3 * Copyright (c) 2012 Martin Storsjo
5 * This file is part of FFmpeg.
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.
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.
20 #include <fdk-aac/aacenc_lib.h>
22 #include "libavutil/channel_layout.h"
23 #include "libavutil/common.h"
24 #include "libavutil/opt.h"
26 #include "audio_frame_queue.h"
30 #ifdef AACENCODER_LIB_VL0
31 #define FDKENC_VER_AT_LEAST(vl0, vl1) \
32 ((AACENCODER_LIB_VL0 > vl0) || \
33 (AACENCODER_LIB_VL0 == vl0 && AACENCODER_LIB_VL1 >= vl1))
35 #define FDKENC_VER_AT_LEAST(vl0, vl1) 0
38 typedef struct AACContext {
40 HANDLE_AACENCODER handle;
52 static const AVOption aac_enc_options[] = {
53 { "afterburner", "Afterburner (improved quality)", offsetof(AACContext, afterburner), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM },
54 { "eld_sbr", "Enable SBR for ELD (for SBR in other configurations, use the -profile parameter)", offsetof(AACContext, eld_sbr), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM },
55 #if FDKENC_VER_AT_LEAST(4, 0) // 4.0.0
56 { "eld_v2", "Enable ELDv2 (LD-MPS extension for ELD stereo signals)", offsetof(AACContext, eld_v2), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM },
58 { "signaling", "SBR/PS signaling style", offsetof(AACContext, signaling), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 2, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM, "signaling" },
59 { "default", "Choose signaling implicitly (explicit hierarchical by default, implicit if global header is disabled)", 0, AV_OPT_TYPE_CONST, { .i64 = -1 }, 0, 0, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM, "signaling" },
60 { "implicit", "Implicit backwards compatible signaling", 0, AV_OPT_TYPE_CONST, { .i64 = 0 }, 0, 0, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM, "signaling" },
61 { "explicit_sbr", "Explicit SBR, implicit PS signaling", 0, AV_OPT_TYPE_CONST, { .i64 = 1 }, 0, 0, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM, "signaling" },
62 { "explicit_hierarchical", "Explicit hierarchical signaling", 0, AV_OPT_TYPE_CONST, { .i64 = 2 }, 0, 0, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM, "signaling" },
63 { "latm", "Output LATM/LOAS encapsulated data", offsetof(AACContext, latm), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM },
64 { "header_period", "StreamMuxConfig and PCE repetition period (in frames)", offsetof(AACContext, header_period), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 0xffff, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM },
65 { "vbr", "VBR mode (1-5)", offsetof(AACContext, vbr), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 5, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM },
70 static const AVClass aac_enc_class = {
71 .class_name = "libfdk_aac",
72 .item_name = av_default_item_name,
73 .option = aac_enc_options,
74 .version = LIBAVUTIL_VERSION_INT,
77 static const char *aac_get_error(AACENC_ERROR err)
82 case AACENC_INVALID_HANDLE:
83 return "Invalid handle";
84 case AACENC_MEMORY_ERROR:
85 return "Memory allocation error";
86 case AACENC_UNSUPPORTED_PARAMETER:
87 return "Unsupported parameter";
88 case AACENC_INVALID_CONFIG:
89 return "Invalid config";
90 case AACENC_INIT_ERROR:
91 return "Initialization error";
92 case AACENC_INIT_AAC_ERROR:
93 return "AAC library initialization error";
94 case AACENC_INIT_SBR_ERROR:
95 return "SBR library initialization error";
