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
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
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"
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
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
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
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
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
60 #ifdef CONFIG_AMR_NB_FIXED
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"
72 #include "amr_float/interf_dec.h"
73 #include "amr_float/interf_enc.h"
76 /* Common code for fixed and float version*/
77 typedef struct AMR_bitrates
85 /* Match desired bitrate with closest one*/
86 static enum Mode getBitrateMode(int bitrate)
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
96 {10000,11999,MR102},//10
97 {12000,64000,MR122},//12
103 if(rates[i].startrate<=bitrate && rates[i].stoprate>=bitrate)
105 return(rates[i].mode);
108 /*Return highest possible*/
112 static void amr_decode_fix_avctx(AVCodecContext * avctx)
114 const int is_amr_wb = 1 + (avctx->codec_id == CODEC_ID_AMR_WB);
116 if(avctx->sample_rate == 0)
118 avctx->sample_rate = 8000 * is_amr_wb;
121 if(avctx->channels == 0)
126 avctx->frame_size = 160 * is_amr_wb;
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)
134 typedef struct AMRContext {
136 Speech_Decode_FrameState *speech_decoder_state;
137 enum RXFrameType rx_type;
140 Word16 reset_flag_old;
142 enum Mode enc_bitrate;
143 Speech_Encode_FrameState *enstate;
144 sid_syncState *sidstate;
145 enum TXFrameType tx_frametype;
150 static int amr_nb_decode_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(Speech_Decode_Frame_init(&s->speech_decoder_state, "Decoder"))
162 av_log(avctx, AV_LOG_ERROR, "Speech_Decode_Frame_init error\n");
166 amr_decode_fix_avctx(avctx);
168 if(avctx->channels > 1)
170 av_log(avctx, AV_LOG_ERROR, "amr_nb: multichannel decoding not supported\n");
177 static int amr_nb_encode_init(AVCodecContext * avctx)
179 AMRContext *s = avctx->priv_data;
181 s->speech_decoder_state=NULL;
182 s->rx_type = (enum RXFrameType)0;
183 s->mode= (enum Mode)0;
187 if(avctx->sample_rate!=8000)
189 av_log(avctx, AV_LOG_ERROR, "Only 8000Hz sample rate supported\n");
193 if(avctx->channels!=1)
195 av_log(avctx, AV_LOG_ERROR, "Only mono supported\n");
199 avctx->frame_size=160;
200 avctx->coded_frame= avcodec_alloc_frame();
202 if(Speech_Encode_Frame_init(&s->enstate, 0, "encoder") || sid_sync_init (&s->sidstate))
204 av_log(avctx, AV_LOG_ERROR, "Speech_Encode_Frame_init error\n");
208 s->enc_bitrate=getBitrateMode(avctx->bit_rate);
213 static int amr_nb_encode_close(AVCodecContext * avctx)
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);
222 static int amr_nb_decode_close(AVCodecContext * avctx)
224 AMRContext *s = avctx->priv_data;
225 Speech_Decode_Frame_exit(&s->speech_decoder_state);
229 static int amr_nb_decode_frame(AVCodecContext * avctx,
230 void *data, int *data_size,
231 uint8_t * buf, int buf_size)
233 AMRContext *s = avctx->priv_data;
240 Word16 serial[SERIAL_FRAMESIZE]; /* coded bits */
244 static Word16 packed_size[16] = {12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0};
247 //printf("amr_decode_frame data_size=%i buf=0x%X buf_size=%d frameCount=%d!!\n",*data_size,buf,buf_size,s->frameCount);
251 // while(offset<buf_size)
254 /* read rest of the frame based on ToC byte */
255 q = (toc >> 2) & 0x01;
256 ft = (toc >> 3) & 0x0F;
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]);
