b19346e3d85e96c5a91b8d4cca70ae06d48c4da8
[libav.git] / libavcodec / amr.c
1 /*
2 * AMR Audio decoder stub
3 * Copyright (c) 2003 the ffmpeg project
4 *
5 * This file is part of FFmpeg.
6 *
7 * FFmpeg 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 * FFmpeg 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 FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21 /*
22 This code implements amr-nb and amr-wb audio encoder/decoder through external reference
23 code from www.3gpp.org. The licence of the code from 3gpp is unclear so you
24 have to download the code separately. Two versions exists: One fixed-point
25 and one with floats. For some reason the float-encoder is significant faster
26 atleast on a P4 1.5GHz (0.9s instead of 9.9s on a 30s audio clip at MR102).
27 Both float and fixed point is supported for amr-nb, but only float for
28 amr-wb.
29
30 --AMR-NB--
31 The fixed-point (TS26.073) can be downloaded from:
32 http://www.3gpp.org/ftp/Specs/archive/26_series/26.073/26073-510.zip
33 Extract the soure into ffmpeg/libavcodec/amr
34 To use the fixed version run "./configure" with "--enable-amr_nb-fixed"
35
36 The float version (default) can be downloaded from:
37 http://www.3gpp.org/ftp/Specs/archive/26_series/26.104/26104-510.zip
38 Extract the soure into ffmpeg/libavcodec/amr_float
39
40 The specification for amr-nb can be found in TS 26.071
41 (http://www.3gpp.org/ftp/Specs/html-info/26071.htm) and some other
42 info at http://www.3gpp.org/ftp/Specs/html-info/26-series.htm
43
44 --AMR-WB--
45 The reference code can be downloaded from:
46 http://www.3gpp.org/ftp/Specs/archive/26_series/26.204/26204-510.zip
47 It should be extracted to "libavcodec/amrwb_float". Enable it with
48 "--enable-amr_wb".
49
50 The specification for amr-wb can be downloaded from:
51 http://www.3gpp.org/ftp/Specs/archive/26_series/26.171/26171-500.zip
52
53 If someone want to use the fixed point version it can be downloaded
54 from: http://www.3gpp.org/ftp/Specs/archive/26_series/26.173/26173-571.zip
55
56 */
57
58 #include "avcodec.h"
59
60 #ifdef CONFIG_AMR_NB_FIXED
61
62 #define MMS_IO
63
64 #include "amr/sp_dec.h"
65 #include "amr/d_homing.h"
66 #include "amr/typedef.h"
67 #include "amr/sp_enc.h"
68 #include "amr/sid_sync.h"
69 #include "amr/e_homing.h"
70
71 #else
72 #include "amr_float/interf_dec.h"
73 #include "amr_float/interf_enc.h"
74 #endif
75
76 /* Common code for fixed and float version*/
77 typedef struct AMR_bitrates
78 {
79 int startrate;
80 int stoprate;
81 enum Mode mode;
82
83 } AMR_bitrates;
84
85 /* Match desired bitrate with closest one*/
86 static enum Mode getBitrateMode(int bitrate)
87 {
88 /* Adjusted so that all bitrates can be used from commandline where
89 only a multiple of 1000 can be specified*/
90 AMR_bitrates rates[]={ {0,4999,MR475}, //4
91 {5000,5899,MR515},//5
92 {5900,6699,MR59},//6
93 {6700,7000,MR67},//7
94 {7001,7949,MR74},//8
95 {7950,9999,MR795},//9
96 {10000,11999,MR102},//10
97 {12000,64000,MR122},//12
98
99 };
100 int i;
101 for(i=0;i<8;i++)
102 {
103 if(rates[i].startrate<=bitrate && rates[i].stoprate>=bitrate)
104 {
105 return(rates[i].mode);
106 }
107 }
108 /*Return highest possible*/
109 return(MR122);
110 }
111
112 static void amr_decode_fix_avctx(AVCodecContext * avctx)
113 {
114 const int is_amr_wb = 1 + (avctx->codec_id == CODEC_ID_AMR_WB);
115
116 if(avctx->sample_rate == 0)
117 {
118 avctx->sample_rate = 8000 * is_amr_wb;
119 }
120
121 if(avctx->channels == 0)
