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
23 * Adaptive Multi-Rate (AMR) Audio decoder stub.
25 * This code implements both an AMR-NarrowBand (AMR-NB) and an AMR-WideBand
26 * (AMR-WB) audio encoder/decoder through external reference code from
27 * http://www.3gpp.org/. The license of the code from 3gpp is unclear so you
28 * have to download the code separately. Two versions exists: One fixed-point
29 * and one floating-point. For some reason the float encoder is significantly
30 * faster at least on a P4 1.5GHz (0.9s instead of 9.9s on a 30s audio clip
31 * at MR102). Both float and fixed point are supported for AMR-NB, but only
37 * The float version (default) can be downloaded from:
38 * http://www.3gpp.org/ftp/Specs/archive/26_series/26.104/26104-610.zip
40 * \subsection Fixed-point
41 * The fixed-point (TS26.073) can be downloaded from:
42 * http://www.3gpp.org/ftp/Specs/archive/26_series/26.073/26073-600.zip
44 * \subsection Specification
45 * The specification for AMR-NB can be found in TS 26.071
46 * (http://www.3gpp.org/ftp/Specs/html-info/26071.htm) and some other
47 * info at http://www.3gpp.org/ftp/Specs/html-info/26-series.htm.
52 * The reference code can be downloaded from:
53 * http://www.3gpp.org/ftp/Specs/archive/26_series/26.204/26204-600.zip
55 * \subsection Fixed-point
56 * If someone wants to use the fixed point version it can be downloaded from:
57 * http://www.3gpp.org/ftp/Specs/archive/26_series/26.173/26173-571.zip.
59 * \subsection Specification
60 * The specification for AMR-WB can be found in TS 26.171
61 * (http://www.3gpp.org/ftp/Specs/html-info/26171.htm) and some other
62 * info at http://www.3gpp.org/ftp/Specs/html-info/26-series.htm.
68 #ifdef CONFIG_LIBAMR_NB_FIXED
72 #include "amr/sp_dec.h"
73 #include "amr/d_homing.h"
74 #include "amr/typedef.h"
75 #include "amr/sp_enc.h"
76 #include "amr/sid_sync.h"
77 #include "amr/e_homing.h"
80 #include <amrnb/interf_dec.h>
81 #include <amrnb/interf_enc.h>
84 static const char *nb_bitrate_unsupported =
85 "bitrate not supported: use one of 4.75k, 5.15k, 5.9k, 6.7k, 7.4k, 7.95k, 10.2k or 12.2k\n";
86 static const char *wb_bitrate_unsupported =
87 "bitrate not supported: use one of 6.6k, 8.85k, 12.65k, 14.25k, 15.85k, 18.25k, 19.85k, 23.05k, or 23.85k\n";
89 /* Common code for fixed and float version*/
90 typedef struct AMR_bitrates
96 /* Match desired bitrate */
97 static int getBitrateMode(int bitrate)
99 /* make the correspondance between bitrate and mode */
100 AMR_bitrates rates[]={ {4750,MR475},
113 if(rates[i].rate==bitrate)
115 return rates[i].mode;
118 /* no bitrate matching, return an error */
122 static void amr_decode_fix_avctx(AVCodecContext * avctx)
124 const int is_amr_wb = 1 + (avctx->codec_id == CODEC_ID_AMR_WB);
126 if(avctx->sample_rate == 0)
128 avctx->sample_rate = 8000 * is_amr_wb;
131 if(avctx->channels == 0)
136 avctx->frame_size = 160 * is_amr_wb;
139 #ifdef CONFIG_LIBAMR_NB_FIXED
140 /* fixed point version*/
141 /* frame size in serial bitstream file (frame type + serial stream + flags) */
