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
62 #include "amr/sp_dec.h"
63 #include "amr/d_homing.h"
64 #include "amr/typedef.h"
65 #include "amr/sp_enc.h"
66 #include "amr/sid_sync.h"
67 #include "amr/e_homing.h"
70 #include "amr_float/interf_dec.h"
71 #include "amr_float/interf_enc.h"
74 /* Common code for fixed and float version*/
75 typedef struct AMR_bitrates
83 /* Match desired bitrate with closest one*/
84 static enum Mode getBitrateMode(int bitrate)
86 /* Adjusted so that all bitrates can be used from commandline where
87 only a multiple of 1000 can be specified*/
88 AMR_bitrates rates[]={ {0,4999,MR475}, //4
94 {10000,11999,MR102},//10
95 {12000,64000,MR122},//12
101 if(rates[i].startrate<=bitrate && rates[i].stoprate>=bitrate)
103 return(rates[i].mode);
106 /*Return highest possible*/
111 /* fixed point version*/
112 /* frame size in serial bitstream file (frame type + serial stream + flags) */
113 #define SERIAL_FRAMESIZE (1+MAX_SERIAL_SIZE+5)
115 typedef struct AMRContext {
117 Speech_Decode_FrameState *speech_decoder_state;
118 enum RXFrameType rx_type;
121 Word16 reset_flag_old;
123 enum Mode enc_bitrate;
124 Speech_Encode_FrameState *enstate;
125 sid_syncState *sidstate;
126 enum TXFrameType tx_frametype;
131 static int amr_nb_decode_init(AVCodecContext * avctx)
133 AMRContext *s = avctx->priv_data;
135 s->speech_decoder_state=NULL;
136 s->rx_type = (enum RXFrameType)0;
137 s->mode= (enum Mode)0;
141 if(Speech_Decode_Frame_init(&s->speech_decoder_state, "Decoder"))
143 av_log(avctx, AV_LOG_ERROR, "Speech_Decode_Frame_init error\n");
149 static int amr_nb_encode_init(AVCodecContext * avctx)
151 AMRContext *s = avctx->priv_data;
153 s->speech_decoder_state=NULL;
154 s->rx_type = (enum RXFrameType)0;
155 s->mode= (enum Mode)0;
159 if(avctx->sample_rate!=8000)
163 av_log(avctx, AV_LOG_DEBUG, "Only 8000Hz sample rate supported\n");
168 if(avctx->channels!=1)
172 av_log(avctx, AV_LOG_DEBUG, "Only mono supported\n");
177 avctx->frame_size=160;
178 avctx->coded_frame= avcodec_alloc_frame();
180 if(Speech_Encode_Frame_init(&s->enstate, 0, "encoder") || sid_sync_init (&s->sidstate))
184 av_log(avctx, AV_LOG_DEBUG, "Speech_Encode_Frame_init error\n");
189 s->enc_bitrate=getBitrateMode(avctx->bit_rate);
194 static int amr_nb_encode_close(AVCodecContext * avctx)
196 AMRContext *s = avctx->priv_data;
197 Speech_Encode_Frame_exit(&s->enstate);
198 sid_sync_exit (&s->sidstate);
199 av_freep(&avctx->coded_frame);
203 static int amr_nb_decode_close(AVCodecContext * avctx)
205 AMRContext *s = avctx->priv_data;
206 Speech_Decode_Frame_exit(&s->speech_decoder_state);
210 static int amr_nb_decode_frame(AVCodecContext * avctx,
211 void *data, int *data_size,
212 uint8_t * buf, int buf_size)
214 AMRContext *s = avctx->priv_data;
221 Word16 serial[SERIAL_FRAMESIZE]; /* coded bits */
225 static Word16 packed_size[16] = {12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0};
228 //printf("amr_decode_frame data_size=%i buf=0x%X buf_size=%d frameCount=%d!!\n",*data_size,buf,buf_size,s->frameCount);
232 // while(offset<buf_size)
235 /* read rest of the frame based on ToC byte */
236 q = (toc >> 2) & 0x01;
237 ft = (toc >> 3) & 0x0F;
