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;
137 avctx->sample_fmt = SAMPLE_FMT_S16;
140 #ifdef CONFIG_LIBAMR_NB_FIXED
141 /* fixed point version*/
142 /* frame size in serial bitstream file (frame type + serial stream + flags) */
143 #define SERIAL_FRAMESIZE (1+MAX_SERIAL_SIZE+5)
145 typedef struct AMRContext {
147 Speech_Decode_FrameState *speech_decoder_state;
148 enum RXFrameType rx_type;
151 Word16 reset_flag_old;
154 Speech_Encode_FrameState *enstate;
155 sid_syncState *sidstate;
156 enum TXFrameType tx_frametype;
159 static int amr_nb_decode_init(AVCodecContext * avctx)
161 AMRContext *s = avctx->priv_data;
164 s->speech_decoder_state=NULL;
165 s->rx_type = (enum RXFrameType)0;
166 s->mode= (enum Mode)0;
170 if(Speech_Decode_Frame_init(&s->speech_decoder_state, "Decoder"))
172 av_log(avctx, AV_LOG_ERROR, "Speech_Decode_Frame_init error\n");
176 amr_decode_fix_avctx(avctx);
178 if(avctx->channels > 1)
180 av_log(avctx, AV_LOG_ERROR, "amr_nb: multichannel decoding not supported\n");
187 static int amr_nb_encode_init(AVCodecContext * avctx)
189 AMRContext *s = avctx->priv_data;
192 s->speech_decoder_state=NULL;
193 s->rx_type = (enum RXFrameType)0;
194 s->mode= (enum Mode)0;
198 if(avctx->sample_rate!=8000)
200 av_log(avctx, AV_LOG_ERROR, "Only 8000Hz sample rate supported\n");
204 if(avctx->channels!=1)
206 av_log(avctx, AV_LOG_ERROR, "Only mono supported\n");
210 avctx->frame_size=160;
211 avctx->coded_frame= avcodec_alloc_frame();
213 if(Speech_Encode_Frame_init(&s->enstate, 0, "encoder") || sid_sync_init (&s->sidstate))
215 av_log(avctx, AV_LOG_ERROR, "Speech_Encode_Frame_init error\n");
219 if((s->enc_bitrate=getBitrateMode(avctx->bit_rate))<0)
221 av_log(avctx, AV_LOG_ERROR, nb_bitrate_unsupported);
228 static int amr_nb_encode_close(AVCodecContext * avctx)
230 AMRContext *s = avctx->priv_data;
232 Speech_Encode_Frame_exit(&s->enstate);
233 sid_sync_exit (&s->sidstate);
234 av_freep(&avctx->coded_frame);
238 static int amr_nb_decode_close(AVCodecContext * avctx)
240 AMRContext *s = avctx->priv_data;
242 Speech_Decode_Frame_exit(&s->speech_decoder_state);
246 static int amr_nb_decode_frame(AVCodecContext * avctx,
247 void *data, int *data_size,
248 const uint8_t * buf, int buf_size)
250 AMRContext *s = avctx->priv_data;
251 const uint8_t*amrData=buf;
254 Word16 serial[SERIAL_FRAMESIZE]; /* coded bits */
257 static Word16 packed_size[16] = {12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0};
260 //printf("amr_decode_frame data_size=%i buf=0x%X buf_size=%d frameCount=%d!!\n",*data_size,buf,buf_size,s->frameCount);
265 /* read rest of the frame based on ToC byte */
266 q = (toc >> 2) & 0x01;
267 ft = (toc >> 3) & 0x0F;
269 //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]);
273 packed_bits=amrData+offset;
275 offset+=packed_size[ft];
277 //Unsort and unpack bits
278 s->rx_type = UnpackBits(q, ft, packed_bits, &s->mode, &serial[1]);
280 //We have a new frame
283 if (s->rx_type == RX_NO_DATA)
285 s->mode = s->speech_decoder_state->prev_mode;
288 s->speech_decoder_state->prev_mode = s->mode;
291 /* if homed: check if this frame is another homing frame */
292 if (s->reset_flag_old == 1)
294 /* only check until end of first subframe */
