mov: Export geotag metadata fields
[libav.git] / libavcodec / libfdk-aacenc.c
CommitLineData
37eeb5e2
MS
1/*
2 * AAC encoder wrapper
3 * Copyright (c) 2012 Martin Storsjo
4 *
5 * This file is part of Libav.
6 *
3828eb85
MS
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions are
9 * met:
37eeb5e2 10 *
3828eb85
MS
11 * * Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
37eeb5e2 13 *
3828eb85
MS
14 * * Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in
16 * the documentation and/or other materials provided with the
17 * distribution.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
20 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
21 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
22 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
23 * HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
24 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
25 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
26 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
27 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
28 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
29 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
37eeb5e2
MS
30 */
31
32#include <fdk-aac/aacenc_lib.h>
33
a903f8f0
JR
34#include "libavutil/channel_layout.h"
35#include "libavutil/common.h"
36#include "libavutil/opt.h"
37eeb5e2
MS
37#include "avcodec.h"
38#include "audio_frame_queue.h"
39#include "internal.h"
37eeb5e2
MS
40
41typedef struct AACContext {
42 const AVClass *class;
43 HANDLE_AACENCODER handle;
44 int afterburner;
45 int eld_sbr;
46 int signaling;
160a27c5
KK
47 int latm;
48 int header_period;
375bff70 49 int vbr;
37eeb5e2
MS
50
51 AudioFrameQueue afq;
52} AACContext;
53
54static const AVOption aac_enc_options[] = {
e6153f17
MS
55 { "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 },
56 { "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 },
57 { "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" },
124134e4
MS
58 { "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" },
59 { "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" },
60 { "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" },
61 { "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" },
e6153f17
MS
62 { "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 },
63 { "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 },
375bff70 64 { "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 },
37eeb5e2
MS
65 { NULL }
66};
67
68static const AVClass aac_enc_class = {
69 "libfdk_aac", av_default_item_name, aac_enc_options, LIBAVUTIL_VERSION_INT
70};
71
72static const char *aac_get_error(AACENC_ERROR err)
73{
74 switch (err) {
75 case AACENC_OK:
76 return "No error";
77 case AACENC_INVALID_HANDLE:
78 return "Invalid handle";
79 case AACENC_MEMORY_ERROR:
80 return "Memory allocation error";
81 case AACENC_UNSUPPORTED_PARAMETER:
82 return "Unsupported parameter";
83 case AACENC_INVALID_CONFIG:
84 return "Invalid config";
85 case AACENC_INIT_ERROR:
86 return "Initialization error";
87 case AACENC_INIT_AAC_ERROR:
88 return "AAC library initialization error";
89 case AACENC_INIT_SBR_ERROR:
90 return "SBR library initialization error";
91 case AACENC_INIT_TP_ERROR:
92 return "Transport library initialization error";
93 case AACENC_INIT_META_ERROR:
94 return "Metadata library initialization error";
95 case AACENC_ENCODE_ERROR:
96 return "Encoding error";
97 case AACENC_ENCODE_EOF:
98 return "End of file";
99 default:
100 return "Unknown error";
101 }
102}
103
104static int aac_encode_close(AVCodecContext *avctx)
105{
106 AACContext *s = avctx->priv_data;
107
108 if (s->handle)
109 aacEncClose(&s->handle);
