2 * AMR Audio decoder stub
3 * Copyright (c) 2003 the ffmpeg project
5 * This file is part of FFmpeg.
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.
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.
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
23 * Adaptive Multi-Rate (AMR) Audio decoder stub.
25 * This code implements both an AMR-NarrowBand (AMR-NB) and an AMR-WideBand
26 * (AMR-WB) audio encoder/decoder through external reference code from
27 * http://www.3gpp.org/. The license of the code from 3gpp is unclear so you
28 * have to download the code separately. Two versions exists: One fixed-point
29 * and one with floats. For some reason the float-encoder is significant faster
30 * at least on a P4 1.5GHz (0.9s instead of 9.9s on a 30s audio clip at MR102).
31 * Both float and fixed point are supported for AMR-NB, but only float for
37 * The float version (default) can be downloaded from:
38 * http://www.3gpp.org/ftp/Specs/archive/26_series/26.104/26104-510.zip
39 * Extract the source into \c "ffmpeg/libavcodec/amr_float".
41 * \subsection Fixed-point
42 * The fixed-point (TS26.073) can be downloaded from:
43 * http://www.3gpp.org/ftp/Specs/archive/26_series/26.073/26073-510.zip.
44 * Extract the source into \c "ffmpeg/libavcodec/amr".
45 * To use the fixed version run \c "./configure" with \c "--enable-amr_nb-fixed".
47 * \subsection Specification
48 * The specification for AMR-NB can be found in TS 26.071
49 * (http://www.3gpp.org/ftp/Specs/html-info/26071.htm) and some other
50 * info at http://www.3gpp.org/ftp/Specs/html-info/26-series.htm.
54 * The reference code can be downloaded from:
55 * http://www.3gpp.org/ftp/Specs/archive/26_series/26.204/26204-510.zip
56 * It should be extracted to \c "ffmpeg/libavcodec/amrwb_float". Enable it with
57 * \c "--enable-amr_wb".
59 * \subsection Fixed-point
60 * If someone wants to use the fixed point version it can be downloaded from:
61 * http://www.3gpp.org/ftp/Specs/archive/26_series/26.173/26173-571.zip.
63 * \subsection Specification
64 * The specification for AMR-WB can be downloaded from:
65 * http://www.3gpp.org/ftp/Specs/archive/26_series/26.171/26171-500.zip.
71 #ifdef CONFIG_AMR_NB_FIXED
75 #include "amr/sp_dec.h"
76 #include "amr/d_homing.h"
77 #include "amr/typedef.h"
78 #include "amr/sp_enc.h"
79 #include "amr/sid_sync.h"
80 #include "amr/e_homing.h"
83 #include "amr_float/interf_dec.h"
84 #include "amr_float/interf_enc.h"
87 /* Common code for fixed and float version*/
88 typedef struct AMR_bitrates
96 /* Match desired bitrate with closest one*/
97 static enum Mode getBitrateMode(int bitrate)
99 /* Adjusted so that all bitrates can be used from commandline where
100 only a multiple of 1000 can be specified*/
101 AMR_bitrates rates[]={ {0,4999,MR475}, //4
102 {5000,5899,MR515},//5
106 {7950,9999,MR795},//9
107 {10000,11999,MR102},//10
108 {12000,64000,MR122},//12
114 if(rates[i].startrate<=bitrate && rates[i].stoprate>=bitrate)
116 return(rates[i].mode);
119 /*Return highest possible*/
123 static void amr_decode_fix_avctx(AVCodecContext * avctx)
125 const int is_amr_wb = 1 + (avctx->codec_id == CODEC_ID_AMR_WB);
127 if(avctx->sample_rate == 0)
129 avctx->sample_rate = 8000 * is_amr_wb;
132 if(avctx->channels == 0)
