?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
fdkaac: simplify parameters checks
[vlc/gmpfix.git] / modules / codec / fdkaac.c
blob59960d19827a8846c6ab3a643e71bbb2a147f210
1 /*****************************************************************************
2 * aac.c: FDK-AAC Encoder plugin for vlc.
3 *****************************************************************************
4 * Copyright (C) 2012 Sergio Ammirata
5 *
6 * Authors: Sergio Ammirata <sergio@ammirata.net>
7 *
8 * This library is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU Lesser General Public
10 * License as published by the Free Software Foundation; either
11 * version 2.1 of the License, or (at your option) any later version.
12 *
13 * This library is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 * Lesser General Public License for more details.
17 *
18 * You should have received a copy of the GNU Lesser General Public
19 * License along with this library; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 *
22 * Alternatively you can redistribute this file under the terms of the
23 * BSD license as stated below:
24 *
25 * Redistribution and use in source and binary forms, with or without
26 * modification, are permitted provided that the following conditions
27 * are met:
28 * 1. Redistributions of source code must retain the above copyright
29 * notice, this list of conditions and the following disclaimer.
30 * 2. Redistributions in binary form must reproduce the above copyright
31 * notice, this list of conditions and the following disclaimer in
32 * the documentation and/or other materials provided with the
33 * distribution.
34 *
35 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
36 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
37 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
38 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
39 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
40 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
41 * TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
42 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
43 * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
44 * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
45 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
46 *
47 *****************************************************************************/
48
49 /*****************************************************************************
50 * Preamble
51 *****************************************************************************/
52 #ifdef HAVE_CONFIG_H
53 # include "config.h"
54 #endif
55
56 #include <fdk-aac/aacenc_lib.h>
57
58 #include <vlc_common.h>
59 #include <vlc_plugin.h>
60 #include <vlc_codec.h>
61
62 static int OpenEncoder(vlc_object_t *);
63 static void CloseEncoder(vlc_object_t *);
64
65 #define ENC_CFG_PREFIX "sout-fdkaac-"
66
67 #define AOT_TEXT N_("Encoder Profile")
68 #define AOT_LONGTEXT N_("Encoder Algorithm to use")
69
70 #define SIDEBAND_TEXT N_("Enable spectral band replication")
71 #define SIDEBAND_LONGTEXT N_("This is an optional feature only for the AAC-ELD profile")
72
73 #define VBR_QUALITY_TEXT N_("VBR Quality")
74 #define VBR_QUALITY_LONGTEXT N_("Quality of the VBR Encoding (0=cbr, 1-5 constant quality vbr, 5 is best")
75
76 #define AFTERBURNER_TEXT N_("Enable afterburner library")
77 #define AFTERBURNER_LONGTEXT N_("This library will produce higher quality audio at the expense of additional CPU usage (default is enabled)")
78
79 #define SIGNALING_TEXT N_("Signaling mode of the extension AOT")
80 #define SIGNALING_LONGTEXT N_("1 is explicit for SBR and implicit for PS (default), 2 is explicit hierarchical")
81