262 packed_bits=amrData+offset;
264 offset+=packed_size[ft];
266 //Unsort and unpack bits
267 s->rx_type = UnpackBits(q, ft, packed_bits, &s->mode, &serial[1]);
269 //We have a new frame
272 if (s->rx_type == RX_NO_DATA)
274 s->mode = s->speech_decoder_state->prev_mode;
277 s->speech_decoder_state->prev_mode = s->mode;
280 /* if homed: check if this frame is another homing frame */
281 if (s->reset_flag_old == 1)
283 /* only check until end of first subframe */
284 s->reset_flag = decoder_homing_frame_test_first(&serial[1], s->mode);
286 /* produce encoder homing frame if homed & input=decoder homing frame */
287 if ((s->reset_flag != 0) && (s->reset_flag_old != 0))
289 for (i = 0; i < L_FRAME; i++)
297 Speech_Decode_Frame(s->speech_decoder_state, s->mode, &serial[1], s->rx_type, synth);
300 //Each AMR-frame results in 160 16-bit samples
304 /* if not homed: check whether current frame is a homing frame */
305 if (s->reset_flag_old == 0)
307 /* check whole frame */
308 s->reset_flag = decoder_homing_frame_test(&serial[1], s->mode);
310 /* reset decoder if current frame is a homing frame */
311 if (s->reset_flag != 0)
313 Speech_Decode_Frame_reset(s->speech_decoder_state);
315 s->reset_flag_old = s->reset_flag;
322 static int amr_nb_encode_frame(AVCodecContext *avctx,
323 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
325 short serial_data[250] = {0};
327 AMRContext *s = avctx->priv_data;
330 s->reset_flag = encoder_homing_frame_test(data);
332 Speech_Encode_Frame(s->enstate, s->enc_bitrate, data, &serial_data[1], &s->mode);
334 /* add frame type and mode */
335 sid_sync (s->sidstate, s->mode, &s->tx_frametype);
337 written = PackBits(s->mode, s->enc_bitrate, s->tx_frametype, &serial_data[1], frame);
339 if (s->reset_flag != 0)
341 Speech_Encode_Frame_reset(s->enstate);
342 sid_sync_reset(s->sidstate);
348 #elif defined(CONFIG_AMR_NB) /* Float point version*/
350 typedef struct AMRContext {
354 enum Mode enc_bitrate;
357 static int amr_nb_decode_init(AVCodecContext * avctx)
359 AMRContext *s = avctx->priv_data;
361 s->decState=Decoder_Interface_init();
364 av_log(avctx, AV_LOG_ERROR, "Decoder_Interface_init error\r\n");
368 amr_decode_fix_avctx(avctx);
370 if(avctx->channels > 1)
372 av_log(avctx, AV_LOG_ERROR, "amr_nb: multichannel decoding not supported\n");
379 static int amr_nb_encode_init(AVCodecContext * avctx)
381 AMRContext *s = avctx->priv_data;
384 if(avctx->sample_rate!=8000)
386 av_log(avctx, AV_LOG_ERROR, "Only 8000Hz sample rate supported\n");
390 if(avctx->channels!=1)
392 av_log(avctx, AV_LOG_ERROR, "Only mono supported\n");
396 avctx->frame_size=160;
397 avctx->coded_frame= avcodec_alloc_frame();
399 s->enstate=Encoder_Interface_init(0);
402 av_log(avctx, AV_LOG_ERROR, "Encoder_Interface_init error\n");
406 s->enc_bitrate=getBitrateMode(avctx->bit_rate);
411 static int amr_nb_decode_close(AVCodecContext * avctx)
413 AMRContext *s = avctx->priv_data;
414 Decoder_Interface_exit(s->decState);
418 static int amr_nb_encode_close(AVCodecContext * avctx)
420 AMRContext *s = avctx->priv_data;
421 Encoder_Interface_exit(s->enstate);
422 av_freep(&avctx->coded_frame);
426 static int amr_nb_decode_frame(AVCodecContext * avctx,
427 void *data, int *data_size,
428 uint8_t * buf, int buf_size)
430 AMRContext *s = (AMRContext*)avctx->priv_data;
433 static short block_size[16]={ 12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0 };
437 /* av_log(NULL,AV_LOG_DEBUG,"amr_decode_frame buf=%p buf_size=%d frameCount=%d!!\n",buf,buf_size,s->frameCount); */