122 {
123 avctx->channels = 1;
124 }
125
126 avctx->frame_size = 160 * is_amr_wb;
127 }
128
129 #ifdef CONFIG_AMR_NB_FIXED
130 /* fixed point version*/
131 /* frame size in serial bitstream file (frame type + serial stream + flags) */
132 #define SERIAL_FRAMESIZE (1+MAX_SERIAL_SIZE+5)
133
134 typedef struct AMRContext {
135 int frameCount;
136 Speech_Decode_FrameState *speech_decoder_state;
137 enum RXFrameType rx_type;
138 enum Mode mode;
139 Word16 reset_flag;
140 Word16 reset_flag_old;
141
142 enum Mode enc_bitrate;
143 Speech_Encode_FrameState *enstate;
144 sid_syncState *sidstate;
145 enum TXFrameType tx_frametype;
146
147
148 } AMRContext;
149
150 static int amr_nb_decode_init(AVCodecContext * avctx)
151 {
152 AMRContext *s = avctx->priv_data;
153 s->frameCount=0;
154 s->speech_decoder_state=NULL;
155 s->rx_type = (enum RXFrameType)0;
156 s->mode= (enum Mode)0;
157 s->reset_flag=0;
158 s->reset_flag_old=1;
159
160 if(Speech_Decode_Frame_init(&s->speech_decoder_state, "Decoder"))
161 {
162 av_log(avctx, AV_LOG_ERROR, "Speech_Decode_Frame_init error\n");
163 return -1;
164 }
165
166 amr_decode_fix_avctx(avctx);
167
168 if(avctx->channels > 1)
169 {
170 av_log(avctx, AV_LOG_ERROR, "amr_nb: multichannel decoding not supported\n");
171 return -1;
172 }
173
174 return 0;
175 }
176
177 static int amr_nb_encode_init(AVCodecContext * avctx)
178 {
179 AMRContext *s = avctx->priv_data;
180 s->frameCount=0;
181 s->speech_decoder_state=NULL;
182 s->rx_type = (enum RXFrameType)0;
183 s->mode= (enum Mode)0;
184 s->reset_flag=0;
185 s->reset_flag_old=1;
186
187 if(avctx->sample_rate!=8000)
188 {
189 av_log(avctx, AV_LOG_ERROR, "Only 8000Hz sample rate supported\n");
190 return -1;
191 }
192
193 if(avctx->channels!=1)
194 {
195 av_log(avctx, AV_LOG_ERROR, "Only mono supported\n");
196 return -1;
197 }
198
199 avctx->frame_size=160;
200 avctx->coded_frame= avcodec_alloc_frame();
201
202 if(Speech_Encode_Frame_init(&s->enstate, 0, "encoder") || sid_sync_init (&s->sidstate))
203 {
204 av_log(avctx, AV_LOG_ERROR, "Speech_Encode_Frame_init error\n");
205 return -1;
206 }
207
208 s->enc_bitrate=getBitrateMode(avctx->bit_rate);
209
210 return 0;
211 }
212
213 static int amr_nb_encode_close(AVCodecContext * avctx)
214 {
215 AMRContext *s = avctx->priv_data;
216 Speech_Encode_Frame_exit(&s->enstate);
217 sid_sync_exit (&s->sidstate);
218 av_freep(&avctx->coded_frame);
219 return 0;
220 }
221
222 static int amr_nb_decode_close(AVCodecContext * avctx)
223 {
224 AMRContext *s = avctx->priv_data;
225 Speech_Decode_Frame_exit(&s->speech_decoder_state);
226 return 0;
227 }
228
229 static int amr_nb_decode_frame(AVCodecContext * avctx,
230 void *data, int *data_size,
231 uint8_t * buf, int buf_size)
232 {
233 AMRContext *s = avctx->priv_data;
234
235 uint8_t*amrData=buf;
236 int offset=0;
237
238 UWord8 toc, q, ft;
239
240 Word16 serial[SERIAL_FRAMESIZE]; /* coded bits */
241 Word16 *synth;
242 UWord8 *packed_bits;
243
244 static Word16 packed_size[16] = {12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0};
245 int i;
246
247 //printf("amr_decode_frame data_size=%i buf=0x%X buf_size=%d frameCount=%d!!\n",*data_size,buf,buf_size,s->frameCount);
248
249 synth=data;
250
251 // while(offset<buf_size)
252 {
253 toc=amrData[offset];
254 /* read rest of the frame based on ToC byte */
255 q = (toc >> 2) & 0x01;
256 ft = (toc >> 3) & 0x0F;
257