137 avctx->frame_size = 160 * is_amr_wb;
140 #ifdef CONFIG_AMR_NB_FIXED
141 /* fixed point version*/
142 /* frame size in serial bitstream file (frame type + serial stream + flags) */
143 #define SERIAL_FRAMESIZE (1+MAX_SERIAL_SIZE+5)
145 typedef struct AMRContext {
147 Speech_Decode_FrameState *speech_decoder_state;
148 enum RXFrameType rx_type;
151 Word16 reset_flag_old;
153 enum Mode enc_bitrate;
154 Speech_Encode_FrameState *enstate;
155 sid_syncState *sidstate;
156 enum TXFrameType tx_frametype;
161 static int amr_nb_decode_init(AVCodecContext * avctx)
163 AMRContext *s = avctx->priv_data;
165 s->speech_decoder_state=NULL;
166 s->rx_type = (enum RXFrameType)0;
167 s->mode= (enum Mode)0;
171 if(Speech_Decode_Frame_init(&s->speech_decoder_state, "Decoder"))
173 av_log(avctx, AV_LOG_ERROR, "Speech_Decode_Frame_init error\n");
177 amr_decode_fix_avctx(avctx);
179 if(avctx->channels > 1)
181 av_log(avctx, AV_LOG_ERROR, "amr_nb: multichannel decoding not supported\n");
188 static int amr_nb_encode_init(AVCodecContext * avctx)
190 AMRContext *s = avctx->priv_data;
192 s->speech_decoder_state=NULL;
193 s->rx_type = (enum RXFrameType)0;
194 s->mode= (enum Mode)0;
198 if(avctx->sample_rate!=8000)
200 av_log(avctx, AV_LOG_ERROR, "Only 8000Hz sample rate supported\n");
204 if(avctx->channels!=1)
206 av_log(avctx, AV_LOG_ERROR, "Only mono supported\n");
210 avctx->frame_size=160;
211 avctx->coded_frame= avcodec_alloc_frame();
213 if(Speech_Encode_Frame_init(&s->enstate, 0, "encoder") || sid_sync_init (&s->sidstate))
215 av_log(avctx, AV_LOG_ERROR, "Speech_Encode_Frame_init error\n");
219 s->enc_bitrate=getBitrateMode(avctx->bit_rate);
224 static int amr_nb_encode_close(AVCodecContext * avctx)
226 AMRContext *s = avctx->priv_data;
227 Speech_Encode_Frame_exit(&s->enstate);
228 sid_sync_exit (&s->sidstate);
229 av_freep(&avctx->coded_frame);
233 static int amr_nb_decode_close(AVCodecContext * avctx)
235 AMRContext *s = avctx->priv_data;
236 Speech_Decode_Frame_exit(&s->speech_decoder_state);
240 static int amr_nb_decode_frame(AVCodecContext * avctx,
241 void *data, int *data_size,
242 uint8_t * buf, int buf_size)
244 AMRContext *s = avctx->priv_data;
251 Word16 serial[SERIAL_FRAMESIZE]; /* coded bits */
255 static Word16 packed_size[16] = {12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0};
258 //printf("amr_decode_frame data_size=%i buf=0x%X buf_size=%d frameCount=%d!!\n",*data_size,buf,buf_size,s->frameCount);
262 // while(offset<buf_size)
265 /* read rest of the frame based on ToC byte */
266 q = (toc >> 2) & 0x01;
267 ft = (toc >> 3) & 0x0F;
269 //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]);
273 packed_bits=amrData+offset;
275 offset+=packed_size[ft];
277 //Unsort and unpack bits
278 s->rx_type = UnpackBits(q, ft, packed_bits, &s->mode, &serial[1]);
280 //We have a new frame
283 if (s->rx_type == RX_NO_DATA)
285 s->mode = s->speech_decoder_state->prev_mode;
288 s->speech_decoder_state->prev_mode = s->mode;
291 /* if homed: check if this frame is another homing frame */
292 if (s->reset_flag_old == 1)
294 /* only check until end of first subframe */
295 s->reset_flag = decoder_homing_frame_test_first(&serial[1], s->mode);