96 case AACENC_INIT_TP_ERROR:
97 return "Transport library initialization error";
98 case AACENC_INIT_META_ERROR:
99 return "Metadata library initialization error";
100 case AACENC_ENCODE_ERROR:
101 return "Encoding error";
102 case AACENC_ENCODE_EOF:
103 return "End of file";
105 return "Unknown error";
109 static int aac_encode_close(AVCodecContext *avctx)
111 AACContext *s = avctx->priv_data;
114 aacEncClose(&s->handle);
115 av_freep(&avctx->extradata);
116 ff_af_queue_close(&s->afq);
121 static av_cold int aac_encode_init(AVCodecContext *avctx)
123 AACContext *s = avctx->priv_data;
124 int ret = AVERROR(EINVAL);
125 AACENC_InfoStruct info = { 0 };
128 int aot = FF_PROFILE_AAC_LOW + 1;
129 int sce = 0, cpe = 0;
131 if ((err = aacEncOpen(&s->handle, 0, avctx->channels)) != AACENC_OK) {
132 av_log(avctx, AV_LOG_ERROR, "Unable to open the encoder: %s\n",
137 if (avctx->profile != FF_PROFILE_UNKNOWN)
138 aot = avctx->profile + 1;
140 if ((err = aacEncoder_SetParam(s->handle, AACENC_AOT, aot)) != AACENC_OK) {
141 av_log(avctx, AV_LOG_ERROR, "Unable to set the AOT %d: %s\n",
142 aot, aac_get_error(err));
146 if (aot == FF_PROFILE_AAC_ELD + 1 && s->eld_sbr) {
147 if ((err = aacEncoder_SetParam(s->handle, AACENC_SBR_MODE,
149 av_log(avctx, AV_LOG_ERROR, "Unable to enable SBR for ELD: %s\n",
155 if ((err = aacEncoder_SetParam(s->handle, AACENC_SAMPLERATE,
156 avctx->sample_rate)) != AACENC_OK) {
157 av_log(avctx, AV_LOG_ERROR, "Unable to set the sample rate %d: %s\n",
158 avctx->sample_rate, aac_get_error(err));
162 switch (avctx->channels) {
163 case 1: mode = MODE_1; sce = 1; cpe = 0; break;
165 #if FDKENC_VER_AT_LEAST(4, 0) // 4.0.0
166 // (profile + 1) to map from profile range to AOT range
167 if (aot == FF_PROFILE_AAC_ELD + 1 && s->eld_v2) {
168 if ((err = aacEncoder_SetParam(s->handle, AACENC_CHANNELMODE,
169 128)) != AACENC_OK) {
170 av_log(avctx, AV_LOG_ERROR, "Unable to enable ELDv2: %s\n",
186 case 3: mode = MODE_1_2; sce = 1; cpe = 1; break;
187 case 4: mode = MODE_1_2_1; sce = 2; cpe = 1; break;
188 case 5: mode = MODE_1_2_2; sce = 1; cpe = 2; break;
189 case 6: mode = MODE_1_2_2_1; sce = 2; cpe = 2; break;
190 /* The version macro is introduced the same time as the 7.1 support, so this
192 #if FDKENC_VER_AT_LEAST(3, 4) // 3.4.12
196 if (avctx->channel_layout == AV_CH_LAYOUT_7POINT1) {
197 mode = MODE_7_1_REAR_SURROUND;
199 // MODE_1_2_2_2_1 and MODE_7_1_FRONT_CENTER use the same channel layout
200 mode = MODE_7_1_FRONT_CENTER;
205 av_log(avctx, AV_LOG_ERROR,
206 "Unsupported number of channels %d\n", avctx->channels);
210 if ((err = aacEncoder_SetParam(s->handle, AACENC_CHANNELMODE,
211 mode)) != AACENC_OK) {
212 av_log(avctx, AV_LOG_ERROR,
213 "Unable to set channel mode %d: %s\n", mode, aac_get_error(err));
217 if ((err = aacEncoder_SetParam(s->handle, AACENC_CHANNELORDER,
219 av_log(avctx, AV_LOG_ERROR,
220 "Unable to set wav channel order %d: %s\n",
221 mode, aac_get_error(err));
225 if (avctx->flags & AV_CODEC_FLAG_QSCALE || s->vbr) {
226 int mode = s->vbr ? s->vbr : avctx->global_quality;
227 if (mode < 1 || mode > 5) {
228 av_log(avctx, AV_LOG_WARNING,
229 "VBR quality %d out of range, should be 1-5\n", mode);
230 mode = av_clip(mode, 1, 5);
232 av_log(avctx, AV_LOG_WARNING,
233 "Note, the VBR setting is unsupported and only works with "
234 "some parameter combinations\n");
235 if ((err = aacEncoder_SetParam(s->handle, AACENC_BITRATEMODE,