82 #define CH_ORDER_MPEG 0 /*!< MPEG channel ordering (e. g. 5.1: C, L, R, SL, SR, LFE) */
83 #define CH_ORDER_WAV 1 /*!< WAV fileformat channel ordering (e. g. 5.1: L, R, C, LFE, SL, SR) */
84 #define CH_ORDER_WG4 2 /*!< WG4 fileformat channel ordering (e. g. 5.1: L, R, SL, SR, C, LFE) */
85
86 #define PROFILE_AAC_LC 2
87 #define PROFILE_AAC_HE 5
88 #define PROFILE_AAC_HE_v2 29
89 #define PROFILE_AAC_LD 23
90 #define PROFILE_AAC_ELD 39
91
92 #define SIGNALING_COMPATIBLE 1
93 #define SIGNALING_HIERARCHICAL 2
94
95 static const int pi_aot_values[] = { PROFILE_AAC_LC, PROFILE_AAC_HE, PROFILE_AAC_HE_v2, PROFILE_AAC_LD, PROFILE_AAC_ELD };
96 static const char *const ppsz_aot_descriptions[] =
97 { N_("AAC-LC"), N_("HE-AAC"), N_("HE-AAC-v2"), N_("AAC-LD"), N_("AAC-ELD") };
98
99 vlc_module_begin ()
100 set_shortname(N_("FDKAAC"))
101 set_description(N_("FDK-AAC Audio encoder"))
102 set_capability("encoder", 150)
103 set_callbacks(OpenEncoder, CloseEncoder)
104 add_shortcut("fdkaac")
105 set_category(CAT_INPUT)
106 set_subcategory(SUBCAT_INPUT_ACODEC)
107 add_integer(ENC_CFG_PREFIX "profile", PROFILE_AAC_LC, AOT_TEXT,
108 AOT_LONGTEXT, false)
109 change_integer_list(pi_aot_values, ppsz_aot_descriptions);
110 add_bool(ENC_CFG_PREFIX "sbr", false, SIDEBAND_TEXT,
111 SIDEBAND_LONGTEXT, false)
112 add_integer(ENC_CFG_PREFIX "vbr", 0, VBR_QUALITY_TEXT,
113 VBR_QUALITY_LONGTEXT, false)
114 change_integer_range (0, 5)
115 add_bool(ENC_CFG_PREFIX "afterburner", true, AFTERBURNER_TEXT,
116 AFTERBURNER_LONGTEXT, true)
117 add_integer(ENC_CFG_PREFIX "signaling", SIGNALING_COMPATIBLE, SIGNALING_TEXT,
118 SIGNALING_LONGTEXT, true)
119 change_integer_range (0, 2)
120 vlc_module_end ()
121
122 /*****************************************************************************
123 * Local prototypes
124 *****************************************************************************/
125 static block_t *EncodeAudio(encoder_t *p_enc, block_t *p_buf);
126
127 static const char *const ppsz_enc_options[] = {
128 "profile", "sbr", "vbr", "afterburner", "signaling", NULL
129 };
130
131 /*****************************************************************************
132 * encoder_sys_t : aac encoder descriptor
133 *****************************************************************************/
134 struct encoder_sys_t
135 {
136 double d_compression_ratio;
137 mtime_t i_pts_last;
138 int i_encoderdelay; /* Samples delay introduced by the profile */
139 int i_frame_size;
140 int i_maxoutputsize; /* Maximum buffer size for encoded output */
141 HANDLE_AACENCODER handle;
142 };
143
144 static const char *aac_get_errorstring(AACENC_ERROR erraac)
145 {
146 switch (erraac) {
147 case AACENC_OK:
148 return "No error";
149 case AACENC_INVALID_HANDLE:
150 return "Invalid handle";
151 case AACENC_MEMORY_ERROR:
152 return "Memory allocation error";
153 case AACENC_UNSUPPORTED_PARAMETER:
154 return "Unsupported parameter";
155 case AACENC_INVALID_CONFIG:
156 return "Invalid config";
157 case AACENC_INIT_ERROR:
158 return "Initialization error";
159 case AACENC_INIT_AAC_ERROR:
160 return "AAC library initialization error";
161 case AACENC_INIT_SBR_ERROR:
162 return "SBR library initialization error";
163 case AACENC_INIT_TP_ERROR:
164 return "Transport library initialization error";
165 case AACENC_INIT_META_ERROR:
166 return "Metadata library initialization error";
167 case AACENC_ENCODE_ERROR:
168 return "Encoding error";
169 case AACENC_ENCODE_EOF:
170 return "End of file";
171 default:
172 return "Unknown error";
173 }
174 }
175
176 /*****************************************************************************
177 * OpenDecoder: open the encoder.
178 *****************************************************************************/