297 /* produce encoder homing frame if homed & input=decoder homing frame */
298 if ((s->reset_flag != 0) && (s->reset_flag_old != 0))
300 for (i = 0; i < L_FRAME; i++)
308 Speech_Decode_Frame(s->speech_decoder_state, s->mode, &serial[1], s->rx_type, synth);
311 //Each AMR-frame results in 160 16-bit samples
315 /* if not homed: check whether current frame is a homing frame */
316 if (s->reset_flag_old == 0)
318 /* check whole frame */
319 s->reset_flag = decoder_homing_frame_test(&serial[1], s->mode);
321 /* reset decoder if current frame is a homing frame */
322 if (s->reset_flag != 0)
324 Speech_Decode_Frame_reset(s->speech_decoder_state);
326 s->reset_flag_old = s->reset_flag;
333 static int amr_nb_encode_frame(AVCodecContext *avctx,
334 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
336 short serial_data[250] = {0};
338 AMRContext *s = avctx->priv_data;
341 s->reset_flag = encoder_homing_frame_test(data);
343 Speech_Encode_Frame(s->enstate, s->enc_bitrate, data, &serial_data[1], &s->mode);
345 /* add frame type and mode */
346 sid_sync (s->sidstate, s->mode, &s->tx_frametype);
348 written = PackBits(s->mode, s->enc_bitrate, s->tx_frametype, &serial_data[1], frame);
350 if (s->reset_flag != 0)
352 Speech_Encode_Frame_reset(s->enstate);
353 sid_sync_reset(s->sidstate);
359 #elif defined(CONFIG_AMR_NB) /* Float point version*/
361 typedef struct AMRContext {
365 enum Mode enc_bitrate;
368 static int amr_nb_decode_init(AVCodecContext * avctx)
370 AMRContext *s = avctx->priv_data;
372 s->decState=Decoder_Interface_init();
375 av_log(avctx, AV_LOG_ERROR, "Decoder_Interface_init error\r\n");
379 amr_decode_fix_avctx(avctx);
381 if(avctx->channels > 1)
383 av_log(avctx, AV_LOG_ERROR, "amr_nb: multichannel decoding not supported\n");
390 static int amr_nb_encode_init(AVCodecContext * avctx)
392 AMRContext *s = avctx->priv_data;
395 if(avctx->sample_rate!=8000)
397 av_log(avctx, AV_LOG_ERROR, "Only 8000Hz sample rate supported\n");
401 if(avctx->channels!=1)
403 av_log(avctx, AV_LOG_ERROR, "Only mono supported\n");
407 avctx->frame_size=160;
408 avctx->coded_frame= avcodec_alloc_frame();
410 s->enstate=Encoder_Interface_init(0);
413 av_log(avctx, AV_LOG_ERROR, "Encoder_Interface_init error\n");
417 s->enc_bitrate=getBitrateMode(avctx->bit_rate);
422 static int amr_nb_decode_close(AVCodecContext * avctx)
424 AMRContext *s = avctx->priv_data;
425 Decoder_Interface_exit(s->decState);
429 static int amr_nb_encode_close(AVCodecContext * avctx)
431 AMRContext *s = avctx->priv_data;
432 Encoder_Interface_exit(s->enstate);
433 av_freep(&avctx->coded_frame);
437 static int amr_nb_decode_frame(AVCodecContext * avctx,
438 void *data, int *data_size,
439 uint8_t * buf, int buf_size)
441 AMRContext *s = (AMRContext*)avctx->priv_data;
444 static short block_size[16]={ 12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0 };
448 /* av_log(NULL,AV_LOG_DEBUG,"amr_decode_frame buf=%p buf_size=%d frameCount=%d!!\n",buf,buf_size,s->frameCount); */
450 dec_mode = (buf[0] >> 3) & 0x000F;
451 packet_size = block_size[dec_mode]+1;
453 if(packet_size > buf_size) {
454 av_log(avctx, AV_LOG_ERROR, "amr frame too short (%u, should be %u)\n", buf_size, packet_size);
459 /* 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]); */
