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)
191 av_log(avctx, AV_LOG_DEBUG, "Only 8000Hz sample rate supported\n");
196 if(avctx->channels!=1)
200 av_log(avctx, AV_LOG_DEBUG, "Only mono supported\n");
205 avctx->frame_size=160;
206 avctx->coded_frame= avcodec_alloc_frame();
208 if(Speech_Encode_Frame_init(&s->enstate, 0, "encoder") || sid_sync_init (&s->sidstate))
212 av_log(avctx, AV_LOG_DEBUG, "Speech_Encode_Frame_init error\n");
217 s->enc_bitrate=getBitrateMode(avctx->bit_rate);
222 static int amr_nb_encode_close(AVCodecContext * avctx)
224 AMRContext *s = avctx->priv_data;
225 Speech_Encode_Frame_exit(&s->enstate);
226 sid_sync_exit (&s->sidstate);
227 av_freep(&avctx->coded_frame);
231 static int amr_nb_decode_close(AVCodecContext * avctx)
233 AMRContext *s = avctx->priv_data;
234 Speech_Decode_Frame_exit(&s->speech_decoder_state);
238 static int amr_nb_decode_frame(AVCodecContext * avctx,
239 void *data, int *data_size,
240 uint8_t * buf, int buf_size)
242 AMRContext *s = avctx->priv_data;
249 Word16 serial[SERIAL_FRAMESIZE]; /* coded bits */
253 static Word16 packed_size[16] = {12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0};
256 //printf("amr_decode_frame data_size=%i buf=0x%X buf_size=%d frameCount=%d!!\n",*data_size,buf,buf_size,s->frameCount);
260 // while(offset<buf_size)
263 /* read rest of the frame based on ToC byte */
264 q = (toc >> 2) & 0x01;
265 ft = (toc >> 3) & 0x0F;
267 //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]);
271 packed_bits=amrData+offset;
273 offset+=packed_size[ft];
275 //Unsort and unpack bits
276 s->rx_type = UnpackBits(q, ft, packed_bits, &s->mode, &serial[1]);
278 //We have a new frame
281 if (s->rx_type == RX_NO_DATA)
283 s->mode = s->speech_decoder_state->prev_mode;
286 s->speech_decoder_state->prev_mode = s->mode;
289 /* if homed: check if this frame is another homing frame */
290 if (s->reset_flag_old == 1)
292 /* only check until end of first subframe */
293 s->reset_flag = decoder_homing_frame_test_first(&serial[1], s->mode);
295 /* produce encoder homing frame if homed & input=decoder homing frame */
296 if ((s->reset_flag != 0) && (s->reset_flag_old != 0))
298 for (i = 0; i < L_FRAME; i++)
306 Speech_Decode_Frame(s->speech_decoder_state, s->mode, &serial[1], s->rx_type, synth);
309 //Each AMR-frame results in 160 16-bit samples
313 /* if not homed: check whether current frame is a homing frame */
314 if (s->reset_flag_old == 0)
316 /* check whole frame */
317 s->reset_flag = decoder_homing_frame_test(&serial[1], s->mode);
319 /* reset decoder if current frame is a homing frame */
320 if (s->reset_flag != 0)
322 Speech_Decode_Frame_reset(s->speech_decoder_state);
324 s->reset_flag_old = s->reset_flag;
331 static int amr_nb_encode_frame(AVCodecContext *avctx,
332 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
334 short serial_data[250] = {0};
336 AMRContext *s = avctx->priv_data;
339 s->reset_flag = encoder_homing_frame_test(data);
341 Speech_Encode_Frame(s->enstate, s->enc_bitrate, data, &serial_data[1], &s->mode);
343 /* add frame type and mode */
344 sid_sync (s->sidstate, s->mode, &s->tx_frametype);
346 written = PackBits(s->mode, s->enc_bitrate, s->tx_frametype, &serial_data[1], frame);
348 if (s->reset_flag != 0)
350 Speech_Encode_Frame_reset(s->enstate);
351 sid_sync_reset(s->sidstate);
357 #elif defined(CONFIG_AMR_NB) /* Float point version*/
359 typedef struct AMRContext {
363 enum Mode enc_bitrate;
366 static int amr_nb_decode_init(AVCodecContext * avctx)
368 AMRContext *s = avctx->priv_data;
370 s->decState=Decoder_Interface_init();
373 av_log(avctx, AV_LOG_ERROR, "Decoder_Interface_init error\r\n");
377 amr_decode_fix_avctx(avctx);
379 if(avctx->channels > 1)
381 av_log(avctx, AV_LOG_ERROR, "amr_nb: multichannel decoding not supported\n");
388 static int amr_nb_encode_init(AVCodecContext * avctx)
390 AMRContext *s = avctx->priv_data;
393 if(avctx->sample_rate!=8000)
397 av_log(avctx, AV_LOG_DEBUG, "Only 8000Hz sample rate supported\n");
402 if(avctx->channels!=1)
406 av_log(avctx, AV_LOG_DEBUG, "Only mono supported\n");
411 avctx->frame_size=160;
412 avctx->coded_frame= avcodec_alloc_frame();
414 s->enstate=Encoder_Interface_init(0);
419 av_log(avctx, AV_LOG_DEBUG, "Encoder_Interface_init error\n");
