| 1 | /* |
| 2 | * AAC decoder wrapper |
| 3 | * Copyright (c) 2012 Martin Storsjo |
| 4 | * |
| 5 | * This file is part of FFmpeg. |
| 6 | * |
| 7 | * Permission to use, copy, modify, and/or distribute this software for any |
| 8 | * purpose with or without fee is hereby granted, provided that the above |
| 9 | * copyright notice and this permission notice appear in all copies. |
| 10 | * |
| 11 | * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES |
| 12 | * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF |
| 13 | * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR |
| 14 | * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES |
| 15 | * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN |
| 16 | * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF |
| 17 | * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. |
| 18 | */ |
| 19 | |
| 20 | #include <fdk-aac/aacdecoder_lib.h> |
| 21 | |
| 22 | #include "libavutil/channel_layout.h" |
| 23 | #include "libavutil/common.h" |
| 24 | #include "libavutil/opt.h" |
| 25 | #include "avcodec.h" |
| 26 | #include "internal.h" |
| 27 | |
| 28 | /* The version macro is introduced the same time as the setting enum was |
| 29 | * changed, so this check should suffice. */ |
| 30 | #ifndef AACDECODER_LIB_VL0 |
| 31 | #define AAC_PCM_MAX_OUTPUT_CHANNELS AAC_PCM_OUTPUT_CHANNELS |
| 32 | #endif |
| 33 | |
| 34 | enum ConcealMethod { |
| 35 | CONCEAL_METHOD_SPECTRAL_MUTING = 0, |
| 36 | CONCEAL_METHOD_NOISE_SUBSTITUTION = 1, |
| 37 | CONCEAL_METHOD_ENERGY_INTERPOLATION = 2, |
| 38 | CONCEAL_METHOD_NB, |
| 39 | }; |
| 40 | |
| 41 | typedef struct FDKAACDecContext { |
| 42 | const AVClass *class; |
| 43 | HANDLE_AACDECODER handle; |
| 44 | int initialized; |
| 45 | uint8_t *decoder_buffer; |
| 46 | uint8_t *anc_buffer; |
| 47 | enum ConcealMethod conceal_method; |
| 48 | int drc_level; |
| 49 | int drc_boost; |
| 50 | int drc_heavy; |
| 51 | int drc_cut; |
| 52 | } FDKAACDecContext; |
| 53 | |
| 54 | |
| 55 | #define DMX_ANC_BUFFSIZE 128 |
| 56 | #define DECODER_MAX_CHANNELS 6 |
| 57 | #define DECODER_BUFFSIZE 2048 * sizeof(INT_PCM) |
| 58 | |
| 59 | #define OFFSET(x) offsetof(FDKAACDecContext, x) |
| 60 | #define AD AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_DECODING_PARAM |
| 61 | static const AVOption fdk_aac_dec_options[] = { |
| 62 | { "conceal", "Error concealment method", OFFSET(conceal_method), AV_OPT_TYPE_INT, { .i64 = CONCEAL_METHOD_NOISE_SUBSTITUTION }, CONCEAL_METHOD_SPECTRAL_MUTING, CONCEAL_METHOD_NB - 1, AD, "conceal" }, |
| 63 | { "spectral", "Spectral muting", 0, AV_OPT_TYPE_CONST, { .i64 = CONCEAL_METHOD_SPECTRAL_MUTING }, INT_MIN, INT_MAX, AD, "conceal" }, |
| 64 | { "noise", "Noise Substitution", 0, AV_OPT_TYPE_CONST, { .i64 = CONCEAL_METHOD_NOISE_SUBSTITUTION }, INT_MIN, INT_MAX, AD, "conceal" }, |
| 65 | { "energy", "Energy Interpolation", 0, AV_OPT_TYPE_CONST, { .i64 = CONCEAL_METHOD_ENERGY_INTERPOLATION }, INT_MIN, INT_MAX, AD, "conceal" }, |
| 66 | { "drc_boost", "Dynamic Range Control: boost, where [0] is none and [127] is max boost", |