179 static int OpenEncoder(vlc_object_t *p_this)
180 {
181 encoder_t *p_enc = (encoder_t *)p_this;
182
183 config_ChainParse(p_enc, ENC_CFG_PREFIX, ppsz_enc_options, p_enc->p_cfg);
184
185 int i_aot;
186 switch (p_enc->fmt_out.i_codec) {
187 case VLC_CODEC_MP4A:
188 i_aot = var_InheritInteger(p_enc, ENC_CFG_PREFIX "profile");
189 break;
190 case VLC_FOURCC('l', 'a', 'a', 'c'):
191 i_aot = PROFILE_AAC_LC;
192 break;
193 case VLC_FOURCC('h', 'a', 'a', 'c'):
194 i_aot = PROFILE_AAC_HE;
195 break;
196 case VLC_FOURCC('s', 'a', 'a', 'c'):
197 i_aot = PROFILE_AAC_HE_v2;
198 break;
199 default:
200 return VLC_EGENERIC;
201 }
202
203 if (p_enc->fmt_in.audio.i_channels != 2)
204 if (i_aot == PROFILE_AAC_HE_v2 || i_aot == PROFILE_AAC_ELD) {
205 msg_Err(p_enc, "Selected profile %d can only be used with stereo", i_aot);
206 return VLC_EGENERIC;
207 }
208
209 uint16_t channel_config;
210 CHANNEL_MODE mode;
211 switch (p_enc->fmt_in.audio.i_channels) {
212 case 1: mode = MODE_1; channel_config = AOUT_CHAN_CENTER; break;
213 case 2: mode = MODE_2; channel_config = AOUT_CHANS_STEREO; break;
214 case 3: mode = MODE_1_2; channel_config = AOUT_CHANS_3_0; break;
215 case 4: mode = MODE_1_2_1; channel_config = AOUT_CHANS_4_CENTER_REAR; break;
216 case 5: mode = MODE_1_2_2; channel_config = AOUT_CHANS_5_0; break;
217 case 6: mode = MODE_1_2_2_1; channel_config = AOUT_CHANS_5_1; break;
218 case 8: mode = MODE_1_2_2_2_1; channel_config = AOUT_CHANS_7_1; break;
219 default:
220 msg_Err(p_enc, "we do not support > 8 input channels, this input has %i",
221 p_enc->fmt_in.audio.i_channels);
222 return VLC_EGENERIC;
223 }
224
225 p_enc->fmt_in.audio.i_physical_channels = channel_config;
226
227 msg_Info(p_enc, "Initializing AAC Encoder, %i channels", p_enc->fmt_in.audio.i_channels);
228
229 /* Allocate the memory needed to store the encoder's structure */
230 encoder_sys_t *p_sys = (encoder_sys_t *)malloc(sizeof(encoder_sys_t));
231 if (unlikely(!p_sys))
232 return VLC_ENOMEM;
233 p_enc->p_sys = p_sys;
234 p_enc->fmt_in.i_codec = VLC_CODEC_S16N;
235 p_enc->fmt_out.i_cat = AUDIO_ES;
236 p_enc->fmt_out.i_codec = VLC_CODEC_MP4A;
237
238 p_sys->i_pts_last = 0;
239
240 AACENC_ERROR erraac;
241 erraac = aacEncOpen(&p_sys->handle, 0, p_enc->fmt_in.audio.i_channels);
242 if (erraac != AACENC_OK) {
243 msg_Err(p_enc, "Unable to open encoder: %s", aac_get_errorstring(erraac));
244 free(p_sys);
245 return VLC_EGENERIC;
246 }
247
248 #define SET_PARAM(P, V) do { \
249 AACENC_ERROR err = aacEncoder_SetParam(p_sys->handle, AACENC_ ## P, V); \
250 if (err != AACENC_OK) { \
251 msg_Err(p_enc, "Couldn't set " #P " to value %d: %s", V, aac_get_errorstring(err)); \
252 goto error; \
253 } \
254 } while(0)
255
256 SET_PARAM(AOT, i_aot);
257 bool b_eld_sbr = var_InheritBool(p_enc, ENC_CFG_PREFIX "sbr");
258 if (i_aot == PROFILE_AAC_ELD && b_eld_sbr)
259 SET_PARAM(SBR_MODE, 1);
260 SET_PARAM(SAMPLERATE, p_enc->fmt_out.audio.i_rate);
261 SET_PARAM(CHANNELMODE, mode);
262 SET_PARAM(CHANNELORDER, CH_ORDER_WG4);
263
264 int i_vbr = var_InheritInteger(p_enc, ENC_CFG_PREFIX "vbr");
265 if (i_vbr != 0) {
266 if ((i_aot == PROFILE_AAC_HE || i_aot == PROFILE_AAC_HE_v2) && i_vbr > 3) {
267 msg_Warn(p_enc, "Maximum VBR quality for this profile is 3, setting vbr=3");
268 i_vbr = 3;
269 }
270 SET_PARAM(BITRATEMODE, i_vbr);
271 } else {
272 int i_bitrate = p_enc->fmt_out.i_bitrate;
273 if (i_bitrate == 0) {
274 i_bitrate = 96 * p_enc->fmt_in.audio.i_channels * p_enc->fmt_out.audio.i_rate / 44;
275 if (i_aot == PROFILE_AAC_HE || i_aot == PROFILE_AAC_HE_v2 || b_eld_sbr)
276 i_bitrate /= 2;
277 p_enc->fmt_out.i_bitrate = i_bitrate;
278 msg_Info(p_enc, "Setting optimal bitrate of %i", i_bitrate);
