lavu: rename audioconvert.* to channel_layout.* and deprecate audioconvert.h
[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
24#include "avcodec.h"
25#include "audio_frame_queue.h"
26#include "internal.h"
27#include "libavutil/audioconvert.h"
1d9c2dc8 28#include "libavutil/common.h"
37eeb5e2
MS
29#include "libavutil/opt.h"
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);
100#if FF_API_OLD_ENCODE_AUDIO
101 av_freep(&avctx->coded_frame);
102#endif
103 av_freep(&avctx->extradata);
104 ff_af_queue_close(&s->afq);
105
106 return 0;
107}
108
109static av_cold int aac_encode_init(AVCodecContext *avctx)
110{
111 AACContext *s = avctx->priv_data;
112 int ret = AVERROR(EINVAL);
113 AACENC_InfoStruct info = { 0 };
114 CHANNEL_MODE mode;
115 AACENC_ERROR err;
116 int aot = FF_PROFILE_AAC_LOW + 1;
117 int sce = 0, cpe = 0;
118
119 if ((err = aacEncOpen(&s->handle, 0, avctx->channels)) != AACENC_OK) {
120 av_log(avctx, AV_LOG_ERROR, "Unable to open the encoder: %s\n",
121 aac_get_error(err));
122 goto error;
123 }
124
125 if (avctx->profile != FF_PROFILE_UNKNOWN)
126 aot = avctx->profile + 1;
127
128 if ((err = aacEncoder_SetParam(s->handle, AACENC_AOT, aot)) != AACENC_OK) {
129 av_log(avctx, AV_LOG_ERROR, "Unable to set the AOT %d: %s\n",
130 aot, aac_get_error(err));
131 goto error;
132 }
133
134 if (aot == FF_PROFILE_AAC_ELD + 1 && s->eld_sbr) {
135 if ((err = aacEncoder_SetParam(s->handle, AACENC_SBR_MODE,
136 1)) != AACENC_OK) {
137 av_log(avctx, AV_LOG_ERROR, "Unable to enable SBR for ELD: %s\n",
138 aac_get_error(err));
139 goto error;
140 }
141 }
142
143 if ((err = aacEncoder_SetParam(s->handle, AACENC_SAMPLERATE,
144 avctx->sample_rate)) != AACENC_OK) {
145 av_log(avctx, AV_LOG_ERROR, "Unable to set the sample rate %d: %s\n",
146 avctx->sample_rate, aac_get_error(err));
147 goto error;
148 }
149
150 switch (avctx->channels) {
151 case 1: mode = MODE_1; sce = 1; cpe = 0; break;
152 case 2: mode = MODE_2; sce = 0; cpe = 1; break;
153 case 3: mode = MODE_1_2; sce = 1; cpe = 1; break;
154 case 4: mode = MODE_1_2_1; sce = 2; cpe = 1; break;
155 case 5: mode = MODE_1_2_2; sce = 1; cpe = 2; break;
156 case 6: mode = MODE_1_2_2_1; sce = 2; cpe = 2; break;
157 default:
158 av_log(avctx, AV_LOG_ERROR,
159 "Unsupported number of channels %d\n", avctx->channels);
160 goto error;
161 }
162
163 if ((err = aacEncoder_SetParam(s->handle, AACENC_CHANNELMODE,
164 mode)) != AACENC_OK) {
165 av_log(avctx, AV_LOG_ERROR,
166 "Unable to set channel mode %d: %s\n", mode, aac_get_error(err));
167 goto error;
168 }
169
170 if ((err = aacEncoder_SetParam(s->handle, AACENC_CHANNELORDER,
171 1)) != AACENC_OK) {
172 av_log(avctx, AV_LOG_ERROR,
173 "Unable to set wav channel order %d: %s\n",
174 mode, aac_get_error(err));
175 goto error;
176 }
177
375bff70
MS
178 if (avctx->flags & CODEC_FLAG_QSCALE || s->vbr) {
179 int mode = s->vbr ? s->vbr : avctx->global_quality;
37eeb5e2
MS
180 if (mode < 1 || mode > 5) {
181 av_log(avctx, AV_LOG_WARNING,
182 "VBR quality %d out of range, should be 1-5\n", mode);
183 mode = av_clip(mode, 1, 5);
184 }
a10190dc
MS
185 av_log(avctx, AV_LOG_WARNING,
186 "Note, the VBR setting is unsupported and only works with "
187 "some parameter combinations\n");
37eeb5e2
MS
188 if ((err = aacEncoder_SetParam(s->handle, AACENC_BITRATEMODE,
189 mode)) != AACENC_OK) {
190 av_log(avctx, AV_LOG_ERROR, "Unable to set the VBR bitrate mode %d: %s\n",
191 mode, aac_get_error(err));
192 goto error;
193 }
194 } else {
195 if (avctx->bit_rate <= 0) {
196 if (avctx->profile == FF_PROFILE_AAC_HE_V2) {
197 sce = 1;
198 cpe = 0;
199 }
200 avctx->bit_rate = (96*sce + 128*cpe) * avctx->sample_rate / 44;