37eeb5e2
MS
110 av_freep(&avctx->extradata);
111 ff_af_queue_close(&s->afq);
112
113 return 0;
114}
115
116static av_cold int aac_encode_init(AVCodecContext *avctx)
117{
118 AACContext *s = avctx->priv_data;
119 int ret = AVERROR(EINVAL);
120 AACENC_InfoStruct info = { 0 };
121 CHANNEL_MODE mode;
122 AACENC_ERROR err;
123 int aot = FF_PROFILE_AAC_LOW + 1;
124 int sce = 0, cpe = 0;
125
126 if ((err = aacEncOpen(&s->handle, 0, avctx->channels)) != AACENC_OK) {
127 av_log(avctx, AV_LOG_ERROR, "Unable to open the encoder: %s\n",
128 aac_get_error(err));
129 goto error;
130 }
131
132 if (avctx->profile != FF_PROFILE_UNKNOWN)
133 aot = avctx->profile + 1;
134
135 if ((err = aacEncoder_SetParam(s->handle, AACENC_AOT, aot)) != AACENC_OK) {
136 av_log(avctx, AV_LOG_ERROR, "Unable to set the AOT %d: %s\n",
137 aot, aac_get_error(err));
138 goto error;
139 }
140
141 if (aot == FF_PROFILE_AAC_ELD + 1 && s->eld_sbr) {
142 if ((err = aacEncoder_SetParam(s->handle, AACENC_SBR_MODE,
143 1)) != AACENC_OK) {
144 av_log(avctx, AV_LOG_ERROR, "Unable to enable SBR for ELD: %s\n",
145 aac_get_error(err));
146 goto error;
147 }
148 }
149
150 if ((err = aacEncoder_SetParam(s->handle, AACENC_SAMPLERATE,
151 avctx->sample_rate)) != AACENC_OK) {
152 av_log(avctx, AV_LOG_ERROR, "Unable to set the sample rate %d: %s\n",
153 avctx->sample_rate, aac_get_error(err));
154 goto error;
155 }
156
157 switch (avctx->channels) {
158 case 1: mode = MODE_1; sce = 1; cpe = 0; break;
159 case 2: mode = MODE_2; sce = 0; cpe = 1; break;
160 case 3: mode = MODE_1_2; sce = 1; cpe = 1; break;
161 case 4: mode = MODE_1_2_1; sce = 2; cpe = 1; break;
162 case 5: mode = MODE_1_2_2; sce = 1; cpe = 2; break;
163 case 6: mode = MODE_1_2_2_1; sce = 2; cpe = 2; break;
15fa856b
JF
164/* The version macro is introduced the same time as the 7.1 support, so this
165 should suffice. */
166#ifdef AACENCODER_LIB_VL0
167 case 8:
168 sce = 2;
169 cpe = 3;
170 if (avctx->channel_layout == AV_CH_LAYOUT_7POINT1) {
171 mode = MODE_7_1_REAR_SURROUND;
172 } else {
173 // MODE_1_2_2_2_1 and MODE_7_1_FRONT_CENTER use the same channel layout
174 mode = MODE_7_1_FRONT_CENTER;
175 }
176 break;
177#endif
37eeb5e2
MS
178 default:
179 av_log(avctx, AV_LOG_ERROR,
180 "Unsupported number of channels %d\n", avctx->channels);
181 goto error;
182 }
183
184 if ((err = aacEncoder_SetParam(s->handle, AACENC_CHANNELMODE,
185 mode)) != AACENC_OK) {
186 av_log(avctx, AV_LOG_ERROR,
187 "Unable to set channel mode %d: %s\n", mode, aac_get_error(err));
188 goto error;
189 }
190
191 if ((err = aacEncoder_SetParam(s->handle, AACENC_CHANNELORDER,
192 1)) != AACENC_OK) {
193 av_log(avctx, AV_LOG_ERROR,
194 "Unable to set wav channel order %d: %s\n",
195 mode, aac_get_error(err));
196 goto error;
197 }
198
375bff70
MS
199 if (avctx->flags & CODEC_FLAG_QSCALE || s->vbr) {
200 int mode = s->vbr ? s->vbr : avctx->global_quality;
37eeb5e2
MS
201 if (mode < 1 || mode > 5) {
202 av_log(avctx, AV_LOG_WARNING,
203 "VBR quality %d out of range, should be 1-5\n", mode);
204 mode = av_clip(mode, 1, 5);
205 }
a10190dc
MS
206 av_log(avctx, AV_LOG_WARNING,
207 "Note, the VBR setting is unsupported and only works with "
208 "some parameter combinations\n");
37eeb5e2
MS
209 if ((err = aacEncoder_SetParam(s->handle, AACENC_BITRATEMODE,
210 mode)) != AACENC_OK) {
211 av_log(avctx, AV_LOG_ERROR, "Unable to set the VBR bitrate mode %d: %s\n",
212 mode, aac_get_error(err));
213 goto error;
214 }
215 } else {
216 if (avctx->bit_rate <= 0) {
217 if (avctx->profile == FF_PROFILE_AAC_HE_V2) {
218 sce = 1;
219 cpe = 0;
220 }
221 avctx->bit_rate = (96*sce + 128*cpe) * avctx->sample_rate / 44;
222 if (avctx->profile == FF_PROFILE_AAC_HE ||