439 dec_mode = (buf[0] >> 3) & 0x000F;
440 packet_size = block_size[dec_mode]+1;
442 if(packet_size > buf_size) {
443 av_log(avctx, AV_LOG_ERROR, "amr frame too short (%u, should be %u)\n", buf_size, packet_size);
448 /* 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]); */
450 Decoder_Interface_Decode(s->decState, amrData, data, 0);
456 static int amr_nb_encode_frame(AVCodecContext *avctx,
457 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
459 AMRContext *s = (AMRContext*)avctx->priv_data;
462 s->enc_bitrate=getBitrateMode(avctx->bit_rate);
464 written = Encoder_Interface_Encode(s->enstate,
469 /* 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] ); */
476 #if defined(CONFIG_AMR_NB) || defined(CONFIG_AMR_NB_FIXED)
478 AVCodec amr_nb_decoder =
490 AVCodec amr_nb_encoder =
504 /* -----------AMR wideband ------------*/
508 //To avoid duplicate typedefs from typdef in amr-nb
512 #include "amrwb_float/enc_if.h"
513 #include "amrwb_float/dec_if.h"
515 /* Common code for fixed and float version*/
516 typedef struct AMRWB_bitrates
524 static int getWBBitrateMode(int bitrate)
526 /* Adjusted so that all bitrates can be used from commandline where
527 only a multiple of 1000 can be specified*/
528 AMRWB_bitrates rates[]={ {0,7999,0}, //6.6kHz
530 {10000,13000,2},//12.65
531 {13001,14999,3},//14.25
532 {15000,17000,4},//15.85
533 {17001,18000,5},//18.25
534 {18001,22000,6},//19.85
535 {22001,23000,7},//23.05
536 {23001,24000,8},//23.85
543 if(rates[i].startrate<=bitrate && rates[i].stoprate>=bitrate)
545 return(rates[i].mode);
548 /*Return highest possible*/
553 typedef struct AMRWBContext {
560 static int amr_wb_encode_init(AVCodecContext * avctx)
562 AMRWBContext *s = (AMRWBContext*)avctx->priv_data;
565 if(avctx->sample_rate!=16000)
567 av_log(avctx, AV_LOG_ERROR, "Only 16000Hz sample rate supported\n");
571 if(avctx->channels!=1)
573 av_log(avctx, AV_LOG_ERROR, "Only mono supported\n");
577 avctx->frame_size=320;
578 avctx->coded_frame= avcodec_alloc_frame();
580 s->state = E_IF_init();
581 s->mode=getWBBitrateMode(avctx->bit_rate);
587 static int amr_wb_encode_close(AVCodecContext * avctx)
589 AMRWBContext *s = (AMRWBContext*) avctx->priv_data;
591 av_freep(&avctx->coded_frame);
596 static int amr_wb_encode_frame(AVCodecContext *avctx,
597 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
601 s = (AMRWBContext*) avctx->priv_data;
602 s->mode=getWBBitrateMode(avctx->bit_rate);
603 size = E_IF_encode(s->state, s->mode, data, frame, s->allow_dtx);
607 static int amr_wb_decode_init(AVCodecContext * avctx)
609 AMRWBContext *s = (AMRWBContext *)avctx->priv_data;
611 s->state = D_IF_init();
613 amr_decode_fix_avctx(avctx);
615 if(avctx->channels > 1)
617 av_log(avctx, AV_LOG_ERROR, "amr_wb: multichannel decoding not supported\n");
624 extern const UWord8 block_size[];
626 static int amr_wb_decode_frame(AVCodecContext * avctx,
627 void *data, int *data_size,
628 uint8_t * buf, int buf_size)
630 AMRWBContext *s = (AMRWBContext*)avctx->priv_data;
641 mode = (amrData[0] >> 3) & 0x000F;
642 packet_size = block_size[mode];
644 if(packet_size > buf_size) {
645 av_log(avctx, AV_LOG_ERROR, "amr frame too short (%u, should be %u)\n", buf_size, packet_size+1);
650 D_IF_decode( s->state, amrData, data, _good_frame);
655 static int amr_wb_decode_close(AVCodecContext * avctx)
657 AMRWBContext *s = (AMRWBContext *)avctx->priv_data;
662 AVCodec amr_wb_decoder =
667 sizeof(AMRWBContext),
674 AVCodec amr_wb_encoder =
679 sizeof(AMRWBContext),
686 #endif //CONFIG_AMR_WB