279 }
280 SET_PARAM(BITRATE, i_bitrate);
281 }
282 SET_PARAM(TRANSMUX, 0);
283 SET_PARAM(SIGNALING_MODE, (int)var_InheritInteger(p_enc, ENC_CFG_PREFIX "signaling"));
284 SET_PARAM(AFTERBURNER, !!var_InheritBool(p_enc, ENC_CFG_PREFIX "afterburner"));
285 #undef SET_PARAM
286
287 erraac = aacEncEncode(p_sys->handle, NULL, NULL, NULL, NULL);
288 if (erraac != AACENC_OK) {
289 msg_Err(p_enc, "Unable to initialize the encoder: %s", aac_get_errorstring(erraac));
290 goto error;
291 }
292
293 AACENC_InfoStruct info = { 0 };
294 erraac = aacEncInfo(p_sys->handle, &info);
295 if (erraac != AACENC_OK) {
296 msg_Err(p_enc, "Unable to get the encoder info: %s", aac_get_errorstring(erraac));
297 goto error;
298 }
299
300 /* The maximum packet size is 6144 bits aka 768 bytes per channel. */
301 p_sys->i_maxoutputsize = 768*p_enc->fmt_in.audio.i_channels;
302 p_enc->fmt_in.audio.i_bitspersample = 16;
303 p_sys->i_frame_size = info.frameLength;
304 p_sys->i_encoderdelay = info.encoderDelay;
305
306 p_enc->fmt_out.i_extra = info.confSize;
307 if (p_enc->fmt_out.i_extra) {
308 p_enc->fmt_out.p_extra = malloc(p_enc->fmt_out.i_extra);
309 if (p_enc->fmt_out.p_extra == NULL) {
310 msg_Err(p_enc, "Unable to allocate fmt_out.p_extra");
311 goto error;
312 }
313 memcpy(p_enc->fmt_out.p_extra, info.confBuf, p_enc->fmt_out.i_extra);
314 }
315
316 p_enc->pf_encode_audio = EncodeAudio;
317
318 #ifndef NDEBUG
319 // TODO: Add more debug info to this config printout
320 msg_Dbg(p_enc, "fmt_out.p_extra = %i", p_enc->fmt_out.i_extra);
321 #endif
322
323 return VLC_SUCCESS;
324
325 error:
326 CloseEncoder(p_this);
327 return VLC_EGENERIC;
328 }
329
330 /****************************************************************************
331 * EncodeAudio: the whole thing
332 ****************************************************************************/
333 static block_t *EncodeAudio(encoder_t *p_enc, block_t *p_aout_buf)
334 {
335 int16_t *p_buffer;
336 int i_samples;
337 mtime_t i_pts_out;
338
339 encoder_sys_t *p_sys = p_enc->p_sys;
340
341 if (likely(p_aout_buf)) {
342 p_buffer = (int16_t *)p_aout_buf->p_buffer;
343 i_samples = p_aout_buf->i_nb_samples;
344 i_pts_out = p_aout_buf->i_pts - (mtime_t)((double)CLOCK_FREQ *
345 (double)p_sys->i_encoderdelay /
346 (double)p_enc->fmt_out.audio.i_rate);
347 if (p_sys->i_pts_last == 0)
348 p_sys->i_pts_last = i_pts_out - (mtime_t)((double)CLOCK_FREQ *
349 (double)(p_sys->i_frame_size) /
350 (double)p_enc->fmt_out.audio.i_rate);
351 } else {
352 i_samples = 0;
353 i_pts_out = p_sys->i_pts_last;
354 }
355
356 int i_samples_left = i_samples;
357 int i_loop_count = 0;
358
359 block_t *p_chain = NULL;
360 while (i_samples_left >= 0) {
361 AACENC_BufDesc in_buf = { 0 }, out_buf = { 0 };
362 AACENC_InArgs in_args = { 0 };
363 AACENC_OutArgs out_args = { 0 };
364 int in_identifier = IN_AUDIO_DATA;
365 int in_size, in_elem_size;
366 int out_identifier = OUT_BITSTREAM_DATA;
367 int out_size, out_elem_size;
368 void *in_ptr, *out_ptr;
369
370 if (unlikely(i_samples == 0)) {
371 // this forces the encoder to purge whatever is left in the internal buffer
372 in_args.numInSamples = -1;
373 } else {
374 in_ptr = p_buffer + (i_samples - i_samples_left)*p_enc->fmt_in.audio.i_channels;
375 in_size = 2*p_enc->fmt_in.audio.i_channels*i_samples_left;
376 in_elem_size = 2;
377 in_args.numInSamples = p_enc->fmt_in.audio.i_channels*i_samples_left;
378 in_buf.numBufs = 1;
379 in_buf.bufs = &in_ptr;
380 in_buf.bufferIdentifiers = &in_identifier;
381 in_buf.bufSizes = &in_size;
382 in_buf.bufElSizes = &in_elem_size;
383 }
384 block_t *p_block;
385 p_block = block_Alloc(p_sys->i_maxoutputsize);
386 p_block->i_buffer = p_sys->i_maxoutputsize;