| 67 | OFFSET(drc_boost), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 127, AD, NULL }, |
| 68 | { "drc_cut", "Dynamic Range Control: attenuation factor, where [0] is none and [127] is max compression", |
| 69 | OFFSET(drc_cut), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 127, AD, NULL }, |
| 70 | { "drc_level", "Dynamic Range Control: reference level, quantized to 0.25dB steps where [0] is 0dB and [127] is -31.75dB", |
| 71 | OFFSET(drc_level), AV_OPT_TYPE_INT, { .i64 = -1}, -1, 127, AD, NULL }, |
| 72 | { "drc_heavy", "Dynamic Range Control: heavy compression, where [1] is on (RF mode) and [0] is off", |
| 73 | OFFSET(drc_heavy), AV_OPT_TYPE_INT, { .i64 = -1}, -1, 1, AD, NULL }, |
| 74 | { NULL } |
| 75 | }; |
| 76 | |
| 77 | static const AVClass fdk_aac_dec_class = { |
| 78 | "libfdk-aac decoder", av_default_item_name, fdk_aac_dec_options, LIBAVUTIL_VERSION_INT |
| 79 | }; |
| 80 | |
| 81 | static int get_stream_info(AVCodecContext *avctx) |
| 82 | { |
| 83 | FDKAACDecContext *s = avctx->priv_data; |
| 84 | CStreamInfo *info = aacDecoder_GetStreamInfo(s->handle); |
| 85 | int channel_counts[0x24] = { 0 }; |
| 86 | int i, ch_error = 0; |
| 87 | uint64_t ch_layout = 0; |
| 88 | |
| 89 | if (!info) { |
| 90 | av_log(avctx, AV_LOG_ERROR, "Unable to get stream info\n"); |
| 91 | return AVERROR_UNKNOWN; |
| 92 | } |
| 93 | |
| 94 | if (info->sampleRate <= 0) { |
| 95 | av_log(avctx, AV_LOG_ERROR, "Stream info not initialized\n"); |
| 96 | return AVERROR_UNKNOWN; |
| 97 | } |
| 98 | avctx->sample_rate = info->sampleRate; |
| 99 | avctx->frame_size = info->frameSize; |
| 100 | |
| 101 | for (i = 0; i < info->numChannels; i++) { |
| 102 | AUDIO_CHANNEL_TYPE ctype = info->pChannelType[i]; |
| 103 | if (ctype <= ACT_NONE || ctype > FF_ARRAY_ELEMS(channel_counts)) { |
| 104 | av_log(avctx, AV_LOG_WARNING, "unknown channel type\n"); |
| 105 | break; |
| 106 | } |
| 107 | channel_counts[ctype]++; |
| 108 | } |
| 109 | av_log(avctx, AV_LOG_DEBUG, |
| 110 | "%d channels - front:%d side:%d back:%d lfe:%d top:%d\n", |
| 111 | info->numChannels, |
| 112 | channel_counts[ACT_FRONT], channel_counts[ACT_SIDE], |
| 113 | channel_counts[ACT_BACK], channel_counts[ACT_LFE], |
| 114 | channel_counts[ACT_FRONT_TOP] + channel_counts[ACT_SIDE_TOP] + |
| 115 | channel_counts[ACT_BACK_TOP] + channel_counts[ACT_TOP]); |
| 116 | |
| 117 | switch (channel_counts[ACT_FRONT]) { |
| 118 | case 4: |
| 119 | ch_layout |= AV_CH_LAYOUT_STEREO | AV_CH_FRONT_LEFT_OF_CENTER | |
| 120 | AV_CH_FRONT_RIGHT_OF_CENTER; |
| 121 | break; |
| 122 | case 3: |
| 123 | ch_layout |= AV_CH_LAYOUT_STEREO | AV_CH_FRONT_CENTER; |
| 124 | break; |
| 125 | case 2: |
| 126 | ch_layout |= AV_CH_LAYOUT_STEREO; |
| 127 | break; |
| 128 | case 1: |
| 129 | ch_layout |= AV_CH_FRONT_CENTER; |
| 130 | break; |
| 131 | default: |
| 132 | av_log(avctx, AV_LOG_WARNING, |
| 133 | "unsupported number of front channels: %d\n", |
| 134 | channel_counts[ACT_FRONT]); |
| 135 | ch_error = 1; |
| 136 | break; |
| 137 | } |
| 138 | if (channel_counts[ACT_SIDE] > 0) { |
| 139 | if (channel_counts[ACT_SIDE] == 2) { |
