avconv: rename OutputStream.opts to OutputStream.encoder_opts
[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 *
7 * Libav is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
11 *
12 * Libav is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
16 *
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with Libav; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22#include <fdk-aac/aacenc_lib.h>
23
a903f8f0
JR
24#include "libavutil/channel_layout.h"
25#include "libavutil/common.h"
26#include "libavutil/opt.h"
37eeb5e2
MS
27#include "avcodec.h"
28#include "audio_frame_queue.h"
29#include "internal.h"
37eeb5e2
MS
30
31typedef struct AACContext {
32 const AVClass *class;
33 HANDLE_AACENCODER handle;
34 int afterburner;
35 int eld_sbr;
36 int signaling;
160a27c5
KK
37 int latm;
38 int header_period;
375bff70 39 int vbr;
37eeb5e2
MS
40
41 AudioFrameQueue afq;
42} AACContext;
43
44static const AVOption aac_enc_options[] = {
e6153f17
MS
45 { "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 },
46 { "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 },
47 { "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
48 { "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" },
49 { "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" },
50 { "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" },
51 { "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
52 { "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 },
53 { "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 54 { "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
55 { NULL }
56};
57
58static const AVClass aac_enc_class = {
59 "libfdk_aac", av_default_item_name, aac_enc_options, LIBAVUTIL_VERSION_INT
60};
61
62static const char *aac_get_error(AACENC_ERROR err)
63{
64 switch (err) {
65 case AACENC_OK:
66 return "No error";
67 case AACENC_INVALID_HANDLE:
68 return "Invalid handle";
69 case AACENC_MEMORY_ERROR:
70 return "Memory allocation error";
71 case AACENC_UNSUPPORTED_PARAMETER:
72 return "Unsupported parameter";
73 case AACENC_INVALID_CONFIG:
74 return "Invalid config";
75 case AACENC_INIT_ERROR:
76 return "Initialization error";
77 case AACENC_INIT_AAC_ERROR:
78 return "AAC library initialization error";
79 case AACENC_INIT_SBR_ERROR:
80 return "SBR library initialization error";
81 case AACENC_INIT_TP_ERROR:
82 return "Transport library initialization error";
83 case AACENC_INIT_META_ERROR:
84 return "Metadata library initialization error";
85 case AACENC_ENCODE_ERROR:
86 return "Encoding error";
87 case AACENC_ENCODE_EOF:
88 return "End of file";
89 default:
90 return "Unknown error";
91 }
92}
93
94static int aac_encode_close(AVCodecContext *avctx)
95{
96 AACContext *s = avctx->priv_data;
97
98 if (s->handle)
99 aacEncClose(&s->handle);
37eeb5e2
MS
100 av_freep(&avctx->extradata);
101 ff_af_queue_close(&s->afq);
102
103 return 0;
104}
105
106static av_cold int aac_encode_init(AVCodecContext *avctx)
107{
108 AACContext *s = avctx->priv_data;
109 int ret = AVERROR(EINVAL);
110 AACENC_InfoStruct info = { 0 };
111 CHANNEL_MODE mode;
112 AACENC_ERROR err;
113 int aot = FF_PROFILE_AAC_LOW + 1;
114 int sce = 0, cpe = 0;
115
116 if ((err = aacEncOpen(&s->handle, 0, avctx->channels)) != AACENC_OK) {
117 av_log(avctx, AV_LOG_ERROR, "Unable to open the encoder: %s\n",