258 //printf("offset=%d, packet_size=%d amrData= 0x%X %X %X %X\n",offset,packed_size[ft],amrData[offset],amrData[offset+1],amrData[offset+2],amrData[offset+3]);
259
260 offset++;
261
262 packed_bits=amrData+offset;
263
264 offset+=packed_size[ft];
265
266 //Unsort and unpack bits
267 s->rx_type = UnpackBits(q, ft, packed_bits, &s->mode, &serial[1]);
268
269 //We have a new frame
270 s->frameCount++;
271
272 if (s->rx_type == RX_NO_DATA)
273 {
274 s->mode = s->speech_decoder_state->prev_mode;
275 }
276 else {
277 s->speech_decoder_state->prev_mode = s->mode;
278 }
279
280 /* if homed: check if this frame is another homing frame */
281 if (s->reset_flag_old == 1)
282 {
283 /* only check until end of first subframe */
284 s->reset_flag = decoder_homing_frame_test_first(&serial[1], s->mode);
285 }
286 /* produce encoder homing frame if homed & input=decoder homing frame */
287 if ((s->reset_flag != 0) && (s->reset_flag_old != 0))
288 {
289 for (i = 0; i < L_FRAME; i++)
290 {
291 synth[i] = EHF_MASK;
292 }
293 }
294 else
295 {
296 /* decode frame */
297 Speech_Decode_Frame(s->speech_decoder_state, s->mode, &serial[1], s->rx_type, synth);
298 }
299
300 //Each AMR-frame results in 160 16-bit samples
301 *data_size+=160*2;
302 synth+=160;
303
304 /* if not homed: check whether current frame is a homing frame */
305 if (s->reset_flag_old == 0)
306 {
307 /* check whole frame */
308 s->reset_flag = decoder_homing_frame_test(&serial[1], s->mode);
309 }
310 /* reset decoder if current frame is a homing frame */
311 if (s->reset_flag != 0)
312 {
313 Speech_Decode_Frame_reset(s->speech_decoder_state);
314 }
315 s->reset_flag_old = s->reset_flag;
316
317 }
318 return offset;
319 }
320
321
322 static int amr_nb_encode_frame(AVCodecContext *avctx,
323 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
324 {
325 short serial_data[250] = {0};
326
327 AMRContext *s = avctx->priv_data;
328 int written;
329
330 s->reset_flag = encoder_homing_frame_test(data);
331
332 Speech_Encode_Frame(s->enstate, s->enc_bitrate, data, &serial_data[1], &s->mode);
333
334 /* add frame type and mode */
335 sid_sync (s->sidstate, s->mode, &s->tx_frametype);
336
337 written = PackBits(s->mode, s->enc_bitrate, s->tx_frametype, &serial_data[1], frame);
338
339 if (s->reset_flag != 0)
340 {
341 Speech_Encode_Frame_reset(s->enstate);
342 sid_sync_reset(s->sidstate);
343 }
344 return written;
345 }
346
347
348 #elif defined(CONFIG_AMR_NB) /* Float point version*/
349
350 typedef struct AMRContext {
351 int frameCount;
352 void * decState;
353 int *enstate;
354 enum Mode enc_bitrate;
355 } AMRContext;
356
357 static int amr_nb_decode_init(AVCodecContext * avctx)
358 {
359 AMRContext *s = avctx->priv_data;
360 s->frameCount=0;
361 s->decState=Decoder_Interface_init();
362 if(!s->decState)
363 {
364 av_log(avctx, AV_LOG_ERROR, "Decoder_Interface_init error\r\n");
365 return -1;
366 }
367
368 amr_decode_fix_avctx(avctx);
369
370 if(avctx->channels > 1)
371 {
372 av_log(avctx, AV_LOG_ERROR, "amr_nb: multichannel decoding not supported\n");
373 return -1;
374 }
375
376 return 0;
377 }
378
379 static int amr_nb_encode_init(AVCodecContext * avctx)
380 {
381 AMRContext *s = avctx->priv_data;
382 s->frameCount=0;
383
384 if(avctx->sample_rate!=8000)
385 {
386 av_log(avctx, AV_LOG_ERROR, "Only 8000Hz sample rate supported\n");
387 return -1;
388 }
389
390 if(avctx->channels!=1)
391 {
392 av_log(avctx, AV_LOG_ERROR, "Only mono supported\n");
393 return -1;
394 }
395
396 avctx->frame_size=160;
397 avctx->coded_frame= avcodec_alloc_frame();