223 avctx->profile == FF_PROFILE_AAC_HE_V2 ||
95d52464 224 avctx->profile == FF_PROFILE_MPEG2_AAC_HE ||
37eeb5e2
MS
225 s->eld_sbr)
226 avctx->bit_rate /= 2;
227 }
228 if ((err = aacEncoder_SetParam(s->handle, AACENC_BITRATE,
229 avctx->bit_rate)) != AACENC_OK) {
230 av_log(avctx, AV_LOG_ERROR, "Unable to set the bitrate %d: %s\n",
231 avctx->bit_rate, aac_get_error(err));
232 goto error;
233 }
234 }
235
236 /* Choose bitstream format - if global header is requested, use
237 * raw access units, otherwise use ADTS. */
238 if ((err = aacEncoder_SetParam(s->handle, AACENC_TRANSMUX,
160a27c5 239 avctx->flags & CODEC_FLAG_GLOBAL_HEADER ? 0 : s->latm ? 10 : 2)) != AACENC_OK) {
37eeb5e2
MS
240 av_log(avctx, AV_LOG_ERROR, "Unable to set the transmux format: %s\n",
241 aac_get_error(err));
242 goto error;
243 }
244
160a27c5
KK
245 if (s->latm && s->header_period) {
246 if ((err = aacEncoder_SetParam(s->handle, AACENC_HEADER_PERIOD,
247 s->header_period)) != AACENC_OK) {
248 av_log(avctx, AV_LOG_ERROR, "Unable to set header period: %s\n",
249 aac_get_error(err));
250 goto error;
251 }
252 }
253
37eeb5e2
MS
254 /* If no signaling mode is chosen, use explicit hierarchical signaling
255 * if using mp4 mode (raw access units, with global header) and
256 * implicit signaling if using ADTS. */
257 if (s->signaling < 0)
258 s->signaling = avctx->flags & CODEC_FLAG_GLOBAL_HEADER ? 2 : 0;
259
260 if ((err = aacEncoder_SetParam(s->handle, AACENC_SIGNALING_MODE,
261 s->signaling)) != AACENC_OK) {
262 av_log(avctx, AV_LOG_ERROR, "Unable to set signaling mode %d: %s\n",
263 s->signaling, aac_get_error(err));
264 goto error;
265 }
266
267 if ((err = aacEncoder_SetParam(s->handle, AACENC_AFTERBURNER,
268 s->afterburner)) != AACENC_OK) {
269 av_log(avctx, AV_LOG_ERROR, "Unable to set afterburner to %d: %s\n",
270 s->afterburner, aac_get_error(err));
271 goto error;
272 }
273
7548831c 274 if (avctx->cutoff > 0) {
375ef652 275 if (avctx->cutoff < (avctx->sample_rate + 255) >> 8 || avctx->cutoff > 20000) {
c37c383e
MA
276 av_log(avctx, AV_LOG_ERROR, "cutoff valid range is %d-20000\n",
277 (avctx->sample_rate + 255) >> 8);
278 goto error;
279 }
7548831c
MS
280 if ((err = aacEncoder_SetParam(s->handle, AACENC_BANDWIDTH,
281 avctx->cutoff)) != AACENC_OK) {
511cf612 282 av_log(avctx, AV_LOG_ERROR, "Unable to set the encoder bandwidth to %d: %s\n",
7548831c
MS
283 avctx->cutoff, aac_get_error(err));
284 goto error;
285 }
286 }
287
37eeb5e2
MS
288 if ((err = aacEncEncode(s->handle, NULL, NULL, NULL, NULL)) != AACENC_OK) {
289 av_log(avctx, AV_LOG_ERROR, "Unable to initialize the encoder: %s\n",
290 aac_get_error(err));
291 return AVERROR(EINVAL);
292 }
293
294 if ((err = aacEncInfo(s->handle, &info)) != AACENC_OK) {
295 av_log(avctx, AV_LOG_ERROR, "Unable to get encoder info: %s\n",
296 aac_get_error(err));
297 goto error;
298 }
299
37eeb5e2
MS
300 avctx->frame_size = info.frameLength;
301 avctx->delay = info.encoderDelay;
302 ff_af_queue_init(avctx, &s->afq);
303
304 if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER) {
305 avctx->extradata_size = info.confSize;
306 avctx->extradata = av_mallocz(avctx->extradata_size +
307 FF_INPUT_BUFFER_PADDING_SIZE);
308 if (!avctx->extradata) {
309 ret = AVERROR(ENOMEM);
310 goto error;
311 }
312
313 memcpy(avctx->extradata, info.confBuf, info.confSize);
314 }
315 return 0;
316error:
317 aac_encode_close(avctx);
318 return ret;
319}
320
321static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
322 const AVFrame *frame, int *got_packet_ptr)
323{
324 AACContext *s = avctx->priv_data;
325 AACENC_BufDesc in_buf = { 0 }, out_buf = { 0 };
326 AACENC_InArgs in_args = { 0 };