118 aac_get_error(err));
119 goto error;
120 }
121
122 if (avctx->profile != FF_PROFILE_UNKNOWN)
123 aot = avctx->profile + 1;
124
125 if ((err = aacEncoder_SetParam(s->handle, AACENC_AOT, aot)) != AACENC_OK) {
126 av_log(avctx, AV_LOG_ERROR, "Unable to set the AOT %d: %s\n",
127 aot, aac_get_error(err));
128 goto error;
129 }
130
131 if (aot == FF_PROFILE_AAC_ELD + 1 && s->eld_sbr) {
132 if ((err = aacEncoder_SetParam(s->handle, AACENC_SBR_MODE,
133 1)) != AACENC_OK) {
134 av_log(avctx, AV_LOG_ERROR, "Unable to enable SBR for ELD: %s\n",
135 aac_get_error(err));
136 goto error;
137 }
138 }
139
140 if ((err = aacEncoder_SetParam(s->handle, AACENC_SAMPLERATE,
141 avctx->sample_rate)) != AACENC_OK) {
142 av_log(avctx, AV_LOG_ERROR, "Unable to set the sample rate %d: %s\n",
143 avctx->sample_rate, aac_get_error(err));
144 goto error;
145 }
146
147 switch (avctx->channels) {
148 case 1: mode = MODE_1; sce = 1; cpe = 0; break;
149 case 2: mode = MODE_2; sce = 0; cpe = 1; break;
150 case 3: mode = MODE_1_2; sce = 1; cpe = 1; break;
151 case 4: mode = MODE_1_2_1; sce = 2; cpe = 1; break;
152 case 5: mode = MODE_1_2_2; sce = 1; cpe = 2; break;
153 case 6: mode = MODE_1_2_2_1; sce = 2; cpe = 2; break;
154 default:
155 av_log(avctx, AV_LOG_ERROR,
156 "Unsupported number of channels %d\n", avctx->channels);
157 goto error;
158 }
159
160 if ((err = aacEncoder_SetParam(s->handle, AACENC_CHANNELMODE,
161 mode)) != AACENC_OK) {
162 av_log(avctx, AV_LOG_ERROR,
163 "Unable to set channel mode %d: %s\n", mode, aac_get_error(err));
164 goto error;
165 }
166
167 if ((err = aacEncoder_SetParam(s->handle, AACENC_CHANNELORDER,
168 1)) != AACENC_OK) {
169 av_log(avctx, AV_LOG_ERROR,
170 "Unable to set wav channel order %d: %s\n",
171 mode, aac_get_error(err));
172 goto error;
173 }
174
375bff70
MS
175 if (avctx->flags & CODEC_FLAG_QSCALE || s->vbr) {
176 int mode = s->vbr ? s->vbr : avctx->global_quality;
37eeb5e2
MS
177 if (mode < 1 || mode > 5) {
178 av_log(avctx, AV_LOG_WARNING,
179 "VBR quality %d out of range, should be 1-5\n", mode);
180 mode = av_clip(mode, 1, 5);
181 }
a10190dc
MS
182 av_log(avctx, AV_LOG_WARNING,
183 "Note, the VBR setting is unsupported and only works with "
184 "some parameter combinations\n");
37eeb5e2
MS
185 if ((err = aacEncoder_SetParam(s->handle, AACENC_BITRATEMODE,
186 mode)) != AACENC_OK) {
187 av_log(avctx, AV_LOG_ERROR, "Unable to set the VBR bitrate mode %d: %s\n",
188 mode, aac_get_error(err));
189 goto error;
190 }
191 } else {
192 if (avctx->bit_rate <= 0) {
193 if (avctx->profile == FF_PROFILE_AAC_HE_V2) {
194 sce = 1;
195 cpe = 0;
196 }
197 avctx->bit_rate = (96*sce + 128*cpe) * avctx->sample_rate / 44;
198 if (avctx->profile == FF_PROFILE_AAC_HE ||
199 avctx->profile == FF_PROFILE_AAC_HE_V2 ||
95d52464 200 avctx->profile == FF_PROFILE_MPEG2_AAC_HE ||
37eeb5e2
MS
201 s->eld_sbr)
202 avctx->bit_rate /= 2;
203 }
204 if ((err = aacEncoder_SetParam(s->handle, AACENC_BITRATE,
205 avctx->bit_rate)) != AACENC_OK) {
206 av_log(avctx, AV_LOG_ERROR, "Unable to set the bitrate %d: %s\n",
207 avctx->bit_rate, aac_get_error(err));
208 goto error;
209 }
210 }
211
212 /* Choose bitstream format - if global header is requested, use
213 * raw access units, otherwise use ADTS. */
214 if ((err = aacEncoder_SetParam(s->handle, AACENC_TRANSMUX,
