2 * AMR Audio decoder stub
3 * Copyright (c) 2003 the ffmpeg project
5 * This library is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU Lesser General Public
7 * License as published by the Free Software Foundation; either
8 * version 2 of the License, or (at your option) any later version.
10 * This library is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * Lesser General Public License for more details.
15 * You should have received a copy of the GNU Lesser General Public
16 * License along with this library; if not, write to the Free Software
17 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20 This code implements amr-nb and amr-wb audio encoder/decoder through external reference
21 code from www.3gpp.org. The licence of the code from 3gpp is unclear so you
22 have to download the code separately. Two versions exists: One fixed-point
23 and one with floats. For some reason the float-encoder is significant faster
24 atleast on a P4 1.5GHz (0.9s instead of 9.9s on a 30s audio clip at MR102).
25 Both float and fixed point is supported for amr-nb, but only float for
29 The fixed-point (TS26.073) can be downloaded from:
30 http://www.3gpp.org/ftp/Specs/archive/26_series/26.073/26073-510.zip
31 Extract the soure into ffmpeg/libavcodec/amr
32 To use the fixed version run "./configure" with "--enable-amr_nb-fixed"
34 The float version (default) can be downloaded from:
35 http://www.3gpp.org/ftp/Specs/archive/26_series/26.104/26104-510.zip
36 Extract the soure into ffmpeg/libavcodec/amr_float
38 The specification for amr-nb can be found in TS 26.071
39 (http://www.3gpp.org/ftp/Specs/html-info/26071.htm) and some other
40 info at http://www.3gpp.org/ftp/Specs/html-info/26-series.htm
43 The reference code can be downloaded from:
44 http://www.3gpp.org/ftp/Specs/archive/26_series/26.204/26204-510.zip
45 It should be extracted to "libavcodec/amrwb_float". Enable it with
48 The specification for amr-wb can be downloaded from:
49 http://www.3gpp.org/ftp/Specs/archive/26_series/26.171/26171-500.zip
51 If someone want to use the fixed point version it can be downloaded
52 from: http://www.3gpp.org/ftp/Specs/archive/26_series/26.173/26173-571.zip
56 #include "../config.h"
63 #include "amr/sp_dec.h"
64 #include "amr/d_homing.h"
65 #include "amr/typedef.h"
66 #include "amr/sp_enc.h"
67 #include "amr/sid_sync.h"
68 #include "amr/e_homing.h"
71 #include "amr_float/interf_dec.h"
72 #include "amr_float/interf_enc.h"
75 /* Common code for fixed and float version*/
76 typedef struct AMR_bitrates
84 /* Match desired bitrate with closest one*/
85 static enum Mode getBitrateMode(int bitrate)
87 /* Adjusted so that all bitrates can be used from commandline where
88 only a multiple of 1000 can be specified*/
89 AMR_bitrates rates[]={ {0,4999,MR475}, //4
95 {10000,11999,MR102},//10
96 {12000,64000,MR122},//12
102 if(rates[i].startrate<=bitrate && rates[i].stoprate>=bitrate)
104 return(rates[i].mode);
107 /*Return highest possible*/
112 /* fixed point version*/
113 /* frame size in serial bitstream file (frame type + serial stream + flags) */
114 #define SERIAL_FRAMESIZE (1+MAX_SERIAL_SIZE+5)
116 typedef struct AMRContext {
118 Speech_Decode_FrameState *speech_decoder_state;
119 enum RXFrameType rx_type;
122 Word16 reset_flag_old;
124 enum Mode enc_bitrate;
125 Speech_Encode_FrameState *enstate;
126 sid_syncState *sidstate;
127 enum TXFrameType tx_frametype;
132 static int amr_nb_decode_init(AVCodecContext * avctx)
134 AMRContext *s = avctx->priv_data;
136 s->speech_decoder_state=NULL;
137 s->rx_type = (enum RXFrameType)0;
138 s->mode= (enum Mode)0;
142 if(Speech_Decode_Frame_init(&s->speech_decoder_state, "Decoder"))
144 av_log(avctx, AV_LOG_ERROR, "Speech_Decode_Frame_init error\n");
150 static int amr_nb_encode_init(AVCodecContext * avctx)
152 AMRContext *s = avctx->priv_data;
154 s->speech_decoder_state=NULL;
155 s->rx_type = (enum RXFrameType)0;
156 s->mode= (enum Mode)0;
160 if(avctx->sample_rate!=8000)