201 if (avctx->profile == FF_PROFILE_AAC_HE ||
202 avctx->profile == FF_PROFILE_AAC_HE_V2 ||
203 s->eld_sbr)
204 avctx->bit_rate /= 2;
205 }
206 if ((err = aacEncoder_SetParam(s->handle, AACENC_BITRATE,
207 avctx->bit_rate)) != AACENC_OK) {
208 av_log(avctx, AV_LOG_ERROR, "Unable to set the bitrate %d: %s\n",
209 avctx->bit_rate, aac_get_error(err));
210 goto error;
211 }
212 }
213
214 /* Choose bitstream format - if global header is requested, use
215 * raw access units, otherwise use ADTS. */
216 if ((err = aacEncoder_SetParam(s->handle, AACENC_TRANSMUX,
160a27c5 217 avctx->flags & CODEC_FLAG_GLOBAL_HEADER ? 0 : s->latm ? 10 : 2)) != AACENC_OK) {
37eeb5e2
MS
218 av_log(avctx, AV_LOG_ERROR, "Unable to set the transmux format: %s\n",
219 aac_get_error(err));
220 goto error;
221 }
222
160a27c5
KK
223 if (s->latm && s->header_period) {
224 if ((err = aacEncoder_SetParam(s->handle, AACENC_HEADER_PERIOD,
225 s->header_period)) != AACENC_OK) {
226 av_log(avctx, AV_LOG_ERROR, "Unable to set header period: %s\n",
227 aac_get_error(err));
228 goto error;
229 }
230 }
231
37eeb5e2
MS
232 /* If no signaling mode is chosen, use explicit hierarchical signaling
233 * if using mp4 mode (raw access units, with global header) and
234 * implicit signaling if using ADTS. */
235 if (s->signaling < 0)
236 s->signaling = avctx->flags & CODEC_FLAG_GLOBAL_HEADER ? 2 : 0;
237
238 if ((err = aacEncoder_SetParam(s->handle, AACENC_SIGNALING_MODE,
239 s->signaling)) != AACENC_OK) {
240 av_log(avctx, AV_LOG_ERROR, "Unable to set signaling mode %d: %s\n",
241 s->signaling, aac_get_error(err));
242 goto error;
243 }
244
245 if ((err = aacEncoder_SetParam(s->handle, AACENC_AFTERBURNER,
246 s->afterburner)) != AACENC_OK) {
247 av_log(avctx, AV_LOG_ERROR, "Unable to set afterburner to %d: %s\n",
248 s->afterburner, aac_get_error(err));
249 goto error;
250 }
251
7548831c 252 if (avctx->cutoff > 0) {
c37c383e
MA
253 if (avctx->cutoff < (avctx->sample_rate + 255) >> 8) {
254 av_log(avctx, AV_LOG_ERROR, "cutoff valid range is %d-20000\n",
255 (avctx->sample_rate + 255) >> 8);
256 goto error;
257 }
7548831c
MS
258 if ((err = aacEncoder_SetParam(s->handle, AACENC_BANDWIDTH,
259 avctx->cutoff)) != AACENC_OK) {
260 av_log(avctx, AV_LOG_ERROR, "Unable to set the encoder bandwith to %d: %s\n",
261 avctx->cutoff, aac_get_error(err));
262 goto error;
263 }
264 }
265
37eeb5e2
MS
266 if ((err = aacEncEncode(s->handle, NULL, NULL, NULL, NULL)) != AACENC_OK) {
267 av_log(avctx, AV_LOG_ERROR, "Unable to initialize the encoder: %s\n",
268 aac_get_error(err));
269 return AVERROR(EINVAL);
270 }
271
272 if ((err = aacEncInfo(s->handle, &info)) != AACENC_OK) {
273 av_log(avctx, AV_LOG_ERROR, "Unable to get encoder info: %s\n",
274 aac_get_error(err));
275 goto error;
276 }
277
278#if FF_API_OLD_ENCODE_AUDIO
279 avctx->coded_frame = avcodec_alloc_frame();
280 if (!avctx->coded_frame) {
281 ret = AVERROR(ENOMEM);
282 goto error;
283 }
284#endif
285 avctx->frame_size = info.frameLength;
286 avctx->delay = info.encoderDelay;
287 ff_af_queue_init(avctx, &s->afq);
288
289 if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER) {
290 avctx->extradata_size = info.confSize;
291 avctx->extradata = av_mallocz(avctx->extradata_size +
292 FF_INPUT_BUFFER_PADDING_SIZE);
293 if (!avctx->extradata) {
294 ret = AVERROR(ENOMEM);
295 goto error;
296 }
297
298 memcpy(avctx->extradata, info.confBuf, info.confSize);
299 }
300 return 0;
301error:
302 aac_encode_close(avctx);
303 return ret;
304}
305
306static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
307 const AVFrame *frame, int *got_packet_ptr)