142 #define SERIAL_FRAMESIZE (1+MAX_SERIAL_SIZE+5)
144 typedef struct AMRContext {
146 Speech_Decode_FrameState *speech_decoder_state;
147 enum RXFrameType rx_type;
150 Word16 reset_flag_old;
153 Speech_Encode_FrameState *enstate;
154 sid_syncState *sidstate;
155 enum TXFrameType tx_frametype;
158 static int amr_nb_decode_init(AVCodecContext * avctx)
160 AMRContext *s = avctx->priv_data;
163 s->speech_decoder_state=NULL;
164 s->rx_type = (enum RXFrameType)0;
165 s->mode= (enum Mode)0;
169 if(Speech_Decode_Frame_init(&s->speech_decoder_state, "Decoder"))
171 av_log(avctx, AV_LOG_ERROR, "Speech_Decode_Frame_init error\n");
175 amr_decode_fix_avctx(avctx);
177 if(avctx->channels > 1)
179 av_log(avctx, AV_LOG_ERROR, "amr_nb: multichannel decoding not supported\n");
186 static int amr_nb_encode_init(AVCodecContext * avctx)
188 AMRContext *s = avctx->priv_data;
191 s->speech_decoder_state=NULL;
192 s->rx_type = (enum RXFrameType)0;
193 s->mode= (enum Mode)0;
197 if(avctx->sample_rate!=8000)
199 av_log(avctx, AV_LOG_ERROR, "Only 8000Hz sample rate supported\n");
203 if(avctx->channels!=1)
205 av_log(avctx, AV_LOG_ERROR, "Only mono supported\n");
209 avctx->frame_size=160;
210 avctx->coded_frame= avcodec_alloc_frame();
212 if(Speech_Encode_Frame_init(&s->enstate, 0, "encoder") || sid_sync_init (&s->sidstate))
214 av_log(avctx, AV_LOG_ERROR, "Speech_Encode_Frame_init error\n");
218 if((s->enc_bitrate=getBitrateMode(avctx->bit_rate))<0)
220 av_log(avctx, AV_LOG_ERROR, nb_bitrate_unsupported);
227 static int amr_nb_encode_close(AVCodecContext * avctx)
229 AMRContext *s = avctx->priv_data;
231 Speech_Encode_Frame_exit(&s->enstate);
232 sid_sync_exit (&s->sidstate);
233 av_freep(&avctx->coded_frame);
237 static int amr_nb_decode_close(AVCodecContext * avctx)
239 AMRContext *s = avctx->priv_data;
241 Speech_Decode_Frame_exit(&s->speech_decoder_state);
245 static int amr_nb_decode_frame(AVCodecContext * avctx,
246 void *data, int *data_size,
247 uint8_t * buf, int buf_size)
249 AMRContext *s = avctx->priv_data;
253 Word16 serial[SERIAL_FRAMESIZE]; /* coded bits */
256 static Word16 packed_size[16] = {12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0};
259 //printf("amr_decode_frame data_size=%i buf=0x%X buf_size=%d frameCount=%d!!\n",*data_size,buf,buf_size,s->frameCount);
264 /* read rest of the frame based on ToC byte */
265 q = (toc >> 2) & 0x01;
266 ft = (toc >> 3) & 0x0F;
268 //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]);
272 packed_bits=amrData+offset;
274 offset+=packed_size[ft];
276 //Unsort and unpack bits
277 s->rx_type = UnpackBits(q, ft, packed_bits, &s->mode, &serial[1]);
279 //We have a new frame
282 if (s->rx_type == RX_NO_DATA)
284 s->mode = s->speech_decoder_state->prev_mode;
287 s->speech_decoder_state->prev_mode = s->mode;
290 /* if homed: check if this frame is another homing frame */
291 if (s->reset_flag_old == 1)
293 /* only check until end of first subframe */
294 s->reset_flag = decoder_homing_frame_test_first(&serial[1], s->mode);
296 /* produce encoder homing frame if homed & input=decoder homing frame */