164 av_log(avctx, AV_LOG_DEBUG, "Only 8000Hz sample rate supported\n");
169 if(avctx->channels!=1)
173 av_log(avctx, AV_LOG_DEBUG, "Only mono supported\n");
178 avctx->frame_size=160;
179 avctx->coded_frame= avcodec_alloc_frame();
181 if(Speech_Encode_Frame_init(&s->enstate, 0, "encoder") || sid_sync_init (&s->sidstate))
185 av_log(avctx, AV_LOG_DEBUG, "Speech_Encode_Frame_init error\n");
190 s->enc_bitrate=getBitrateMode(avctx->bit_rate);
195 static int amr_nb_encode_close(AVCodecContext * avctx)
197 AMRContext *s = avctx->priv_data;
198 Speech_Encode_Frame_exit(&s->enstate);
199 sid_sync_exit (&s->sidstate);
200 av_freep(&avctx->coded_frame);
204 static int amr_nb_decode_close(AVCodecContext * avctx)
206 AMRContext *s = avctx->priv_data;
207 Speech_Decode_Frame_exit(&s->speech_decoder_state);
211 static int amr_nb_decode_frame(AVCodecContext * avctx,
212 void *data, int *data_size,
213 uint8_t * buf, int buf_size)
215 AMRContext *s = avctx->priv_data;
222 Word16 serial[SERIAL_FRAMESIZE]; /* coded bits */
227 static Word16 packed_size[16] = {12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0};
230 //printf("amr_decode_frame data_size=%i buf=0x%X buf_size=%d frameCount=%d!!\n",*data_size,buf,buf_size,s->frameCount);
234 // while(offset<buf_size)
237 /* read rest of the frame based on ToC byte */
238 q = (toc >> 2) & 0x01;
239 ft = (toc >> 3) & 0x0F;
241 //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]);
245 packed_bits=amrData+offset;
247 offset+=packed_size[ft];
249 //Unsort and unpack bits
250 s->rx_type = UnpackBits(q, ft, packed_bits, &s->mode, &serial[1]);
252 //We have a new frame
255 if (s->rx_type == RX_NO_DATA)
257 s->mode = s->speech_decoder_state->prev_mode;
260 s->speech_decoder_state->prev_mode = s->mode;
263 /* if homed: check if this frame is another homing frame */
264 if (s->reset_flag_old == 1)
266 /* only check until end of first subframe */
267 s->reset_flag = decoder_homing_frame_test_first(&serial[1], s->mode);
269 /* produce encoder homing frame if homed & input=decoder homing frame */
270 if ((s->reset_flag != 0) && (s->reset_flag_old != 0))
272 for (i = 0; i < L_FRAME; i++)
280 Speech_Decode_Frame(s->speech_decoder_state, s->mode, &serial[1], s->rx_type, synth);
283 //Each AMR-frame results in 160 16-bit samples
287 /* if not homed: check whether current frame is a homing frame */
288 if (s->reset_flag_old == 0)
290 /* check whole frame */
291 s->reset_flag = decoder_homing_frame_test(&serial[1], s->mode);
293 /* reset decoder if current frame is a homing frame */
294 if (s->reset_flag != 0)
296 Speech_Decode_Frame_reset(s->speech_decoder_state);
298 s->reset_flag_old = s->reset_flag;
305 static int amr_nb_encode_frame(AVCodecContext *avctx,
306 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
308 short serial_data[250] = {0};
310 AMRContext *s = avctx->priv_data;
313 s->reset_flag = encoder_homing_frame_test(data);
315 Speech_Encode_Frame(s->enstate, s->enc_bitrate, data, &serial_data[1], &s->mode);
317 /* add frame type and mode */
318 sid_sync (s->sidstate, s->mode, &s->tx_frametype);
320 written = PackBits(s->mode, s->enc_bitrate, s->tx_frametype, &serial_data[1], frame);
322 if (s->reset_flag != 0)
324 Speech_Encode_Frame_reset(s->enstate);
325 sid_sync_reset(s->sidstate);
331 #else /* Float point version*/
333 typedef struct AMRContext {
337 enum Mode enc_bitrate;
340 static int amr_nb_decode_init(AVCodecContext * avctx)
342 AMRContext *s = avctx->priv_data;
344 s->decState=Decoder_Interface_init();
347 av_log(avctx, AV_LOG_ERROR, "Decoder_Interface_init error\r\n");
353 static int amr_nb_encode_init(AVCodecContext * avctx)
355 AMRContext *s = avctx->priv_data;
358 if(avctx->sample_rate!=8000)
362 av_log(avctx, AV_LOG_DEBUG, "Only 8000Hz sample rate supported\n");
367 if(avctx->channels!=1)
371 av_log(avctx, AV_LOG_DEBUG, "Only mono supported\n");
376 avctx->frame_size=160;
377 avctx->coded_frame= avcodec_alloc_frame();