160a27c5 215 avctx->flags & CODEC_FLAG_GLOBAL_HEADER ? 0 : s->latm ? 10 : 2)) != AACENC_OK) {
37eeb5e2
MS
216 av_log(avctx, AV_LOG_ERROR, "Unable to set the transmux format: %s\n",
217 aac_get_error(err));
218 goto error;
219 }
220
160a27c5
KK
221 if (s->latm && s->header_period) {
222 if ((err = aacEncoder_SetParam(s->handle, AACENC_HEADER_PERIOD,
223 s->header_period)) != AACENC_OK) {
224 av_log(avctx, AV_LOG_ERROR, "Unable to set header period: %s\n",
225 aac_get_error(err));
226 goto error;
227 }
228 }
229
37eeb5e2
MS
230 /* If no signaling mode is chosen, use explicit hierarchical signaling
231 * if using mp4 mode (raw access units, with global header) and
232 * implicit signaling if using ADTS. */
233 if (s->signaling < 0)
234 s->signaling = avctx->flags & CODEC_FLAG_GLOBAL_HEADER ? 2 : 0;
235
236 if ((err = aacEncoder_SetParam(s->handle, AACENC_SIGNALING_MODE,
237 s->signaling)) != AACENC_OK) {
238 av_log(avctx, AV_LOG_ERROR, "Unable to set signaling mode %d: %s\n",
239 s->signaling, aac_get_error(err));
240 goto error;
241 }
242
243 if ((err = aacEncoder_SetParam(s->handle, AACENC_AFTERBURNER,
244 s->afterburner)) != AACENC_OK) {
245 av_log(avctx, AV_LOG_ERROR, "Unable to set afterburner to %d: %s\n",
246 s->afterburner, aac_get_error(err));
247 goto error;
248 }
249
7548831c 250 if (avctx->cutoff > 0) {
375ef652 251 if (avctx->cutoff < (avctx->sample_rate + 255) >> 8 || avctx->cutoff > 20000) {
c37c383e
MA
252 av_log(avctx, AV_LOG_ERROR, "cutoff valid range is %d-20000\n",
253 (avctx->sample_rate + 255) >> 8);
254 goto error;
255 }
7548831c
MS
256 if ((err = aacEncoder_SetParam(s->handle, AACENC_BANDWIDTH,
257 avctx->cutoff)) != AACENC_OK) {
511cf612 258 av_log(avctx, AV_LOG_ERROR, "Unable to set the encoder bandwidth to %d: %s\n",
7548831c
MS
259 avctx->cutoff, aac_get_error(err));
260 goto error;
261 }
262 }
263
37eeb5e2
MS
264 if ((err = aacEncEncode(s->handle, NULL, NULL, NULL, NULL)) != AACENC_OK) {
265 av_log(avctx, AV_LOG_ERROR, "Unable to initialize the encoder: %s\n",
266 aac_get_error(err));
267 return AVERROR(EINVAL);
268 }
269
270 if ((err = aacEncInfo(s->handle, &info)) != AACENC_OK) {
271 av_log(avctx, AV_LOG_ERROR, "Unable to get encoder info: %s\n",
272 aac_get_error(err));
273 goto error;
274 }
275
37eeb5e2
MS
276 avctx->frame_size = info.frameLength;
277 avctx->delay = info.encoderDelay;
278 ff_af_queue_init(avctx, &s->afq);
279
280 if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER) {
281 avctx->extradata_size = info.confSize;
282 avctx->extradata = av_mallocz(avctx->extradata_size +
283 FF_INPUT_BUFFER_PADDING_SIZE);
284 if (!avctx->extradata) {
285 ret = AVERROR(ENOMEM);
286 goto error;
287 }
288
289 memcpy(avctx->extradata, info.confBuf, info.confSize);
290 }
291 return 0;
292error:
293 aac_encode_close(avctx);
294 return ret;
295}
296
297static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
298 const AVFrame *frame, int *got_packet_ptr)
299{
300 AACContext *s = avctx->priv_data;
301 AACENC_BufDesc in_buf = { 0 }, out_buf = { 0 };
302 AACENC_InArgs in_args = { 0 };
303 AACENC_OutArgs out_args = { 0 };
304 int in_buffer_identifier = IN_AUDIO_DATA;
305 int in_buffer_size, in_buffer_element_size;
306 int out_buffer_identifier = OUT_BITSTREAM_DATA;
307 int out_buffer_size, out_buffer_element_size;