| 140 | ch_layout |= AV_CH_SIDE_LEFT | AV_CH_SIDE_RIGHT; |
| 141 | } else { |
| 142 | av_log(avctx, AV_LOG_WARNING, |
| 143 | "unsupported number of side channels: %d\n", |
| 144 | channel_counts[ACT_SIDE]); |
| 145 | ch_error = 1; |
| 146 | } |
| 147 | } |
| 148 | if (channel_counts[ACT_BACK] > 0) { |
| 149 | switch (channel_counts[ACT_BACK]) { |
| 150 | case 3: |
| 151 | ch_layout |= AV_CH_BACK_LEFT | AV_CH_BACK_RIGHT | AV_CH_BACK_CENTER; |
| 152 | break; |
| 153 | case 2: |
| 154 | ch_layout |= AV_CH_BACK_LEFT | AV_CH_BACK_RIGHT; |
| 155 | break; |
| 156 | case 1: |
| 157 | ch_layout |= AV_CH_BACK_CENTER; |
| 158 | break; |
| 159 | default: |
| 160 | av_log(avctx, AV_LOG_WARNING, |
| 161 | "unsupported number of back channels: %d\n", |
| 162 | channel_counts[ACT_BACK]); |
| 163 | ch_error = 1; |
| 164 | break; |
| 165 | } |
| 166 | } |
| 167 | if (channel_counts[ACT_LFE] > 0) { |
| 168 | if (channel_counts[ACT_LFE] == 1) { |
| 169 | ch_layout |= AV_CH_LOW_FREQUENCY; |
| 170 | } else { |
| 171 | av_log(avctx, AV_LOG_WARNING, |
| 172 | "unsupported number of LFE channels: %d\n", |
| 173 | channel_counts[ACT_LFE]); |
| 174 | ch_error = 1; |
| 175 | } |
| 176 | } |
| 177 | if (!ch_error && |
| 178 | av_get_channel_layout_nb_channels(ch_layout) != info->numChannels) { |
| 179 | av_log(avctx, AV_LOG_WARNING, "unsupported channel configuration\n"); |
| 180 | ch_error = 1; |
| 181 | } |
| 182 | if (ch_error) |
| 183 | avctx->channel_layout = 0; |
| 184 | else |
| 185 | avctx->channel_layout = ch_layout; |
| 186 | |
| 187 | avctx->channels = info->numChannels; |
| 188 | |
| 189 | return 0; |
| 190 | } |
| 191 | |
| 192 | static av_cold int fdk_aac_decode_close(AVCodecContext *avctx) |
| 193 | { |
| 194 | FDKAACDecContext *s = avctx->priv_data; |
| 195 | |
| 196 | if (s->handle) |
| 197 | aacDecoder_Close(s->handle); |
| 198 | av_freep(&s->decoder_buffer); |
| 199 | av_freep(&s->anc_buffer); |
| 200 | |
| 201 | return 0; |
| 202 | } |
| 203 | |
| 204 | static av_cold int fdk_aac_decode_init(AVCodecContext *avctx) |
| 205 | { |
| 206 | FDKAACDecContext *s = avctx->priv_data; |
| 207 | AAC_DECODER_ERROR err; |
| 208 | int ret; |
| 209 | |
| 210 | s->handle = aacDecoder_Open(avctx->extradata_size ? TT_MP4_RAW : TT_MP4_ADTS, 1); |
| 211 | if (!s->handle) { |
| 212 | av_log(avctx, AV_LOG_ERROR, "Error opening decoder\n"); |
| 213 | return AVERROR_UNKNOWN; |
| 214 | } |
| 215 | |
| 216 | if (avctx->extradata_size) { |
| 217 | if ((err = aacDecoder_ConfigRaw(s->handle, &avctx->extradata, |
| 218 | &avctx->extradata_size)) != AAC_DEC_OK) { |
| 219 | av_log(avctx, AV_LOG_ERROR, "Unable to set extradata\n"); |
| 220 | return AVERROR_INVALIDDATA; |
| 221 | } |
| 222 | } |
| 223 | |
| 224 | if ((err = aacDecoder_SetParam(s->handle, AAC_CONCEAL_METHOD, |
| 225 | s->conceal_method)) != AAC_DEC_OK) { |
| 226 | av_log(avctx, AV_LOG_ERROR, "Unable to set error concealment method\n"); |
| 227 | return AVERROR_UNKNOWN; |
| 228 | } |
| 229 | |
| 230 | if (avctx->request_channel_layout > 0 && |
| 231 | avctx->request_channel_layout != AV_CH_LAYOUT_NATIVE) { |
| 232 | int downmix_channels = -1; |
| 233 | |