236 mode)) != AACENC_OK) {
237 av_log(avctx, AV_LOG_ERROR, "Unable to set the VBR bitrate mode %d: %s\n",
238 mode, aac_get_error(err));
242 if (avctx->bit_rate <= 0) {
243 if (avctx->profile == FF_PROFILE_AAC_HE_V2) {
247 avctx->bit_rate = (96*sce + 128*cpe) * avctx->sample_rate / 44;
248 if (avctx->profile == FF_PROFILE_AAC_HE ||
249 avctx->profile == FF_PROFILE_AAC_HE_V2 ||
250 avctx->profile == FF_PROFILE_MPEG2_AAC_HE ||
252 avctx->bit_rate /= 2;
254 if ((err = aacEncoder_SetParam(s->handle, AACENC_BITRATE,
255 avctx->bit_rate)) != AACENC_OK) {
256 av_log(avctx, AV_LOG_ERROR, "Unable to set the bitrate %"PRId64": %s\n",
257 avctx->bit_rate, aac_get_error(err));
262 /* Choose bitstream format - if global header is requested, use
263 * raw access units, otherwise use ADTS. */
264 if ((err = aacEncoder_SetParam(s->handle, AACENC_TRANSMUX,
265 avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER ? TT_MP4_RAW :
266 s->latm ? TT_MP4_LOAS : TT_MP4_ADTS)) != AACENC_OK) {
267 av_log(avctx, AV_LOG_ERROR, "Unable to set the transmux format: %s\n",
272 if (s->latm && s->header_period) {
273 if ((err = aacEncoder_SetParam(s->handle, AACENC_HEADER_PERIOD,
274 s->header_period)) != AACENC_OK) {
275 av_log(avctx, AV_LOG_ERROR, "Unable to set header period: %s\n",
281 /* If no signaling mode is chosen, use explicit hierarchical signaling
282 * if using mp4 mode (raw access units, with global header) and
283 * implicit signaling if using ADTS. */
284 if (s->signaling < 0)
285 s->signaling = avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER ? 2 : 0;
287 if ((err = aacEncoder_SetParam(s->handle, AACENC_SIGNALING_MODE,
288 s->signaling)) != AACENC_OK) {
289 av_log(avctx, AV_LOG_ERROR, "Unable to set signaling mode %d: %s\n",
290 s->signaling, aac_get_error(err));
294 if ((err = aacEncoder_SetParam(s->handle, AACENC_AFTERBURNER,
295 s->afterburner)) != AACENC_OK) {
296 av_log(avctx, AV_LOG_ERROR, "Unable to set afterburner to %d: %s\n",
297 s->afterburner, aac_get_error(err));
301 if (avctx->cutoff > 0) {
302 if (avctx->cutoff < (avctx->sample_rate + 255) >> 8 || avctx->cutoff > 20000) {
303 av_log(avctx, AV_LOG_ERROR, "cutoff valid range is %d-20000\n",
304 (avctx->sample_rate + 255) >> 8);
307 if ((err = aacEncoder_SetParam(s->handle, AACENC_BANDWIDTH,
308 avctx->cutoff)) != AACENC_OK) {
309 av_log(avctx, AV_LOG_ERROR, "Unable to set the encoder bandwidth to %d: %s\n",
310 avctx->cutoff, aac_get_error(err));
315 if ((err = aacEncEncode(s->handle, NULL, NULL, NULL, NULL)) != AACENC_OK) {
316 av_log(avctx, AV_LOG_ERROR, "Unable to initialize the encoder: %s\n",
318 return AVERROR(EINVAL);
321 if ((err = aacEncInfo(s->handle, &info)) != AACENC_OK) {
322 av_log(avctx, AV_LOG_ERROR, "Unable to get encoder info: %s\n",
327 avctx->frame_size = info.frameLength;
328 #if FDKENC_VER_AT_LEAST(4, 0) // 4.0.0
329 avctx->initial_padding = info.nDelay;
331 avctx->initial_padding = info.encoderDelay;
333 ff_af_queue_init(avctx, &s->afq);
335 if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER) {
336 avctx->extradata_size = info.confSize;
337 avctx->extradata = av_mallocz(avctx->extradata_size +
338 AV_INPUT_BUFFER_PADDING_SIZE);
339 if (!avctx->extradata) {
340 ret = AVERROR(ENOMEM);
344 memcpy(avctx->extradata, info.confBuf, info.confSize);
348 aac_encode_close(avctx);
352 static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