295 s->reset_flag = decoder_homing_frame_test_first(&serial[1], s->mode);
297 /* produce encoder homing frame if homed & input=decoder homing frame */
298 if ((s->reset_flag != 0) && (s->reset_flag_old != 0))
300 for (i = 0; i < L_FRAME; i++)
308 Speech_Decode_Frame(s->speech_decoder_state, s->mode, &serial[1], s->rx_type, synth);
311 //Each AMR-frame results in 160 16-bit samples
314 /* if not homed: check whether current frame is a homing frame */
315 if (s->reset_flag_old == 0)
317 /* check whole frame */
318 s->reset_flag = decoder_homing_frame_test(&serial[1], s->mode);
320 /* reset decoder if current frame is a homing frame */
321 if (s->reset_flag != 0)
323 Speech_Decode_Frame_reset(s->speech_decoder_state);
325 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};
335 AMRContext *s = avctx->priv_data;
338 s->reset_flag = encoder_homing_frame_test(data);
340 Speech_Encode_Frame(s->enstate, s->enc_bitrate, data, &serial_data[1], &s->mode);
342 /* add frame type and mode */
343 sid_sync (s->sidstate, s->mode, &s->tx_frametype);
345 written = PackBits(s->mode, s->enc_bitrate, s->tx_frametype, &serial_data[1], frame);
347 if (s->reset_flag != 0)
349 Speech_Encode_Frame_reset(s->enstate);
350 sid_sync_reset(s->sidstate);
356 #elif defined(CONFIG_LIBAMR_NB) /* Float point version*/
358 typedef struct AMRContext {
365 static int amr_nb_decode_init(AVCodecContext * avctx)
367 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;
394 if(avctx->sample_rate!=8000)
396 av_log(avctx, AV_LOG_ERROR, "Only 8000Hz sample rate supported\n");
400 if(avctx->channels!=1)
402 av_log(avctx, AV_LOG_ERROR, "Only mono supported\n");
406 avctx->frame_size=160;
407 avctx->coded_frame= avcodec_alloc_frame();
409 s->enstate=Encoder_Interface_init(0);
412 av_log(avctx, AV_LOG_ERROR, "Encoder_Interface_init error\n");
416 if((s->enc_bitrate=getBitrateMode(avctx->bit_rate))<0)
418 av_log(avctx, AV_LOG_ERROR, nb_bitrate_unsupported);
425 static int amr_nb_decode_close(AVCodecContext * avctx)
427 AMRContext *s = avctx->priv_data;
429 Decoder_Interface_exit(s->decState);
433 static int amr_nb_encode_close(AVCodecContext * avctx)
435 AMRContext *s = avctx->priv_data;
437 Encoder_Interface_exit(s->enstate);
438 av_freep(&avctx->coded_frame);
442 static int amr_nb_decode_frame(AVCodecContext * avctx,
443 void *data, int *data_size,
444 const uint8_t * buf, int buf_size)
446 AMRContext *s = avctx->priv_data;
447 const uint8_t*amrData=buf;
448 static const uint8_t block_size[16]={ 12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0 };
452 /* av_log(NULL,AV_LOG_DEBUG,"amr_decode_frame buf=%p buf_size=%d frameCount=%d!!\n",buf,buf_size,s->frameCount); */
454 dec_mode = (buf[0] >> 3) & 0x000F;
455 packet_size = block_size[dec_mode]+1;
457 if(packet_size > buf_size) {
458 av_log(avctx, AV_LOG_ERROR, "amr frame too short (%u, should be %u)\n", buf_size, packet_size);
463 /* 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]); */
465 Decoder_Interface_Decode(s->decState, amrData, data, 0);
471 static int amr_nb_encode_frame(AVCodecContext *avctx,
472 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
474 AMRContext *s = avctx->priv_data;