| 234 | switch (avctx->request_channel_layout) { |
| 235 | case AV_CH_LAYOUT_STEREO: |
| 236 | case AV_CH_LAYOUT_STEREO_DOWNMIX: |
| 237 | downmix_channels = 2; |
| 238 | break; |
| 239 | case AV_CH_LAYOUT_MONO: |
| 240 | downmix_channels = 1; |
| 241 | break; |
| 242 | default: |
| 243 | av_log(avctx, AV_LOG_WARNING, "Invalid request_channel_layout\n"); |
| 244 | break; |
| 245 | } |
| 246 | |
| 247 | if (downmix_channels != -1) { |
| 248 | if (aacDecoder_SetParam(s->handle, AAC_PCM_MAX_OUTPUT_CHANNELS, |
| 249 | downmix_channels) != AAC_DEC_OK) { |
| 250 | av_log(avctx, AV_LOG_WARNING, "Unable to set output channels in the decoder\n"); |
| 251 | } else { |
| 252 | s->anc_buffer = av_malloc(DMX_ANC_BUFFSIZE); |
| 253 | if (!s->anc_buffer) { |
| 254 | av_log(avctx, AV_LOG_ERROR, "Unable to allocate ancillary buffer for the decoder\n"); |
| 255 | ret = AVERROR(ENOMEM); |
| 256 | goto fail; |
| 257 | } |
| 258 | if (aacDecoder_AncDataInit(s->handle, s->anc_buffer, DMX_ANC_BUFFSIZE)) { |
| 259 | av_log(avctx, AV_LOG_ERROR, "Unable to register downmix ancillary buffer in the decoder\n"); |
| 260 | ret = AVERROR_UNKNOWN; |
| 261 | goto fail; |
| 262 | } |
| 263 | } |
| 264 | } |
| 265 | } |
| 266 | |
| 267 | if (s->drc_boost != -1) { |
| 268 | if (aacDecoder_SetParam(s->handle, AAC_DRC_BOOST_FACTOR, s->drc_boost) != AAC_DEC_OK) { |
| 269 | av_log(avctx, AV_LOG_ERROR, "Unable to set DRC boost factor in the decoder\n"); |
| 270 | return AVERROR_UNKNOWN; |
| 271 | } |
| 272 | } |
| 273 | |
| 274 | if (s->drc_cut != -1) { |
| 275 | if (aacDecoder_SetParam(s->handle, AAC_DRC_ATTENUATION_FACTOR, s->drc_cut) != AAC_DEC_OK) { |
| 276 | av_log(avctx, AV_LOG_ERROR, "Unable to set DRC attenuation factor in the decoder\n"); |
| 277 | return AVERROR_UNKNOWN; |
| 278 | } |
| 279 | } |
| 280 | |
| 281 | if (s->drc_level != -1) { |
| 282 | if (aacDecoder_SetParam(s->handle, AAC_DRC_REFERENCE_LEVEL, s->drc_level) != AAC_DEC_OK) { |
| 283 | av_log(avctx, AV_LOG_ERROR, "Unable to set DRC reference level in the decoder\n"); |
| 284 | return AVERROR_UNKNOWN; |
| 285 | } |
| 286 | } |
| 287 | |
| 288 | if (s->drc_heavy != -1) { |
| 289 | if (aacDecoder_SetParam(s->handle, AAC_DRC_HEAVY_COMPRESSION, s->drc_heavy) != AAC_DEC_OK) { |
| 290 | av_log(avctx, AV_LOG_ERROR, "Unable to set DRC heavy compression in the decoder\n"); |
| 291 | return AVERROR_UNKNOWN; |
| 292 | } |
| 293 | } |
| 294 | |
| 295 | avctx->sample_fmt = AV_SAMPLE_FMT_S16; |
| 296 | |
| 297 | return 0; |
| 298 | fail: |
| 299 | fdk_aac_decode_close(avctx); |
| 300 | return ret; |
| 301 | } |
| 302 | |
| 303 | static int fdk_aac_decode_frame(AVCodecContext *avctx, void *data, |
| 304 | int *got_frame_ptr, AVPacket *avpkt) |
| 305 | { |
| 306 | FDKAACDecContext *s = avctx->priv_data; |
| 307 | AVFrame *frame = data; |
| 308 | int ret; |
| 309 | AAC_DECODER_ERROR err; |
| 310 | UINT valid = avpkt->size; |
| 311 | uint8_t *buf, *tmpptr = NULL; |
| 312 | int buf_size; |
| 313 | |
| 314 | err = aacDecoder_Fill(s->handle, &avpkt->data, &avpkt->size, &valid); |
| 315 | if (err != AAC_DEC_OK) { |