461 Decoder_Interface_Decode(s->decState, amrData, data, 0);
467 static int amr_nb_encode_frame(AVCodecContext *avctx,
468 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
470 AMRContext *s = (AMRContext*)avctx->priv_data;
473 s->enc_bitrate=getBitrateMode(avctx->bit_rate);
475 written = Encoder_Interface_Encode(s->enstate,
480 /* 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] ); */
487 #if defined(CONFIG_AMR_NB) || defined(CONFIG_AMR_NB_FIXED)
489 AVCodec amr_nb_decoder =
501 AVCodec amr_nb_encoder =
515 /* -----------AMR wideband ------------*/
519 //To avoid duplicate typedefs from typdef in amr-nb
523 #include "amrwb_float/enc_if.h"
524 #include "amrwb_float/dec_if.h"
526 /* Common code for fixed and float version*/
527 typedef struct AMRWB_bitrates
535 static int getWBBitrateMode(int bitrate)
537 /* Adjusted so that all bitrates can be used from commandline where
538 only a multiple of 1000 can be specified*/
539 AMRWB_bitrates rates[]={ {0,7999,0}, //6.6kHz
541 {10000,13000,2},//12.65
542 {13001,14999,3},//14.25
543 {15000,17000,4},//15.85
544 {17001,18000,5},//18.25
545 {18001,22000,6},//19.85
546 {22001,23000,7},//23.05
547 {23001,24000,8},//23.85
554 if(rates[i].startrate<=bitrate && rates[i].stoprate>=bitrate)
556 return(rates[i].mode);
559 /*Return highest possible*/
564 typedef struct AMRWBContext {
571 static int amr_wb_encode_init(AVCodecContext * avctx)
573 AMRWBContext *s = (AMRWBContext*)avctx->priv_data;
576 if(avctx->sample_rate!=16000)
578 av_log(avctx, AV_LOG_ERROR, "Only 16000Hz sample rate supported\n");
582 if(avctx->channels!=1)
584 av_log(avctx, AV_LOG_ERROR, "Only mono supported\n");
588 avctx->frame_size=320;
589 avctx->coded_frame= avcodec_alloc_frame();
591 s->state = E_IF_init();
592 s->mode=getWBBitrateMode(avctx->bit_rate);
598 static int amr_wb_encode_close(AVCodecContext * avctx)
600 AMRWBContext *s = (AMRWBContext*) avctx->priv_data;
602 av_freep(&avctx->coded_frame);
607 static int amr_wb_encode_frame(AVCodecContext *avctx,
608 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
612 s = (AMRWBContext*) avctx->priv_data;
613 s->mode=getWBBitrateMode(avctx->bit_rate);
614 size = E_IF_encode(s->state, s->mode, data, frame, s->allow_dtx);
618 static int amr_wb_decode_init(AVCodecContext * avctx)
620 AMRWBContext *s = (AMRWBContext *)avctx->priv_data;
622 s->state = D_IF_init();
624 amr_decode_fix_avctx(avctx);
626 if(avctx->channels > 1)
628 av_log(avctx, AV_LOG_ERROR, "amr_wb: multichannel decoding not supported\n");
635 extern const UWord8 block_size[];
637 static int amr_wb_decode_frame(AVCodecContext * avctx,
638 void *data, int *data_size,
639 uint8_t * buf, int buf_size)
641 AMRWBContext *s = (AMRWBContext*)avctx->priv_data;
652 mode = (amrData[0] >> 3) & 0x000F;
653 packet_size = block_size[mode];
655 if(packet_size > buf_size) {
656 av_log(avctx, AV_LOG_ERROR, "amr frame too short (%u, should be %u)\n", buf_size, packet_size+1);
661 D_IF_decode( s->state, amrData, data, _good_frame);
666 static int amr_wb_decode_close(AVCodecContext * avctx)
668 AMRWBContext *s = (AMRWBContext *)avctx->priv_data;
673 AVCodec amr_wb_decoder =
678 sizeof(AMRWBContext),
685 AVCodec amr_wb_encoder =
690 sizeof(AMRWBContext),
697 #endif //CONFIG_AMR_WB