239 //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]);
243 packed_bits=amrData+offset;
245 offset+=packed_size[ft];
247 //Unsort and unpack bits
248 s->rx_type = UnpackBits(q, ft, packed_bits, &s->mode, &serial[1]);
250 //We have a new frame
253 if (s->rx_type == RX_NO_DATA)
255 s->mode = s->speech_decoder_state->prev_mode;
258 s->speech_decoder_state->prev_mode = s->mode;
261 /* if homed: check if this frame is another homing frame */
262 if (s->reset_flag_old == 1)
264 /* only check until end of first subframe */
265 s->reset_flag = decoder_homing_frame_test_first(&serial[1], s->mode);
267 /* produce encoder homing frame if homed & input=decoder homing frame */
268 if ((s->reset_flag != 0) && (s->reset_flag_old != 0))
270 for (i = 0; i < L_FRAME; i++)
278 Speech_Decode_Frame(s->speech_decoder_state, s->mode, &serial[1], s->rx_type, synth);
281 //Each AMR-frame results in 160 16-bit samples
285 /* if not homed: check whether current frame is a homing frame */
286 if (s->reset_flag_old == 0)
288 /* check whole frame */
289 s->reset_flag = decoder_homing_frame_test(&serial[1], s->mode);
291 /* reset decoder if current frame is a homing frame */
292 if (s->reset_flag != 0)
294 Speech_Decode_Frame_reset(s->speech_decoder_state);
296 s->reset_flag_old = s->reset_flag;
303 static int amr_nb_encode_frame(AVCodecContext *avctx,
304 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
306 short serial_data[250] = {0};
308 AMRContext *s = avctx->priv_data;
311 s->reset_flag = encoder_homing_frame_test(data);
313 Speech_Encode_Frame(s->enstate, s->enc_bitrate, data, &serial_data[1], &s->mode);
315 /* add frame type and mode */
316 sid_sync (s->sidstate, s->mode, &s->tx_frametype);
318 written = PackBits(s->mode, s->enc_bitrate, s->tx_frametype, &serial_data[1], frame);
320 if (s->reset_flag != 0)
322 Speech_Encode_Frame_reset(s->enstate);
323 sid_sync_reset(s->sidstate);
329 #else /* Float point version*/
331 typedef struct AMRContext {
335 enum Mode enc_bitrate;
338 static int amr_nb_decode_init(AVCodecContext * avctx)
340 AMRContext *s = avctx->priv_data;
342 s->decState=Decoder_Interface_init();
345 av_log(avctx, AV_LOG_ERROR, "Decoder_Interface_init error\r\n");
351 static int amr_nb_encode_init(AVCodecContext * avctx)
353 AMRContext *s = avctx->priv_data;
356 if(avctx->sample_rate!=8000)
360 av_log(avctx, AV_LOG_DEBUG, "Only 8000Hz sample rate supported\n");
365 if(avctx->channels!=1)
369 av_log(avctx, AV_LOG_DEBUG, "Only mono supported\n");
374 avctx->frame_size=160;
375 avctx->coded_frame= avcodec_alloc_frame();
377 s->enstate=Encoder_Interface_init(0);
382 av_log(avctx, AV_LOG_DEBUG, "Encoder_Interface_init error\n");
387 s->enc_bitrate=getBitrateMode(avctx->bit_rate);
392 static int amr_nb_decode_close(AVCodecContext * avctx)
394 AMRContext *s = avctx->priv_data;
395 Decoder_Interface_exit(s->decState);
399 static int amr_nb_encode_close(AVCodecContext * avctx)
401 AMRContext *s = avctx->priv_data;
402 Encoder_Interface_exit(s->enstate);
403 av_freep(&avctx->coded_frame);
407 static int amr_nb_decode_frame(AVCodecContext * avctx,
408 void *data, int *data_size,
409 uint8_t * buf, int buf_size)
411 AMRContext *s = (AMRContext*)avctx->priv_data;
