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 */
226 static Word16 packed_size[16] = {12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0};
229 //printf("amr_decode_frame data_size=%i buf=0x%X buf_size=%d frameCount=%d!!\n",*data_size,buf,buf_size,s->frameCount);
233 // while(offset<buf_size)
236 /* read rest of the frame based on ToC byte */
237 q = (toc >> 2) & 0x01;
238 ft = (toc >> 3) & 0x0F;
240 //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]);
244 packed_bits=amrData+offset;
246 offset+=packed_size[ft];
248 //Unsort and unpack bits
249 s->rx_type = UnpackBits(q, ft, packed_bits, &s->mode, &serial[1]);
251 //We have a new frame
254 if (s->rx_type == RX_NO_DATA)
256 s->mode = s->speech_decoder_state->prev_mode;
259 s->speech_decoder_state->prev_mode = s->mode;
262 /* if homed: check if this frame is another homing frame */
263 if (s->reset_flag_old == 1)
265 /* only check until end of first subframe */
266 s->reset_flag = decoder_homing_frame_test_first(&serial[1], s->mode);
268 /* produce encoder homing frame if homed & input=decoder homing frame */
269 if ((s->reset_flag != 0) && (s->reset_flag_old != 0))
271 for (i = 0; i < L_FRAME; i++)
279 Speech_Decode_Frame(s->speech_decoder_state, s->mode, &serial[1], s->rx_type, synth);
282 //Each AMR-frame results in 160 16-bit samples
286 /* if not homed: check whether current frame is a homing frame */
287 if (s->reset_flag_old == 0)
289 /* check whole frame */
290 s->reset_flag = decoder_homing_frame_test(&serial[1], s->mode);
292 /* reset decoder if current frame is a homing frame */
293 if (s->reset_flag != 0)
295 Speech_Decode_Frame_reset(s->speech_decoder_state);
297 s->reset_flag_old = s->reset_flag;
304 static int amr_nb_encode_frame(AVCodecContext *avctx,
305 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
307 short serial_data[250] = {0};
309 AMRContext *s = avctx->priv_data;
312 s->reset_flag = encoder_homing_frame_test(data);
314 Speech_Encode_Frame(s->enstate, s->enc_bitrate, data, &serial_data[1], &s->mode);
316 /* add frame type and mode */
317 sid_sync (s->sidstate, s->mode, &s->tx_frametype);
319 written = PackBits(s->mode, s->enc_bitrate, s->tx_frametype, &serial_data[1], frame);
321 if (s->reset_flag != 0)
323 Speech_Encode_Frame_reset(s->enstate);
324 sid_sync_reset(s->sidstate);
330 #else /* Float point version*/
332 typedef struct AMRContext {
336 enum Mode enc_bitrate;
339 static int amr_nb_decode_init(AVCodecContext * avctx)
341 AMRContext *s = avctx->priv_data;
343 s->decState=Decoder_Interface_init();
346 av_log(avctx, AV_LOG_ERROR, "Decoder_Interface_init error\r\n");
352 static int amr_nb_encode_init(AVCodecContext * avctx)
354 AMRContext *s = avctx->priv_data;
357 if(avctx->sample_rate!=8000)
361 av_log(avctx, AV_LOG_DEBUG, "Only 8000Hz sample rate supported\n");
366 if(avctx->channels!=1)
370 av_log(avctx, AV_LOG_DEBUG, "Only mono supported\n");
375 avctx->frame_size=160;
376 avctx->coded_frame= avcodec_alloc_frame();
378 s->enstate=Encoder_Interface_init(0);
383 av_log(avctx, AV_LOG_DEBUG, "Encoder_Interface_init error\n");
388 s->enc_bitrate=getBitrateMode(avctx->bit_rate);
393 static int amr_nb_decode_close(AVCodecContext * avctx)
395 AMRContext *s = avctx->priv_data;
396 Decoder_Interface_exit(s->decState);
400 static int amr_nb_encode_close(AVCodecContext * avctx)
402 AMRContext *s = avctx->priv_data;