477 if((s->enc_bitrate=getBitrateMode(avctx->bit_rate))<0)
479 av_log(avctx, AV_LOG_ERROR, nb_bitrate_unsupported);
483 written = Encoder_Interface_Encode(s->enstate,
488 /* 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] ); */
495 #if defined(CONFIG_LIBAMR_NB) || defined(CONFIG_LIBAMR_NB_FIXED)
497 AVCodec libamr_nb_decoder =
507 .long_name = NULL_IF_CONFIG_SMALL("libamr-nb Adaptive Multi-Rate (AMR) Narrow-Band"),
510 AVCodec libamr_nb_encoder =
520 .sample_fmts = (enum SampleFormat[]){SAMPLE_FMT_S16,SAMPLE_FMT_NONE},
521 .long_name = NULL_IF_CONFIG_SMALL("libamr-nb Adaptive Multi-Rate (AMR) Narrow-Band"),
526 /* -----------AMR wideband ------------*/
527 #ifdef CONFIG_LIBAMR_WB
530 //To avoid duplicate typedefs from typedef in amr-nb
534 #include <amrwb/enc_if.h>
535 #include <amrwb/dec_if.h>
536 #include <amrwb/if_rom.h>
538 /* Common code for fixed and float version*/
539 typedef struct AMRWB_bitrates
545 static int getWBBitrateMode(int bitrate)
547 /* make the correspondance between bitrate and mode */
548 AMRWB_bitrates rates[]={ {6600,0},
562 if(rates[i].rate==bitrate)
564 return rates[i].mode;
567 /* no bitrate matching, return an error */
572 typedef struct AMRWBContext {
579 static int amr_wb_encode_init(AVCodecContext * avctx)
581 AMRWBContext *s = avctx->priv_data;
585 if(avctx->sample_rate!=16000)
587 av_log(avctx, AV_LOG_ERROR, "Only 16000Hz sample rate supported\n");
591 if(avctx->channels!=1)
593 av_log(avctx, AV_LOG_ERROR, "Only mono supported\n");
597 if((s->mode=getWBBitrateMode(avctx->bit_rate))<0)
599 av_log(avctx, AV_LOG_ERROR, wb_bitrate_unsupported);
603 avctx->frame_size=320;
604 avctx->coded_frame= avcodec_alloc_frame();
606 s->state = E_IF_init();
612 static int amr_wb_encode_close(AVCodecContext * avctx)
614 AMRWBContext *s = avctx->priv_data;
617 av_freep(&avctx->coded_frame);
622 static int amr_wb_encode_frame(AVCodecContext *avctx,
623 unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
625 AMRWBContext *s = avctx->priv_data;
628 if((s->mode=getWBBitrateMode(avctx->bit_rate))<0)
630 av_log(avctx, AV_LOG_ERROR, wb_bitrate_unsupported);
633 size = E_IF_encode(s->state, s->mode, data, frame, s->allow_dtx);
637 static int amr_wb_decode_init(AVCodecContext * avctx)
639 AMRWBContext *s = avctx->priv_data;
642 s->state = D_IF_init();
644 amr_decode_fix_avctx(avctx);
646 if(avctx->channels > 1)
648 av_log(avctx, AV_LOG_ERROR, "amr_wb: multichannel decoding not supported\n");
655 static int amr_wb_decode_frame(AVCodecContext * avctx,
656 void *data, int *data_size,
657 const uint8_t * buf, int buf_size)
659 AMRWBContext *s = avctx->priv_data;
660 const uint8_t*amrData=buf;
663 static const uint8_t block_size[16] = {18, 23, 33, 37, 41, 47, 51, 59, 61, 6, 6, 0, 0, 0, 1, 1};
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 = avctx->priv_data;
692 AVCodec libamr_wb_decoder =
697 sizeof(AMRWBContext),
702 .long_name = NULL_IF_CONFIG_SMALL("libamr-wb Adaptive Multi-Rate (AMR) Wide-Band"),
705 AVCodec libamr_wb_encoder =
710 sizeof(AMRWBContext),
715 .sample_fmts = (enum SampleFormat[]){SAMPLE_FMT_S16,SAMPLE_FMT_NONE},
716 .long_name = NULL_IF_CONFIG_SMALL("libamr-wb Adaptive Multi-Rate (AMR) Wide-Band"),
719 #endif //CONFIG_LIBAMR_WB