424 s->enc_bitrate=getBitrateMode(avctx->bit_rate);
429 static int amr_nb_decode_close(AVCodecContext * avctx)
431 AMRContext *s = avctx->priv_data;
432 Decoder_Interface_exit(s->decState);
436 static int amr_nb_encode_close(AVCodecContext * avctx)
438 AMRContext *s = avctx->priv_data;
439 Encoder_Interface_exit(s->enstate);
440 av_freep(&avctx->coded_frame);
444 static int amr_nb_decode_frame(AVCodecContext * avctx,
445 void *data, int *data_size,
446 uint8_t * buf, int buf_size)
448 AMRContext *s = (AMRContext*)avctx->priv_data;
451 static short block_size[16]={ 12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0 };
455 /* av_log(NULL,AV_LOG_DEBUG,"amr_decode_frame buf=%p buf_size=%d frameCount=%d!!\n",buf,buf_size,s->frameCount); */
462 dec_mode = (buf[0] >> 3) & 0x000F;
463 packet_size = block_size[dec_mode]+1;
465 if(packet_size > buf_size) {
466 av_log(avctx, AV_LOG_ERROR, "amr frame too short (%u, should be %u)\n", buf_size, packet_size);
471 /* 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]); */
473 Decoder_Interface_Decode(s->decState, amrData, data, 0);
479 static int amr_nb_encode_frame(AVCodecContext *avctx,
480 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
482 AMRContext *s = (AMRContext*)avctx->priv_data;
485 s->enc_bitrate=getBitrateMode(avctx->bit_rate);
487 written = Encoder_Interface_Encode(s->enstate,
492 /* 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] ); */
499 #if defined(CONFIG_AMR_NB) || defined(CONFIG_AMR_NB_FIXED)
501 AVCodec amr_nb_decoder =
513 AVCodec amr_nb_encoder =
527 /* -----------AMR wideband ------------*/
531 //To avoid duplicate typedefs from typdef in amr-nb
535 #include "amrwb_float/enc_if.h"
536 #include "amrwb_float/dec_if.h"
538 /* Common code for fixed and float version*/
539 typedef struct AMRWB_bitrates
547 static int getWBBitrateMode(int bitrate)
549 /* Adjusted so that all bitrates can be used from commandline where
550 only a multiple of 1000 can be specified*/
551 AMRWB_bitrates rates[]={ {0,7999,0}, //6.6kHz
553 {10000,13000,2},//12.65
554 {13001,14999,3},//14.25
555 {15000,17000,4},//15.85
556 {17001,18000,5},//18.25
557 {18001,22000,6},//19.85
558 {22001,23000,7},//23.05
559 {23001,24000,8},//23.85
566 if(rates[i].startrate<=bitrate && rates[i].stoprate>=bitrate)
568 return(rates[i].mode);
571 /*Return highest possible*/
576 typedef struct AMRWBContext {
583 static int amr_wb_encode_init(AVCodecContext * avctx)
585 AMRWBContext *s = (AMRWBContext*)avctx->priv_data;
588 if(avctx->sample_rate!=16000)
592 av_log(avctx, AV_LOG_DEBUG, "Only 16000Hz sample rate supported\n");
597 if(avctx->channels!=1)
601 av_log(avctx, AV_LOG_DEBUG, "Only mono supported\n");
606 avctx->frame_size=320;
607 avctx->coded_frame= avcodec_alloc_frame();
609 s->state = E_IF_init();
610 s->mode=getWBBitrateMode(avctx->bit_rate);
616 static int amr_wb_encode_close(AVCodecContext * avctx)
618 AMRWBContext *s = (AMRWBContext*) avctx->priv_data;
620 av_freep(&avctx->coded_frame);
625 static int amr_wb_encode_frame(AVCodecContext *avctx,
626 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
630 s = (AMRWBContext*) avctx->priv_data;
631 s->mode=getWBBitrateMode(avctx->bit_rate);
632 size = E_IF_encode(s->state, s->mode, data, frame, s->allow_dtx);
636 static int amr_wb_decode_init(AVCodecContext * avctx)
638 AMRWBContext *s = (AMRWBContext *)avctx->priv_data;
640 s->state = D_IF_init();
642 amr_decode_fix_avctx(avctx);
644 if(avctx->channels > 1)
646 av_log(avctx, AV_LOG_ERROR, "amr_wb: multichannel decoding not supported\n");
653 extern const UWord8 block_size[];
655 static int amr_wb_decode_frame(AVCodecContext * avctx,
656 void *data, int *data_size,
657 uint8_t * buf, int buf_size)
659 AMRWBContext *s = (AMRWBContext*)avctx->priv_data;
670 mode = (amrData[0] >> 3) & 0x000F;
671 packet_size = block_size[mode];
673 if(packet_size > buf_size) {
674 av_log(avctx, AV_LOG_ERROR, "amr frame too short (%u, should be %u)\n", buf_size, packet_size+1);
679 D_IF_decode( s->state, amrData, data, _good_frame);
684 static int amr_wb_decode_close(AVCodecContext * avctx)
686 AMRWBContext *s = (AMRWBContext *)avctx->priv_data;
691 AVCodec amr_wb_decoder =
696 sizeof(AMRWBContext),
703 AVCodec amr_wb_encoder =
708 sizeof(AMRWBContext),
715 #endif //CONFIG_AMR_WB