403 Encoder_Interface_exit(s->enstate);
404 av_freep(&avctx->coded_frame);
408 static int amr_nb_decode_frame(AVCodecContext * avctx,
409 void *data, int *data_size,
410 uint8_t * buf, int buf_size)
412 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 };
421 //printf("amr_decode_frame data_size=%i buf=0x%X buf_size=%d frameCount=%d!!\n",*data_size,buf,buf_size,s->frameCount);
423 while(offset<buf_size)
425 dec_mode = (amrData[offset] >> 3) & 0x000F;
426 packet_size = block_size[dec_mode];
429 //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]);
431 Decoder_Interface_Decode(s->decState, &amrData[offset], data+*data_size, 0);
434 offset+=packet_size+1;
439 static int amr_nb_encode_frame(AVCodecContext *avctx,
440 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
442 AMRContext *s = (AMRContext*)avctx->priv_data;
445 written = Encoder_Interface_Encode(s->enstate,
456 AVCodec amr_nb_decoder =
468 AVCodec amr_nb_encoder =
480 /* -----------AMR wideband ------------*/
484 //To avoid duplicate typedefs from typdef in amr-nb
488 #include "amrwb_float/enc_if.h"
489 #include "amrwb_float/dec_if.h"
491 /* Common code for fixed and float version*/
492 typedef struct AMRWB_bitrates
500 static int getWBBitrateMode(int bitrate)
502 /* Adjusted so that all bitrates can be used from commandline where
503 only a multiple of 1000 can be specified*/
504 AMRWB_bitrates rates[]={ {0,7999,0}, //6.6kHz
506 {10000,13000,2},//12.65
507 {13001,14999,3},//14.25
508 {15000,17000,4},//15.85
509 {17001,18000,5},//18.25
510 {18001,22000,6},//19.85
511 {22001,23000,7},//23.05
512 {23001,24000,8},//23.85
519 if(rates[i].startrate<=bitrate && rates[i].stoprate>=bitrate)
521 return(rates[i].mode);
524 /*Return highest possible*/
529 typedef struct AMRWBContext {
536 static int amr_wb_encode_init(AVCodecContext * avctx)
538 AMRWBContext *s = (AMRWBContext*)avctx->priv_data;
541 if(avctx->sample_rate!=16000)
545 av_log(avctx, AV_LOG_DEBUG, "Only 16000Hz sample rate supported\n");
550 if(avctx->channels!=1)
554 av_log(avctx, AV_LOG_DEBUG, "Only mono supported\n");
559 avctx->frame_size=320;
560 avctx->coded_frame= avcodec_alloc_frame();
562 s->state = E_IF_init();
563 s->mode=getWBBitrateMode(avctx->bit_rate);
569 static int amr_wb_encode_close(AVCodecContext * avctx)
571 AMRWBContext *s = (AMRWBContext*) avctx->priv_data;
573 av_freep(&avctx->coded_frame);
578 static int amr_wb_encode_frame(AVCodecContext *avctx,
579 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
581 AMRWBContext *s = (AMRWBContext*) avctx->priv_data;
582 int size = E_IF_encode(s->state, s->mode, data, frame, s->allow_dtx);
586 static int amr_wb_decode_init(AVCodecContext * avctx)
588 AMRWBContext *s = (AMRWBContext *)avctx->priv_data;
590 s->state = D_IF_init();
594 extern const UWord8 block_size[];
596 static int amr_wb_decode_frame(AVCodecContext * avctx,
597 void *data, int *data_size,
598 uint8_t * buf, int buf_size)
600 AMRWBContext *s = (AMRWBContext*)avctx->priv_data;
608 while(offset<buf_size)
611 mode = (Word16)((amrData[offset] >> 3) & 0x0F);
612 packet_size = block_size[mode];
613 D_IF_decode( s->state, &amrData[offset], data+*data_size, _good_frame);
620 static int amr_wb_decode_close(AVCodecContext * avctx)
622 AMRWBContext *s = (AMRWBContext *)avctx->priv_data;
627 AVCodec amr_wb_decoder =
632 sizeof(AMRWBContext),
639 AVCodec amr_wb_encoder =
644 sizeof(AMRWBContext),