415 static short block_size[16]={ 12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0 };
419 //printf("amr_decode_frame data_size=%i buf=0x%X buf_size=%d frameCount=%d!!\n",*data_size,buf,buf_size,s->frameCount);
421 while(offset<buf_size)
423 dec_mode = (amrData[offset] >> 3) & 0x000F;
424 packet_size = block_size[dec_mode];
427 //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]);
429 Decoder_Interface_Decode(s->decState, &amrData[offset], data+*data_size, 0);
432 offset+=packet_size+1;
437 static int amr_nb_encode_frame(AVCodecContext *avctx,
438 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
440 AMRContext *s = (AMRContext*)avctx->priv_data;
443 written = Encoder_Interface_Encode(s->enstate,
454 AVCodec amr_nb_decoder =
466 AVCodec amr_nb_encoder =
478 /* -----------AMR wideband ------------*/
482 //To avoid duplicate typedefs from typdef in amr-nb
486 #include "amrwb_float/enc_if.h"
487 #include "amrwb_float/dec_if.h"
489 /* Common code for fixed and float version*/
490 typedef struct AMRWB_bitrates
498 static int getWBBitrateMode(int bitrate)
500 /* Adjusted so that all bitrates can be used from commandline where
501 only a multiple of 1000 can be specified*/
502 AMRWB_bitrates rates[]={ {0,7999,0}, //6.6kHz
504 {10000,13000,2},//12.65
505 {13001,14999,3},//14.25
506 {15000,17000,4},//15.85
507 {17001,18000,5},//18.25
508 {18001,22000,6},//19.85
509 {22001,23000,7},//23.05
510 {23001,24000,8},//23.85
517 if(rates[i].startrate<=bitrate && rates[i].stoprate>=bitrate)
519 return(rates[i].mode);
522 /*Return highest possible*/
527 typedef struct AMRWBContext {
534 static int amr_wb_encode_init(AVCodecContext * avctx)
536 AMRWBContext *s = (AMRWBContext*)avctx->priv_data;
539 if(avctx->sample_rate!=16000)
543 av_log(avctx, AV_LOG_DEBUG, "Only 16000Hz sample rate supported\n");
548 if(avctx->channels!=1)
552 av_log(avctx, AV_LOG_DEBUG, "Only mono supported\n");
557 avctx->frame_size=320;
558 avctx->coded_frame= avcodec_alloc_frame();
560 s->state = E_IF_init();
561 s->mode=getWBBitrateMode(avctx->bit_rate);
567 static int amr_wb_encode_close(AVCodecContext * avctx)
569 AMRWBContext *s = (AMRWBContext*) avctx->priv_data;
571 av_freep(&avctx->coded_frame);
576 static int amr_wb_encode_frame(AVCodecContext *avctx,
577 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
579 AMRWBContext *s = (AMRWBContext*) avctx->priv_data;
580 int size = E_IF_encode(s->state, s->mode, data, frame, s->allow_dtx);
584 static int amr_wb_decode_init(AVCodecContext * avctx)
586 AMRWBContext *s = (AMRWBContext *)avctx->priv_data;
588 s->state = D_IF_init();
592 extern const UWord8 block_size[];
594 static int amr_wb_decode_frame(AVCodecContext * avctx,
595 void *data, int *data_size,
596 uint8_t * buf, int buf_size)
598 AMRWBContext *s = (AMRWBContext*)avctx->priv_data;
605 while(offset<buf_size)
608 mode = (Word16)((amrData[offset] >> 3) & 0x0F);
609 packet_size = block_size[mode];
610 D_IF_decode( s->state, &amrData[offset], data+*data_size, _good_frame);
617 static int amr_wb_decode_close(AVCodecContext * avctx)
619 AMRWBContext *s = (AMRWBContext *)avctx->priv_data;
624 AVCodec amr_wb_decoder =
629 sizeof(AMRWBContext),
636 AVCodec amr_wb_encoder =
641 sizeof(AMRWBContext),