379 s->enstate=Encoder_Interface_init(0);
384 av_log(avctx, AV_LOG_DEBUG, "Encoder_Interface_init error\n");
389 s->enc_bitrate=getBitrateMode(avctx->bit_rate);
394 static int amr_nb_decode_close(AVCodecContext * avctx)
396 AMRContext *s = avctx->priv_data;
397 Decoder_Interface_exit(s->decState);
401 static int amr_nb_encode_close(AVCodecContext * avctx)
403 AMRContext *s = avctx->priv_data;
404 Encoder_Interface_exit(s->enstate);
405 av_freep(&avctx->coded_frame);
409 static int amr_nb_decode_frame(AVCodecContext * avctx,
410 void *data, int *data_size,
411 uint8_t * buf, int buf_size)
413 AMRContext *s = (AMRContext*)avctx->priv_data;
417 static short block_size[16]={ 12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0 };
422 //printf("amr_decode_frame data_size=%i buf=0x%X buf_size=%d frameCount=%d!!\n",*data_size,buf,buf_size,s->frameCount);
424 while(offset<buf_size)
426 dec_mode = (amrData[offset] >> 3) & 0x000F;
427 packet_size = block_size[dec_mode];
430 //printf("offset=%d, packet_size=%d amrData= 0x%X %X %X %X\n",offset,packet_size,amrData[offset],amrData[offset+1],amrData[offset+2],amrData[offset+3]);
432 Decoder_Interface_Decode(s->decState, &amrData[offset], data+*data_size, 0);
435 offset+=packet_size+1;
440 static int amr_nb_encode_frame(AVCodecContext *avctx,
441 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
443 AMRContext *s = (AMRContext*)avctx->priv_data;
446 written = Encoder_Interface_Encode(s->enstate,
457 AVCodec amr_nb_decoder =
469 AVCodec amr_nb_encoder =
481 /* -----------AMR wideband ------------*/
485 //To avoid duplicate typedefs from typdef in amr-nb
489 #include "amrwb_float/enc_if.h"
490 #include "amrwb_float/dec_if.h"
492 /* Common code for fixed and float version*/
493 typedef struct AMRWB_bitrates
501 static int getWBBitrateMode(int bitrate)
503 /* Adjusted so that all bitrates can be used from commandline where
504 only a multiple of 1000 can be specified*/
505 AMRWB_bitrates rates[]={ {0,7999,0}, //6.6kHz
507 {10000,13000,2},//12.65
508 {13001,14999,3},//14.25
509 {15000,17000,4},//15.85
510 {17001,18000,5},//18.25
511 {18001,22000,6},//19.85
512 {22001,23000,7},//23.05
513 {23001,24000,8},//23.85
520 if(rates[i].startrate<=bitrate && rates[i].stoprate>=bitrate)
522 return(rates[i].mode);
525 /*Return highest possible*/
530 typedef struct AMRWBContext {
537 static int amr_wb_encode_init(AVCodecContext * avctx)
539 AMRWBContext *s = (AMRWBContext*)avctx->priv_data;
542 if(avctx->sample_rate!=16000)
546 av_log(avctx, AV_LOG_DEBUG, "Only 16000Hz sample rate supported\n");
551 if(avctx->channels!=1)
555 av_log(avctx, AV_LOG_DEBUG, "Only mono supported\n");
560 avctx->frame_size=320;
561 avctx->coded_frame= avcodec_alloc_frame();
563 s->state = E_IF_init();
564 s->mode=getWBBitrateMode(avctx->bit_rate);
570 static int amr_wb_encode_close(AVCodecContext * avctx)
572 AMRWBContext *s = (AMRWBContext*) avctx->priv_data;
574 av_freep(&avctx->coded_frame);
579 static int amr_wb_encode_frame(AVCodecContext *avctx,
580 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
582 AMRWBContext *s = (AMRWBContext*) avctx->priv_data;
583 int size = E_IF_encode(s->state, s->mode, data, frame, s->allow_dtx);
587 static int amr_wb_decode_init(AVCodecContext * avctx)
589 AMRWBContext *s = (AMRWBContext *)avctx->priv_data;
591 s->state = D_IF_init();
595 extern const UWord8 block_size[];
597 static int amr_wb_decode_frame(AVCodecContext * avctx,
598 void *data, int *data_size,
599 uint8_t * buf, int buf_size)
601 AMRWBContext *s = (AMRWBContext*)avctx->priv_data;
609 while(offset<buf_size)
612 mode = (Word16)((amrData[offset] >> 3) & 0x0F);
613 packet_size = block_size[mode];
614 D_IF_decode( s->state, &amrData[offset], data+*data_size, _good_frame);
621 static int amr_wb_decode_close(AVCodecContext * avctx)
623 AMRWBContext *s = (AMRWBContext *)avctx->priv_data;
628 AVCodec amr_wb_decoder =
633 sizeof(AMRWBContext),
640 AVCodec amr_wb_encoder =
645 sizeof(AMRWBContext),