| 316 | av_log(avctx, AV_LOG_ERROR, "aacDecoder_Fill() failed: %x\n", err); |
| 317 | return AVERROR_INVALIDDATA; |
| 318 | } |
| 319 | |
| 320 | if (s->initialized) { |
| 321 | frame->nb_samples = avctx->frame_size; |
| 322 | if ((ret = ff_get_buffer(avctx, frame, 0)) < 0) |
| 323 | return ret; |
| 324 | |
| 325 | if (s->anc_buffer) { |
| 326 | buf_size = DECODER_BUFFSIZE * DECODER_MAX_CHANNELS; |
| 327 | buf = s->decoder_buffer; |
| 328 | } else { |
| 329 | buf = frame->extended_data[0]; |
| 330 | buf_size = avctx->channels * frame->nb_samples * |
| 331 | av_get_bytes_per_sample(avctx->sample_fmt); |
| 332 | } |
| 333 | } else { |
| 334 | buf_size = DECODER_BUFFSIZE * DECODER_MAX_CHANNELS; |
| 335 | |
| 336 | if (!s->decoder_buffer) |
| 337 | s->decoder_buffer = av_malloc(buf_size); |
| 338 | if (!s->decoder_buffer) |
| 339 | return AVERROR(ENOMEM); |
| 340 | |
| 341 | buf = tmpptr = s->decoder_buffer; |
| 342 | } |
| 343 | |
| 344 | err = aacDecoder_DecodeFrame(s->handle, (INT_PCM *) buf, buf_size, 0); |
| 345 | if (err == AAC_DEC_NOT_ENOUGH_BITS) { |
| 346 | ret = avpkt->size - valid; |
| 347 | goto end; |
| 348 | } |
| 349 | if (err != AAC_DEC_OK) { |
| 350 | av_log(avctx, AV_LOG_ERROR, |
| 351 | "aacDecoder_DecodeFrame() failed: %x\n", err); |
| 352 | ret = AVERROR_UNKNOWN; |
| 353 | goto end; |
| 354 | } |
| 355 | |
| 356 | if (!s->initialized) { |
| 357 | if ((ret = get_stream_info(avctx)) < 0) |
| 358 | goto end; |
| 359 | s->initialized = 1; |
| 360 | frame->nb_samples = avctx->frame_size; |
| 361 | } |
| 362 | |
| 363 | if (tmpptr) { |
| 364 | frame->nb_samples = avctx->frame_size; |
| 365 | if ((ret = ff_get_buffer(avctx, frame, 0)) < 0) |
| 366 | goto end; |
| 367 | } |
| 368 | if (s->decoder_buffer) { |
| 369 | memcpy(frame->extended_data[0], buf, |
| 370 | avctx->channels * avctx->frame_size * |
| 371 | av_get_bytes_per_sample(avctx->sample_fmt)); |
| 372 | |
| 373 | if (!s->anc_buffer) |
| 374 | av_freep(&s->decoder_buffer); |
| 375 | } |
| 376 | |
| 377 | *got_frame_ptr = 1; |
| 378 | ret = avpkt->size - valid; |
| 379 | |
| 380 | end: |
| 381 | return ret; |
| 382 | } |
| 383 | |
| 384 | static av_cold void fdk_aac_decode_flush(AVCodecContext *avctx) |
| 385 | { |
| 386 | FDKAACDecContext *s = avctx->priv_data; |
| 387 | AAC_DECODER_ERROR err; |
| 388 | |
| 389 | if (!s->handle) |
| 390 | return; |
| 391 | |
| 392 | if ((err = aacDecoder_SetParam(s->handle, |
| 393 | AAC_TPDEC_CLEAR_BUFFER, 1)) != AAC_DEC_OK) |
| 394 | av_log(avctx, AV_LOG_WARNING, "failed to clear buffer when flushing\n"); |
| 395 | } |
| 396 | |
| 397 | AVCodec ff_libfdk_aac_decoder = { |
| 398 | .name = "libfdk_aac", |
| 399 | .long_name = NULL_IF_CONFIG_SMALL("Fraunhofer FDK AAC"), |
| 400 | .type = AVMEDIA_TYPE_AUDIO, |
| 401 | .id = AV_CODEC_ID_AAC, |
| 402 | .priv_data_size = sizeof(FDKAACDecContext), |
| 403 | .init = fdk_aac_decode_init, |
| 404 | .decode = fdk_aac_decode_frame, |
| 405 | .close = fdk_aac_decode_close, |
| 406 | .flush = fdk_aac_decode_flush, |
| 407 | .capabilities = CODEC_CAP_DR1 | CODEC_CAP_CHANNEL_CONF, |
| 408 | .priv_class = &fdk_aac_dec_class, |
| 409 | }; |