327 AACENC_OutArgs out_args = { 0 };
328 int in_buffer_identifier = IN_AUDIO_DATA;
329 int in_buffer_size, in_buffer_element_size;
330 int out_buffer_identifier = OUT_BITSTREAM_DATA;
331 int out_buffer_size, out_buffer_element_size;
332 void *in_ptr, *out_ptr;
333 int ret;
334 AACENC_ERROR err;
335
336 /* handle end-of-stream small frame and flushing */
337 if (!frame) {
338 in_args.numInSamples = -1;
339 } else {
340 in_ptr = frame->data[0];
341 in_buffer_size = 2 * avctx->channels * frame->nb_samples;
342 in_buffer_element_size = 2;
343
344 in_args.numInSamples = avctx->channels * frame->nb_samples;
345 in_buf.numBufs = 1;
346 in_buf.bufs = &in_ptr;
347 in_buf.bufferIdentifiers = &in_buffer_identifier;
348 in_buf.bufSizes = &in_buffer_size;
349 in_buf.bufElSizes = &in_buffer_element_size;
350
351 /* add current frame to the queue */
1d7ffd06 352 if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
37eeb5e2
MS
353 return ret;
354 }
355
356 /* The maximum packet size is 6144 bits aka 768 bytes per channel. */
357 if ((ret = ff_alloc_packet(avpkt, FFMAX(8192, 768 * avctx->channels)))) {
358 av_log(avctx, AV_LOG_ERROR, "Error getting output packet\n");
359 return ret;
360 }
361
362 out_ptr = avpkt->data;
363 out_buffer_size = avpkt->size;
364 out_buffer_element_size = 1;
365 out_buf.numBufs = 1;
366 out_buf.bufs = &out_ptr;
367 out_buf.bufferIdentifiers = &out_buffer_identifier;
368 out_buf.bufSizes = &out_buffer_size;
369 out_buf.bufElSizes = &out_buffer_element_size;
370
371 if ((err = aacEncEncode(s->handle, &in_buf, &out_buf, &in_args,
372 &out_args)) != AACENC_OK) {
373 if (!frame && err == AACENC_ENCODE_EOF)
374 return 0;
375 av_log(avctx, AV_LOG_ERROR, "Unable to encode frame: %s\n",
376 aac_get_error(err));
377 return AVERROR(EINVAL);
378 }
379
380 if (!out_args.numOutBytes)
381 return 0;
382
383 /* Get the next frame pts & duration */
384 ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts,
385 &avpkt->duration);
386
387 avpkt->size = out_args.numOutBytes;
388 *got_packet_ptr = 1;
389 return 0;
390}
391
392static const AVProfile profiles[] = {
393 { FF_PROFILE_AAC_LOW, "LC" },
394 { FF_PROFILE_AAC_HE, "HE-AAC" },
395 { FF_PROFILE_AAC_HE_V2, "HE-AACv2" },
396 { FF_PROFILE_AAC_LD, "LD" },
397 { FF_PROFILE_AAC_ELD, "ELD" },
398 { FF_PROFILE_UNKNOWN },
399};
400
401static const AVCodecDefault aac_encode_defaults[] = {
402 { "b", "0" },
403 { NULL }
404};
405
406static const uint64_t aac_channel_layout[] = {
407 AV_CH_LAYOUT_MONO,
408 AV_CH_LAYOUT_STEREO,
409 AV_CH_LAYOUT_SURROUND,
410 AV_CH_LAYOUT_4POINT0,
411 AV_CH_LAYOUT_5POINT0_BACK,
412 AV_CH_LAYOUT_5POINT1_BACK,
15fa856b
JF
413#ifdef AACENCODER_LIB_VL0
414 AV_CH_LAYOUT_7POINT1_WIDE_BACK,
415 AV_CH_LAYOUT_7POINT1,
416#endif
37eeb5e2
MS
417 0,
418};
419
94a69dee
NC
420static const int aac_sample_rates[] = {
421 96000, 88200, 64000, 48000, 44100, 32000,
422 24000, 22050, 16000, 12000, 11025, 8000, 0
423};
424
37eeb5e2 425AVCodec ff_libfdk_aac_encoder = {
c8ba8be2 426 .name = "libfdk_aac",
b2bed932 427 .long_name = NULL_IF_CONFIG_SMALL("Fraunhofer FDK AAC"),
c8ba8be2
NC
428 .type = AVMEDIA_TYPE_AUDIO,
429 .id = AV_CODEC_ID_AAC,
430 .priv_data_size = sizeof(AACContext),
431 .init = aac_encode_init,
432 .encode2 = aac_encode_frame,
433 .close = aac_encode_close,
434 .capabilities = CODEC_CAP_SMALL_LAST_FRAME | CODEC_CAP_DELAY,
435 .sample_fmts = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_S16,
436 AV_SAMPLE_FMT_NONE },
c8ba8be2
NC
437 .priv_class = &aac_enc_class,
438 .defaults = aac_encode_defaults,
439 .profiles = profiles,
94a69dee 440 .supported_samplerates = aac_sample_rates,
c8ba8be2 441 .channel_layouts = aac_channel_layout,
37eeb5e2 442};