR-WB stream. */
 213     lsmash_audio_summary_t *summary = (lsmash_audio_summary_t *)lsmash_create_summary( LSMASH_SUMMARY_TYPE_AUDIO );
 214     if( !summary )
 215         return NULL;
 216     summary->sample_type      = wb ? ISOM_CODEC_TYPE_SAWB_AUDIO : ISOM_CODEC_TYPE_SAMR_AUDIO;
 217     summary->max_au_length    = wb ? 61 : 32;
 218     summary->aot              = MP4A_AUDIO_OBJECT_TYPE_NULL;    /* no effect */
 219     summary->frequency        = (8000 << wb);
 220     summary->channels         = 1;                              /* always single channel */
 221     summary->sample_size      = 16;
 222     summary->samples_in_frame = (160 << wb);
 223     summary->sbr_mode         = MP4A_AAC_SBR_NOT_SPECIFIED;     /* no effect */
 224     if( amr_create_damr( summary, wb ) < 0
 225      || lsmash_list_add_entry( importer->summaries, summary ) < 0 )
 226     {
 227         lsmash_cleanup_summary( (lsmash_summary_t *)summary );
 228         return NULL;
 229     }
 230     return summary;
 231 }
 232
 233 static int amr_probe
 234 (
 235     importer_t *importer
 236 )
 237 {
 238
 239     amr_importer_t *amr_imp = create_amr_importer( importer );
 240     if( !amr_imp )
 241         return LSMASH_ERR_MEMORY_ALLOC;
 242     int err;
 243     int wb = amr_check_magic_number( importer->bs );
 244     if( wb < 0 )
 245     {
 246         err = wb;
 247         goto fail;
 248     }
 249     lsmash_audio_summary_t *summary = amr_create_summary( importer, wb );
 250     if( !summary )
 251     {
 252         err = LSMASH_ERR_NAMELESS;
 253         goto fail;
 254     }
 255     amr_imp->wb               = wb;
 256     amr_imp->samples_in_frame = summary->samples_in_frame;
 257     amr_imp->au_number        = 0;
 258     importer->info   = amr_imp;
 259     importer->status = IMPORTER_OK;
 260     return 0;
 261 fail:
 262     remove_amr_importer( amr_imp );
 263     return err;
 264 }
 265
 266 static uint32_t amr_get_last_delta
 267 (
 268     importer_t *importer,
 269     uint32_t    track_number
 270 )
 271 {
 272     debug_if( !importer || !importer->info )
 273         return 0;
 274     amr_importer_t *amr_imp = (amr_importer_t *)importer->info;
 275     if( !amr_imp || track_number != 1 )
 276         return 0;
 277     return amr_imp->samples_in_frame;
 278 }
 279
 280 const importer_functions amr_importer =
 281 {
 282     { "AMR", offsetof( importer_t, log_level ) },
 283     1,
 284     amr_probe,
 285     amr_get_accessunit,
 286     amr_get_last_delta,
 287     amr_cleanup
 288 };