297 if ((s->reset_flag != 0) && (s->reset_flag_old != 0))
299 for (i = 0; i < L_FRAME; i++)
307 Speech_Decode_Frame(s->speech_decoder_state, s->mode, &serial[1], s->rx_type, synth);
310 //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;
330 static int amr_nb_encode_frame(AVCodecContext *avctx,
331 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
333 short serial_data[250] = {0};
334 AMRContext *s = avctx->priv_data;
337 s->reset_flag = encoder_homing_frame_test(data);
339 Speech_Encode_Frame(s->enstate, s->enc_bitrate, data, &serial_data[1], &s->mode);
341 /* add frame type and mode */
342 sid_sync (s->sidstate, s->mode, &s->tx_frametype);
344 written = PackBits(s->mode, s->enc_bitrate, s->tx_frametype, &serial_data[1], frame);
346 if (s->reset_flag != 0)
348 Speech_Encode_Frame_reset(s->enstate);
349 sid_sync_reset(s->sidstate);
355 #elif defined(CONFIG_LIBAMR_NB) /* Float point version*/
357 typedef struct AMRContext {
364 static int amr_nb_decode_init(AVCodecContext * avctx)
366 AMRContext *s = avctx->priv_data;
369 s->decState=Decoder_Interface_init();
372 av_log(avctx, AV_LOG_ERROR, "Decoder_Interface_init error\r\n");
376 amr_decode_fix_avctx(avctx);
378 if(avctx->channels > 1)
380 av_log(avctx, AV_LOG_ERROR, "amr_nb: multichannel decoding not supported\n");
387 static int amr_nb_encode_init(AVCodecContext * avctx)
389 AMRContext *s = avctx->priv_data;
393 if(avctx->sample_rate!=8000)
395 av_log(avctx, AV_LOG_ERROR, "Only 8000Hz sample rate supported\n");
399 if(avctx->channels!=1)
401 av_log(avctx, AV_LOG_ERROR, "Only mono supported\n");
405 avctx->frame_size=160;
406 avctx->coded_frame= avcodec_alloc_frame();
408 s->enstate=Encoder_Interface_init(0);
411 av_log(avctx, AV_LOG_ERROR, "Encoder_Interface_init error\n");
415 if((s->enc_bitrate=getBitrateMode(avctx->bit_rate))<0)
417 av_log(avctx, AV_LOG_ERROR, nb_bitrate_unsupported);
424 static int amr_nb_decode_close(AVCodecContext * avctx)
426 AMRContext *s = avctx->priv_data;
428 Decoder_Interface_exit(s->decState);
432 static int amr_nb_encode_close(AVCodecContext * avctx)
434 AMRContext *s = avctx->priv_data;
436 Encoder_Interface_exit(s->enstate);
437 av_freep(&avctx->coded_frame);
441 static int amr_nb_decode_frame(AVCodecContext * avctx,
442 void *data, int *data_size,
443 uint8_t * buf, int buf_size)
445 AMRContext *s = avctx->priv_data;
447 static const uint8_t block_size[16]={ 12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0 };
451 /* av_log(NULL,AV_LOG_DEBUG,"amr_decode_frame buf=%p buf_size=%d frameCount=%d!!\n",buf,buf_size,s->frameCount); */
453 dec_mode = (buf[0] >> 3) & 0x000F;
454 packet_size = block_size[dec_mode]+1;
456 if(packet_size > buf_size) {
457 av_log(avctx, AV_LOG_ERROR, "amr frame too short (%u, should be %u)\n", buf_size, packet_size);
462 /* 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]); */
464 Decoder_Interface_Decode(s->decState, amrData, data, 0);
470 static int amr_nb_encode_frame(AVCodecContext *avctx,
471 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
473 AMRContext *s = avctx->priv_data;