308 void *in_ptr, *out_ptr;
309 int ret;
310 AACENC_ERROR err;
311
312 /* handle end-of-stream small frame and flushing */
313 if (!frame) {
314 in_args.numInSamples = -1;
315 } else {
316 in_ptr = frame->data[0];
317 in_buffer_size = 2 * avctx->channels * frame->nb_samples;
318 in_buffer_element_size = 2;
319
320 in_args.numInSamples = avctx->channels * frame->nb_samples;
321 in_buf.numBufs = 1;
322 in_buf.bufs = &in_ptr;
323 in_buf.bufferIdentifiers = &in_buffer_identifier;
324 in_buf.bufSizes = &in_buffer_size;
325 in_buf.bufElSizes = &in_buffer_element_size;
326
327 /* add current frame to the queue */
1d7ffd06 328 if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
37eeb5e2
MS
329 return ret;
330 }
331
332 /* The maximum packet size is 6144 bits aka 768 bytes per channel. */
333 if ((ret = ff_alloc_packet(avpkt, FFMAX(8192, 768 * avctx->channels)))) {
334 av_log(avctx, AV_LOG_ERROR, "Error getting output packet\n");
335 return ret;
336 }
337
338 out_ptr = avpkt->data;
339 out_buffer_size = avpkt->size;
340 out_buffer_element_size = 1;
341 out_buf.numBufs = 1;
342 out_buf.bufs = &out_ptr;
343 out_buf.bufferIdentifiers = &out_buffer_identifier;
344 out_buf.bufSizes = &out_buffer_size;
345 out_buf.bufElSizes = &out_buffer_element_size;
346
347 if ((err = aacEncEncode(s->handle, &in_buf, &out_buf, &in_args,
348 &out_args)) != AACENC_OK) {
349 if (!frame && err == AACENC_ENCODE_EOF)
350 return 0;
351 av_log(avctx, AV_LOG_ERROR, "Unable to encode frame: %s\n",
352 aac_get_error(err));
353 return AVERROR(EINVAL);
354 }
355
356 if (!out_args.numOutBytes)
357 return 0;
358
359 /* Get the next frame pts & duration */
360 ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts,
361 &avpkt->duration);
362
363 avpkt->size = out_args.numOutBytes;
364 *got_packet_ptr = 1;
365 return 0;
366}
367
368static const AVProfile profiles[] = {
369 { FF_PROFILE_AAC_LOW, "LC" },
370 { FF_PROFILE_AAC_HE, "HE-AAC" },
371 { FF_PROFILE_AAC_HE_V2, "HE-AACv2" },
372 { FF_PROFILE_AAC_LD, "LD" },
373 { FF_PROFILE_AAC_ELD, "ELD" },
374 { FF_PROFILE_UNKNOWN },
375};
376
377static const AVCodecDefault aac_encode_defaults[] = {
378 { "b", "0" },
379 { NULL }
380};
381
382static const uint64_t aac_channel_layout[] = {
383 AV_CH_LAYOUT_MONO,
384 AV_CH_LAYOUT_STEREO,
385 AV_CH_LAYOUT_SURROUND,
386 AV_CH_LAYOUT_4POINT0,
387 AV_CH_LAYOUT_5POINT0_BACK,
388 AV_CH_LAYOUT_5POINT1_BACK,
389 0,
390};
391
94a69dee
NC
392static const int aac_sample_rates[] = {
393 96000, 88200, 64000, 48000, 44100, 32000,
394 24000, 22050, 16000, 12000, 11025, 8000, 0
395};
396
37eeb5e2 397AVCodec ff_libfdk_aac_encoder = {
c8ba8be2 398 .name = "libfdk_aac",
b2bed932 399 .long_name = NULL_IF_CONFIG_SMALL("Fraunhofer FDK AAC"),
c8ba8be2
NC
400 .type = AVMEDIA_TYPE_AUDIO,
401 .id = AV_CODEC_ID_AAC,
402 .priv_data_size = sizeof(AACContext),
403 .init = aac_encode_init,
404 .encode2 = aac_encode_frame,
405 .close = aac_encode_close,
406 .capabilities = CODEC_CAP_SMALL_LAST_FRAME | CODEC_CAP_DELAY,
407 .sample_fmts = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_S16,
408 AV_SAMPLE_FMT_NONE },
c8ba8be2
NC
409 .priv_class = &aac_enc_class,
410 .defaults = aac_encode_defaults,
411 .profiles = profiles,
94a69dee 412 .supported_samplerates = aac_sample_rates,
c8ba8be2 413 .channel_layouts = aac_channel_layout,
37eeb5e2 414};