398
399 s->enstate=Encoder_Interface_init(0);
400 if(!s->enstate)
401 {
402 av_log(avctx, AV_LOG_ERROR, "Encoder_Interface_init error\n");
403 return -1;
404 }
405
406 s->enc_bitrate=getBitrateMode(avctx->bit_rate);
407
408 return 0;
409 }
410
411 static int amr_nb_decode_close(AVCodecContext * avctx)
412 {
413 AMRContext *s = avctx->priv_data;
414 Decoder_Interface_exit(s->decState);
415 return 0;
416 }
417
418 static int amr_nb_encode_close(AVCodecContext * avctx)
419 {
420 AMRContext *s = avctx->priv_data;
421 Encoder_Interface_exit(s->enstate);
422 av_freep(&avctx->coded_frame);
423 return 0;
424 }
425
426 static int amr_nb_decode_frame(AVCodecContext * avctx,
427 void *data, int *data_size,
428 uint8_t * buf, int buf_size)
429 {
430 AMRContext *s = (AMRContext*)avctx->priv_data;
431
432 uint8_t*amrData=buf;
433 static short block_size[16]={ 12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0 };
434 enum Mode dec_mode;
435 int packet_size;
436
437 /* av_log(NULL,AV_LOG_DEBUG,"amr_decode_frame buf=%p buf_size=%d frameCount=%d!!\n",buf,buf_size,s->frameCount); */
438
439 if(buf_size==0) {
440 /* nothing to do */
441 return 0;
442 }
443
444 dec_mode = (buf[0] >> 3) & 0x000F;
445 packet_size = block_size[dec_mode]+1;
446
447 if(packet_size > buf_size) {
448 av_log(avctx, AV_LOG_ERROR, "amr frame too short (%u, should be %u)\n", buf_size, packet_size);
449 return -1;
450 }
451
452 s->frameCount++;
453 /* av_log(NULL,AV_LOG_DEBUG,"packet_size=%d amrData= 0x%X %X %X %X\n",packet_size,amrData[0],amrData[1],amrData[2],amrData[3]); */
454 /* call decoder */
455 Decoder_Interface_Decode(s->decState, amrData, data, 0);
456 *data_size=160*2;
457
458 return packet_size;
459 }
460
461 static int amr_nb_encode_frame(AVCodecContext *avctx,
462 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
463 {
464 AMRContext *s = (AMRContext*)avctx->priv_data;
465 int written;
466
467 s->enc_bitrate=getBitrateMode(avctx->bit_rate);
468
469 written = Encoder_Interface_Encode(s->enstate,
470 s->enc_bitrate,
471 data,
472 frame,
473 0);
474 /* av_log(NULL,AV_LOG_DEBUG,"amr_nb_encode_frame encoded %u bytes, bitrate %u, first byte was %#02x\n",written, s->enc_bitrate, frame[0] ); */
475
476 return written;
477 }
478
479 #endif
480
481 #if defined(CONFIG_AMR_NB) || defined(CONFIG_AMR_NB_FIXED)
482
483 AVCodec amr_nb_decoder =
484 {
485 "amr_nb",
486 CODEC_TYPE_AUDIO,
487 CODEC_ID_AMR_NB,
488 sizeof(AMRContext),
489 amr_nb_decode_init,
490 NULL,
491 amr_nb_decode_close,
492 amr_nb_decode_frame,
493 };
494
495 AVCodec amr_nb_encoder =
496 {
497 "amr_nb",
498 CODEC_TYPE_AUDIO,
499 CODEC_ID_AMR_NB,
500 sizeof(AMRContext),
501 amr_nb_encode_init,
502 amr_nb_encode_frame,
503 amr_nb_encode_close,
504 NULL,
505 };
506
507 #endif
508
509 /* -----------AMR wideband ------------*/
510 #ifdef CONFIG_AMR_WB
511
512 #ifdef _TYPEDEF_H
513 //To avoid duplicate typedefs from typdef in amr-nb
514 #define typedef_h
515 #endif
516
517 #include "amrwb_float/enc_if.h"
518 #include "amrwb_float/dec_if.h"
519
520 /* Common code for fixed and float version*/
521 typedef struct AMRWB_bitrates
522 {
523 int startrate;
524 int stoprate;
525 int mode;
526
527 } AMRWB_bitrates;
528
529 static int getWBBitrateMode(int bitrate)
530 {
531 /* Adjusted so that all bitrates can be used from commandline where
532 only a multiple of 1000 can be specified*/
533 AMRWB_bitrates rates[]={ {0,7999,0}, //6.6kHz
534 {8000,9999,1},//8.85