476 if((s->enc_bitrate=getBitrateMode(avctx->bit_rate))<0)
478 av_log(avctx, AV_LOG_ERROR, nb_bitrate_unsupported);
482 written = Encoder_Interface_Encode(s->enstate,
487 /* 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] ); */
494 #if defined(CONFIG_LIBAMR_NB) || defined(CONFIG_LIBAMR_NB_FIXED)
496 AVCodec libamr_nb_decoder =
506 .long_name = "libamr-nb Adaptive Multi-Rate (AMR) Narrow-Band",
509 AVCodec libamr_nb_encoder =
519 .long_name = "libamr-nb Adaptive Multi-Rate (AMR) Narrow-Band",
524 /* -----------AMR wideband ------------*/
525 #ifdef CONFIG_LIBAMR_WB
528 //To avoid duplicate typedefs from typedef in amr-nb
532 #include <amrwb/enc_if.h>
533 #include <amrwb/dec_if.h>
534 #include <amrwb/if_rom.h>
536 /* Common code for fixed and float version*/
537 typedef struct AMRWB_bitrates
543 static int getWBBitrateMode(int bitrate)
545 /* make the correspondance between bitrate and mode */
546 AMRWB_bitrates rates[]={ {6600,0},
560 if(rates[i].rate==bitrate)
562 return rates[i].mode;
565 /* no bitrate matching, return an error */
570 typedef struct AMRWBContext {
577 static int amr_wb_encode_init(AVCodecContext * avctx)
579 AMRWBContext *s = avctx->priv_data;
583 if(avctx->sample_rate!=16000)
585 av_log(avctx, AV_LOG_ERROR, "Only 16000Hz sample rate supported\n");
589 if(avctx->channels!=1)
591 av_log(avctx, AV_LOG_ERROR, "Only mono supported\n");
595 if((s->mode=getWBBitrateMode(avctx->bit_rate))<0)
597 av_log(avctx, AV_LOG_ERROR, wb_bitrate_unsupported);
601 avctx->frame_size=320;
602 avctx->coded_frame= avcodec_alloc_frame();
604 s->state = E_IF_init();
610 static int amr_wb_encode_close(AVCodecContext * avctx)
612 AMRWBContext *s = avctx->priv_data;
615 av_freep(&avctx->coded_frame);
620 static int amr_wb_encode_frame(AVCodecContext *avctx,
621 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
623 AMRWBContext *s = avctx->priv_data;
626 if((s->mode=getWBBitrateMode(avctx->bit_rate))<0)
628 av_log(avctx, AV_LOG_ERROR, wb_bitrate_unsupported);
631 size = E_IF_encode(s->state, s->mode, data, frame, s->allow_dtx);
635 static int amr_wb_decode_init(AVCodecContext * avctx)
637 AMRWBContext *s = 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 static int amr_wb_decode_frame(AVCodecContext * avctx,
654 void *data, int *data_size,
655 uint8_t * buf, int buf_size)
657 AMRWBContext *s = avctx->priv_data;
661 static const uint8_t block_size[16] = {18, 23, 33, 37, 41, 47, 51, 59, 61, 6, 6, 0, 0, 0, 1, 1};
668 mode = (amrData[0] >> 3) & 0x000F;
669 packet_size = block_size[mode];
671 if(packet_size > buf_size) {
672 av_log(avctx, AV_LOG_ERROR, "amr frame too short (%u, should be %u)\n", buf_size, packet_size+1);
677 D_IF_decode( s->state, amrData, data, _good_frame);
682 static int amr_wb_decode_close(AVCodecContext * avctx)
684 AMRWBContext *s = avctx->priv_data;
690 AVCodec libamr_wb_decoder =
695 sizeof(AMRWBContext),
700 .long_name = "libamr-wb Adaptive Multi-Rate (AMR) Wide-Band",
703 AVCodec libamr_wb_encoder =
708 sizeof(AMRWBContext),
713 .long_name = "libamr-wb Adaptive Multi-Rate (AMR) Wide-Band",
716 #endif //CONFIG_LIBAMR_WB