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 #ifdef CONFIG_AMR_NB_FIXED
113 /* fixed point version*/
114 /* frame size in serial bitstream file (frame type + serial stream + flags) */
115 #define SERIAL_FRAMESIZE (1+MAX_SERIAL_SIZE+5)
117 typedef struct AMRContext {
119 Speech_Decode_FrameState *speech_decoder_state;
120 enum RXFrameType rx_type;
123 Word16 reset_flag_old;
125 enum Mode enc_bitrate;
126 Speech_Encode_FrameState *enstate;
127 sid_syncState *sidstate;
128 enum TXFrameType tx_frametype;
133 static int amr_nb_decode_init(AVCodecContext * avctx)
135 AMRContext *s = avctx->priv_data;
137 s->speech_decoder_state=NULL;
138 s->rx_type = (enum RXFrameType)0;
139 s->mode= (enum Mode)0;
143 if(Speech_Decode_Frame_init(&s->speech_decoder_state, "Decoder"))
145 av_log(avctx, AV_LOG_ERROR, "Speech_Decode_Frame_init error\n");
151 static int amr_nb_encode_init(AVCodecContext * avctx)
153 AMRContext *s = avctx->priv_data;
155 s->speech_decoder_state=NULL;
156 s->rx_type = (enum RXFrameType)0;
157 s->mode= (enum Mode)0;
161 if(avctx->sample_rate!=8000)
165 av_log(avctx, AV_LOG_DEBUG, "Only 8000Hz sample rate supported\n");
170 if(avctx->channels!=1)
174 av_log(avctx, AV_LOG_DEBUG, "Only mono supported\n");
179 avctx->frame_size=160;
180 avctx->coded_frame= avcodec_alloc_frame();
182 if(Speech_Encode_Frame_init(&s->enstate, 0, "encoder") || sid_sync_init (&s->sidstate))
186 av_log(avctx, AV_LOG_DEBUG, "Speech_Encode_Frame_init error\n");
191 s->enc_bitrate=getBitrateMode(avctx->bit_rate);
196 static int amr_nb_encode_close(AVCodecContext * avctx)
198 AMRContext *s = avctx->priv_data;
199 Speech_Encode_Frame_exit(&s->enstate);
200 sid_sync_exit (&s->sidstate);
201 av_freep(&avctx->coded_frame);
205 static int amr_nb_decode_close(AVCodecContext * avctx)
207 AMRContext *s = avctx->priv_data;
208 Speech_Decode_Frame_exit(&s->speech_decoder_state);
212 static int amr_nb_decode_frame(AVCodecContext * avctx,
213 void *data, int *data_size,
214 uint8_t * buf, int buf_size)
216 AMRContext *s = avctx->priv_data;
223 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 #elif defined(CONFIG_AMR_NB) /* 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;
416 static short block_size[16]={ 12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0 };
420 /* av_log(NULL,AV_LOG_DEBUG,"amr_decode_frame buf=%p buf_size=%d frameCount=%d!!\n",buf,buf_size,s->frameCount); */
427 dec_mode = (buf[0] >> 3) & 0x000F;
428 packet_size = block_size[dec_mode]+1;
430 if(packet_size > buf_size) {
431 av_log(avctx, AV_LOG_ERROR, "amr frame too short (%u, should be %u)\n", buf_size, packet_size);
436 /* 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]); */
438 Decoder_Interface_Decode(s->decState, amrData, data, 0);
444 static int amr_nb_encode_frame(AVCodecContext *avctx,
445 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
447 AMRContext *s = (AMRContext*)avctx->priv_data;
450 written = Encoder_Interface_Encode(s->enstate,
455 /* 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] ); */
462 #if defined(CONFIG_AMR_NB) || defined(CONFIG_AMR_NB_FIXED)
464 AVCodec amr_nb_decoder =
476 AVCodec amr_nb_encoder =
490 /* -----------AMR wideband ------------*/
494 //To avoid duplicate typedefs from typdef in amr-nb
498 #include "amrwb_float/enc_if.h"
499 #include "amrwb_float/dec_if.h"
501 /* Common code for fixed and float version*/
502 typedef struct AMRWB_bitrates
510 static int getWBBitrateMode(int bitrate)
512 /* Adjusted so that all bitrates can be used from commandline where
513 only a multiple of 1000 can be specified*/
514 AMRWB_bitrates rates[]={ {0,7999,0}, //6.6kHz
516 {10000,13000,2},//12.65
517 {13001,14999,3},//14.25
518 {15000,17000,4},//15.85
519 {17001,18000,5},//18.25
520 {18001,22000,6},//19.85
521 {22001,23000,7},//23.05
522 {23001,24000,8},//23.85
529 if(rates[i].startrate<=bitrate && rates[i].stoprate>=bitrate)
531 return(rates[i].mode);
534 /*Return highest possible*/
539 typedef struct AMRWBContext {
546 static int amr_wb_encode_init(AVCodecContext * avctx)
548 AMRWBContext *s = (AMRWBContext*)avctx->priv_data;
551 if(avctx->sample_rate!=16000)
555 av_log(avctx, AV_LOG_DEBUG, "Only 16000Hz sample rate supported\n");
560 if(avctx->channels!=1)
564 av_log(avctx, AV_LOG_DEBUG, "Only mono supported\n");
569 avctx->frame_size=320;
570 avctx->coded_frame= avcodec_alloc_frame();
572 s->state = E_IF_init();
573 s->mode=getWBBitrateMode(avctx->bit_rate);
579 static int amr_wb_encode_close(AVCodecContext * avctx)
581 AMRWBContext *s = (AMRWBContext*) avctx->priv_data;
583 av_freep(&avctx->coded_frame);
588 static int amr_wb_encode_frame(AVCodecContext *avctx,
589 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
591 AMRWBContext *s = (AMRWBContext*) avctx->priv_data;
592 int size = E_IF_encode(s->state, s->mode, data, frame, s->allow_dtx);
596 static int amr_wb_decode_init(AVCodecContext * avctx)
598 AMRWBContext *s = (AMRWBContext *)avctx->priv_data;
600 s->state = D_IF_init();
604 extern const UWord8 block_size[];
606 static int amr_wb_decode_frame(AVCodecContext * avctx,
607 void *data, int *data_size,
608 uint8_t * buf, int buf_size)
610 AMRWBContext *s = (AMRWBContext*)avctx->priv_data;
621 mode = (amrData[0] >> 3) & 0x000F;
622 packet_size = block_size[mode];
624 if(packet_size > buf_size) {
625 av_log(avctx, AV_LOG_ERROR, "amr frame too short (%u, should be %u)\n", buf_size, packet_size+1);
630 D_IF_decode( s->state, amrData, data, _good_frame);
635 static int amr_wb_decode_close(AVCodecContext * avctx)
637 AMRWBContext *s = (AMRWBContext *)avctx->priv_data;
642 AVCodec amr_wb_decoder =
647 sizeof(AMRWBContext),
654 AVCodec amr_wb_encoder =
659 sizeof(AMRWBContext),
666 #endif //CONFIG_AMR_WB