535 {10000,13000,2},//12.65
536 {13001,14999,3},//14.25
537 {15000,17000,4},//15.85
538 {17001,18000,5},//18.25
539 {18001,22000,6},//19.85
540 {22001,23000,7},//23.05
541 {23001,24000,8},//23.85
542
543 };
544 int i;
545
546 for(i=0;i<9;i++)
547 {
548 if(rates[i].startrate<=bitrate && rates[i].stoprate>=bitrate)
549 {
550 return(rates[i].mode);
551 }
552 }
553 /*Return highest possible*/
554 return(8);
555 }
556
557
558 typedef struct AMRWBContext {
559 int frameCount;
560 void *state;
561 int mode;
562 Word16 allow_dtx;
563 } AMRWBContext;
564
565 static int amr_wb_encode_init(AVCodecContext * avctx)
566 {
567 AMRWBContext *s = (AMRWBContext*)avctx->priv_data;
568 s->frameCount=0;
569
570 if(avctx->sample_rate!=16000)
571 {
572 av_log(avctx, AV_LOG_ERROR, "Only 16000Hz sample rate supported\n");
573 return -1;
574 }
575
576 if(avctx->channels!=1)
577 {
578 av_log(avctx, AV_LOG_ERROR, "Only mono supported\n");
579 return -1;
580 }
581
582 avctx->frame_size=320;
583 avctx->coded_frame= avcodec_alloc_frame();
584
585 s->state = E_IF_init();
586 s->mode=getWBBitrateMode(avctx->bit_rate);
587 s->allow_dtx=0;
588
589 return 0;
590 }
591
592 static int amr_wb_encode_close(AVCodecContext * avctx)
593 {
594 AMRWBContext *s = (AMRWBContext*) avctx->priv_data;
595 E_IF_exit(s->state);
596 av_freep(&avctx->coded_frame);
597 s->frameCount++;
598 return 0;
599 }
600
601 static int amr_wb_encode_frame(AVCodecContext *avctx,
602 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
603 {
604 AMRWBContext *s;
605 int size;
606 s = (AMRWBContext*) avctx->priv_data;
607 s->mode=getWBBitrateMode(avctx->bit_rate);
608 size = E_IF_encode(s->state, s->mode, data, frame, s->allow_dtx);
609 return size;
610 }
611
612 static int amr_wb_decode_init(AVCodecContext * avctx)
613 {
614 AMRWBContext *s = (AMRWBContext *)avctx->priv_data;
615 s->frameCount=0;
616 s->state = D_IF_init();
617
618 amr_decode_fix_avctx(avctx);
619
620 if(avctx->channels > 1)
621 {
622 av_log(avctx, AV_LOG_ERROR, "amr_wb: multichannel decoding not supported\n");
623 return -1;
624 }
625
626 return 0;
627 }
628
629 extern const UWord8 block_size[];
630
631 static int amr_wb_decode_frame(AVCodecContext * avctx,
632 void *data, int *data_size,
633 uint8_t * buf, int buf_size)
634 {
635 AMRWBContext *s = (AMRWBContext*)avctx->priv_data;
636
637 uint8_t*amrData=buf;
638 int mode;
639 int packet_size;
640
641 if(buf_size==0) {
642 /* nothing to do */
643 return 0;
644 }
645
646 mode = (amrData[0] >> 3) & 0x000F;
647 packet_size = block_size[mode];
648
649 if(packet_size > buf_size) {
650 av_log(avctx, AV_LOG_ERROR, "amr frame too short (%u, should be %u)\n", buf_size, packet_size+1);
651 return -1;
652 }
653
654 s->frameCount++;
655 D_IF_decode( s->state, amrData, data, _good_frame);
656 *data_size=320*2;
657 return packet_size;
658 }
659
660 static int amr_wb_decode_close(AVCodecContext * avctx)
661 {
662 AMRWBContext *s = (AMRWBContext *)avctx->priv_data;
663 D_IF_exit(s->state);
664 return 0;
665 }
666
667 AVCodec amr_wb_decoder =
668 {
669 "amr_wb",
670 CODEC_TYPE_AUDIO,
671 CODEC_ID_AMR_WB,
672 sizeof(AMRWBContext),
673 amr_wb_decode_init,
674 NULL,
675 amr_wb_decode_close,
676 amr_wb_decode_frame,
677 };
678
679 AVCodec amr_wb_encoder =
680 {
681 "amr_wb",
682 CODEC_TYPE_AUDIO,
683 CODEC_ID_AMR_WB,
684 sizeof(AMRWBContext),
685 amr_wb_encode_init,
686 amr_wb_encode_frame,
687 amr_wb_encode_close,
688 NULL,
689 };
690
691 #endif //CONFIG_AMR_WB