308{
309 AACContext *s = avctx->priv_data;
310 AACENC_BufDesc in_buf = { 0 }, out_buf = { 0 };
311 AACENC_InArgs in_args = { 0 };
312 AACENC_OutArgs out_args = { 0 };
313 int in_buffer_identifier = IN_AUDIO_DATA;
314 int in_buffer_size, in_buffer_element_size;
315 int out_buffer_identifier = OUT_BITSTREAM_DATA;
316 int out_buffer_size, out_buffer_element_size;
317 void *in_ptr, *out_ptr;
318 int ret;
319 AACENC_ERROR err;
320
321 /* handle end-of-stream small frame and flushing */
322 if (!frame) {
323 in_args.numInSamples = -1;
324 } else {
325 in_ptr = frame->data[0];
326 in_buffer_size = 2 * avctx->channels * frame->nb_samples;
327 in_buffer_element_size = 2;
328
329 in_args.numInSamples = avctx->channels * frame->nb_samples;
330 in_buf.numBufs = 1;
331 in_buf.bufs = &in_ptr;
332 in_buf.bufferIdentifiers = &in_buffer_identifier;
333 in_buf.bufSizes = &in_buffer_size;
334 in_buf.bufElSizes = &in_buffer_element_size;
335
336 /* add current frame to the queue */
337 if ((ret = ff_af_queue_add(&s->afq, frame) < 0))
338 return ret;
339 }
340
341 /* The maximum packet size is 6144 bits aka 768 bytes per channel. */
342 if ((ret = ff_alloc_packet(avpkt, FFMAX(8192, 768 * avctx->channels)))) {
343 av_log(avctx, AV_LOG_ERROR, "Error getting output packet\n");
344 return ret;
345 }
346
347 out_ptr = avpkt->data;
348 out_buffer_size = avpkt->size;
349 out_buffer_element_size = 1;
350 out_buf.numBufs = 1;
351 out_buf.bufs = &out_ptr;
352 out_buf.bufferIdentifiers = &out_buffer_identifier;
353 out_buf.bufSizes = &out_buffer_size;
354 out_buf.bufElSizes = &out_buffer_element_size;
355
356 if ((err = aacEncEncode(s->handle, &in_buf, &out_buf, &in_args,
357 &out_args)) != AACENC_OK) {
358 if (!frame && err == AACENC_ENCODE_EOF)
359 return 0;
360 av_log(avctx, AV_LOG_ERROR, "Unable to encode frame: %s\n",
361 aac_get_error(err));
362 return AVERROR(EINVAL);
363 }
364
365 if (!out_args.numOutBytes)
366 return 0;
367
368 /* Get the next frame pts & duration */
369 ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts,
370 &avpkt->duration);
371
372 avpkt->size = out_args.numOutBytes;
373 *got_packet_ptr = 1;
374 return 0;
375}
376
377static const AVProfile profiles[] = {
378 { FF_PROFILE_AAC_LOW, "LC" },
379 { FF_PROFILE_AAC_HE, "HE-AAC" },
380 { FF_PROFILE_AAC_HE_V2, "HE-AACv2" },
381 { FF_PROFILE_AAC_LD, "LD" },
382 { FF_PROFILE_AAC_ELD, "ELD" },
383 { FF_PROFILE_UNKNOWN },
384};
385
386static const AVCodecDefault aac_encode_defaults[] = {
387 { "b", "0" },
388 { NULL }
389};
390
391static const uint64_t aac_channel_layout[] = {
392 AV_CH_LAYOUT_MONO,
393 AV_CH_LAYOUT_STEREO,
394 AV_CH_LAYOUT_SURROUND,
395 AV_CH_LAYOUT_4POINT0,
396 AV_CH_LAYOUT_5POINT0_BACK,
397 AV_CH_LAYOUT_5POINT1_BACK,
398 0,
399};
400
94a69dee
NC
401static const int aac_sample_rates[] = {
402 96000, 88200, 64000, 48000, 44100, 32000,
403 24000, 22050, 16000, 12000, 11025, 8000, 0
404};
405
37eeb5e2 406AVCodec ff_libfdk_aac_encoder = {
c8ba8be2
NC
407 .name = "libfdk_aac",
408 .type = AVMEDIA_TYPE_AUDIO,
409 .id = AV_CODEC_ID_AAC,
410 .priv_data_size = sizeof(AACContext),
411 .init = aac_encode_init,
412 .encode2 = aac_encode_frame,
413 .close = aac_encode_close,
414 .capabilities = CODEC_CAP_SMALL_LAST_FRAME | CODEC_CAP_DELAY,
415 .sample_fmts = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_S16,
416 AV_SAMPLE_FMT_NONE },
417 .long_name = NULL_IF_CONFIG_SMALL("Fraunhofer FDK AAC"),
418 .priv_class = &aac_enc_class,
419 .defaults = aac_encode_defaults,
420 .profiles = profiles,
94a69dee 421 .supported_samplerates = aac_sample_rates,
c8ba8be2 422 .channel_layouts = aac_channel_layout,
37eeb5e2 423};