387 out_ptr = p_block->p_buffer;
388 out_size = p_block->i_buffer;
389 out_elem_size = 1;
390 out_buf.numBufs = 1;
391 out_buf.bufs = &out_ptr;
392 out_buf.bufferIdentifiers = &out_identifier;
393 out_buf.bufSizes = &out_size;
394 out_buf.bufElSizes = &out_elem_size;
395
396 AACENC_ERROR erraac;
397 if ((erraac = aacEncEncode(p_sys->handle, &in_buf, &out_buf, &in_args, &out_args)) != AACENC_OK) {
398 if (erraac == AACENC_ENCODE_EOF) {
399 msg_Info(p_enc, "Encoding final bytes (EOF)");
400 } else {
401 msg_Err(p_enc, "Encoding failed: %s", aac_get_errorstring(erraac));
402 block_Release(p_block);
403 break;
404 }
405 }
406 if (out_args.numOutBytes > 0) {
407 p_block->i_buffer = out_args.numOutBytes;
408 if (unlikely(i_samples == 0)) {
409 // I only have the numOutBytes so approximate based on compression factor
410 double d_samples_forward = p_sys->d_compression_ratio*(double)out_args.numOutBytes;
411 i_pts_out += (mtime_t)d_samples_forward;
412 p_block->i_length = (mtime_t)d_samples_forward;
413 // TODO: It would be more precise (a few microseconds) to use d_samples_forward =
414 // (mtime_t)CLOCK_FREQ * (mtime_t)p_sys->i_frame_size/(mtime_t)p_enc->fmt_out.audio.i_rate
415 // but I am not sure if the lib always outputs a full frame when
416 // emptying the internal buffer in the EOF scenario
417 } else {
418 if (i_loop_count == 0) {
419 // There can be an implicit delay in the first loop cycle because leftover bytes
420 // in the library buffer from the prior block
421 double d_samples_delay = (double)p_sys->i_frame_size - (double)out_args.numInSamples /
422 (double)p_enc->fmt_in.audio.i_channels;
423 i_pts_out -= (mtime_t)((double)CLOCK_FREQ * d_samples_delay /
424 (double)p_enc->fmt_out.audio.i_rate);
425 p_block->i_length = (mtime_t)((double)CLOCK_FREQ * (double)p_sys->i_frame_size /
426 (double)p_enc->fmt_out.audio.i_rate);
427 p_block->i_nb_samples = d_samples_delay;
428 //p_block->i_length = i_pts_out - p_sys->i_pts_last;
429 } else {
430 double d_samples_forward = (double)out_args.numInSamples/(double)p_enc->fmt_in.audio.i_channels;
431 double d_length = ((double)CLOCK_FREQ * d_samples_forward /
432 (double)p_enc->fmt_out.audio.i_rate);
433 i_pts_out += (mtime_t) d_length;
434 p_block->i_length = (mtime_t) d_length;
435 p_block->i_nb_samples = d_samples_forward;
436 }
437 }
438 p_block->i_dts = p_block->i_pts = i_pts_out;
439 block_ChainAppend(&p_chain, p_block);
440 #if 0
441 msg_Dbg(p_enc, "dts %"PRId64", length %"PRId64", " "pts_last "
442 "%"PRId64" numOutBytes = %i, numInSamples = %i, "
443 "i_samples %i, i_loop_count %i",
444 p_block->i_dts, p_block->i_length,
445 p_sys->i_pts_last, out_args.numOutBytes,
446 out_args.numInSamples, i_samples, i_loop_count);
447 #endif
448 if (likely(i_samples > 0)) {
449 p_sys->d_compression_ratio = (double)p_block->i_length / (double)out_args.numOutBytes;
450 i_samples_left -= out_args.numInSamples/p_enc->fmt_in.audio.i_channels;
451 p_sys->i_pts_last = i_pts_out;
452 }
453 } else {
454 block_Release(p_block);
455 //msg_Dbg(p_enc, "aac_encode_audio: not enough data yet");
456 break;
457 }
458 if (unlikely(i_loop_count++ > 100)) {
459 msg_Err(p_enc, "Loop count greater than 100!!!, something must be wrong with the encoder library");
460 break;
461 }
462 }
463
464 return p_chain;
465 }
466
467 /*****************************************************************************
468 * CloseDecoder: decoder destruction
469 *****************************************************************************/
470 static void CloseEncoder(vlc_object_t *p_this)
471 {
472 encoder_t *p_enc = (encoder_t *)p_this;
473 encoder_sys_t *p_sys = p_enc->p_sys;
474
475 aacEncClose(&p_sys->handle);
476 free(p_sys);
477 }