353 const AVFrame *frame, int *got_packet_ptr)
355 AACContext *s = avctx->priv_data;
356 AACENC_BufDesc in_buf = { 0 }, out_buf = { 0 };
357 AACENC_InArgs in_args = { 0 };
358 AACENC_OutArgs out_args = { 0 };
359 int in_buffer_identifier = IN_AUDIO_DATA;
360 int in_buffer_size, in_buffer_element_size;
361 int out_buffer_identifier = OUT_BITSTREAM_DATA;
362 int out_buffer_size, out_buffer_element_size;
363 void *in_ptr, *out_ptr;
365 uint8_t dummy_buf[1];
368 /* handle end-of-stream small frame and flushing */
370 /* Must be a non-null pointer, even if it's a dummy. We could use
371 * the address of anything else on the stack as well. */
375 in_args.numInSamples = -1;
377 in_ptr = frame->data[0];
378 in_buffer_size = 2 * avctx->channels * frame->nb_samples;
380 in_args.numInSamples = avctx->channels * frame->nb_samples;
382 /* add current frame to the queue */
383 if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
387 in_buffer_element_size = 2;
389 in_buf.bufs = &in_ptr;
390 in_buf.bufferIdentifiers = &in_buffer_identifier;
391 in_buf.bufSizes = &in_buffer_size;
392 in_buf.bufElSizes = &in_buffer_element_size;
394 /* The maximum packet size is 6144 bits aka 768 bytes per channel. */
395 if ((ret = ff_alloc_packet2(avctx, avpkt, FFMAX(8192, 768 * avctx->channels), 0)) < 0)
398 out_ptr = avpkt->data;
399 out_buffer_size = avpkt->size;
400 out_buffer_element_size = 1;
402 out_buf.bufs = &out_ptr;
403 out_buf.bufferIdentifiers = &out_buffer_identifier;
404 out_buf.bufSizes = &out_buffer_size;
405 out_buf.bufElSizes = &out_buffer_element_size;
407 if ((err = aacEncEncode(s->handle, &in_buf, &out_buf, &in_args,
408 &out_args)) != AACENC_OK) {
409 if (!frame && err == AACENC_ENCODE_EOF)
411 av_log(avctx, AV_LOG_ERROR, "Unable to encode frame: %s\n",
413 return AVERROR(EINVAL);
416 if (!out_args.numOutBytes)
419 /* Get the next frame pts & duration */
420 ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts,
423 avpkt->size = out_args.numOutBytes;
428 static const AVProfile profiles[] = {
429 { FF_PROFILE_AAC_LOW, "LC" },
430 { FF_PROFILE_AAC_HE, "HE-AAC" },
431 { FF_PROFILE_AAC_HE_V2, "HE-AACv2" },
432 { FF_PROFILE_AAC_LD, "LD" },
433 { FF_PROFILE_AAC_ELD, "ELD" },
434 { FF_PROFILE_UNKNOWN },
437 static const AVCodecDefault aac_encode_defaults[] = {
442 static const uint64_t aac_channel_layout[] = {
445 AV_CH_LAYOUT_SURROUND,
446 AV_CH_LAYOUT_4POINT0,
447 AV_CH_LAYOUT_5POINT0_BACK,
448 AV_CH_LAYOUT_5POINT1_BACK,
449 #if FDKENC_VER_AT_LEAST(3, 4) // 3.4.12
450 AV_CH_LAYOUT_7POINT1_WIDE_BACK,
451 AV_CH_LAYOUT_7POINT1,
456 static const int aac_sample_rates[] = {
457 96000, 88200, 64000, 48000, 44100, 32000,
458 24000, 22050, 16000, 12000, 11025, 8000, 0
461 AVCodec ff_libfdk_aac_encoder = {
462 .name = "libfdk_aac",
463 .long_name = NULL_IF_CONFIG_SMALL("Fraunhofer FDK AAC"),
464 .type = AVMEDIA_TYPE_AUDIO,
465 .id = AV_CODEC_ID_AAC,
466 .priv_data_size = sizeof(AACContext),
467 .init = aac_encode_init,
468 .encode2 = aac_encode_frame,
469 .close = aac_encode_close,
470 .capabilities = AV_CODEC_CAP_SMALL_LAST_FRAME | AV_CODEC_CAP_DELAY,
471 .sample_fmts = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_S16,
472 AV_SAMPLE_FMT_NONE },
473 .priv_class = &aac_enc_class,
474 .defaults = aac_encode_defaults,
475 .profiles = profiles,
476 .supported_samplerates = aac_sample_rates,
477 .channel_layouts = aac_channel_layout,
478 .wrapper_name = "libfdk",