1 /*****************************************************************************
2 * audio.c: audio decoder using ffmpeg library
3 *****************************************************************************
4 * Copyright (C) 1999-2003 the VideoLAN team
7 * Authors: Laurent Aimar <fenrir@via.ecp.fr>
8 * Gildas Bazin <gbazin@videolan.org>
10 * This program is free software; you can redistribute it and/or modify
11 * it under the terms of the GNU General Public License as published by
12 * the Free Software Foundation; either version 2 of the License, or
13 * (at your option) any later version.
15 * This program is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 * GNU General Public License for more details.
20 * You should have received a copy of the GNU General Public License
21 * along with this program; if not, write to the Free Software
22 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
23 *****************************************************************************/
25 /*****************************************************************************
27 *****************************************************************************/
32 #include <vlc_common.h>
34 #include <vlc_codec.h>
35 #include <vlc_avcodec.h>
38 #include <libavcodec/avcodec.h>
39 #include <libavutil/mem.h>
41 #if LIBAVUTIL_VERSION_INT >= ((50<<16)+(38<<8)+0)
42 # include "libavutil/audioconvert.h"
47 /*****************************************************************************
48 * decoder_sys_t : decoder descriptor
49 *****************************************************************************/
52 AVCODEC_COMMON_MEMBERS
54 /* Temporary buffer for libavcodec */
61 audio_sample_format_t aout_format;
75 int pi_extraction[AOUT_CHAN_MAX];
76 int i_previous_channels;
77 int64_t i_previous_layout;
80 #define BLOCK_FLAG_PRIVATE_REALLOCATED (1 << BLOCK_FLAG_PRIVATE_SHIFT)
82 static void SetupOutputFormat( decoder_t *p_dec, bool b_trust );
84 static void InitDecoderConfig( decoder_t *p_dec, AVCodecContext *p_context )
86 if( p_dec->fmt_in.i_extra > 0 )
88 const uint8_t * const p_src = p_dec->fmt_in.p_extra;
92 if( p_dec->fmt_in.i_codec == VLC_CODEC_FLAC )
95 i_size = p_dec->fmt_in.i_extra - 8;
97 else if( p_dec->fmt_in.i_codec == VLC_CODEC_ALAC )
99 static const uint8_t p_pattern[] = { 0, 0, 0, 36, 'a', 'l', 'a', 'c' };
100 /* Find alac atom XXX it is a bit ugly */
101 for( i_offset = 0; i_offset < p_dec->fmt_in.i_extra - sizeof(p_pattern); i_offset++ )
103 if( !memcmp( &p_src[i_offset], p_pattern, sizeof(p_pattern) ) )
106 i_size = __MIN( p_dec->fmt_in.i_extra - i_offset, 36 );
113 i_size = p_dec->fmt_in.i_extra;
118 p_context->extradata =
119 malloc( i_size + FF_INPUT_BUFFER_PADDING_SIZE );
120 if( p_context->extradata )
122 uint8_t *p_dst = p_context->extradata;
124 p_context->extradata_size = i_size;
126 memcpy( &p_dst[0], &p_src[i_offset], i_size );
127 memset( &p_dst[i_size], 0, FF_INPUT_BUFFER_PADDING_SIZE );
133 p_context->extradata_size = 0;
134 p_context->extradata = NULL;
138 /*****************************************************************************
139 * InitAudioDec: initialize audio decoder
140 *****************************************************************************
141 * The ffmpeg codec will be opened, some memory allocated.
142 *****************************************************************************/
143 int InitAudioDec( decoder_t *p_dec, AVCodecContext *p_context,
144 AVCodec *p_codec, int i_codec_id, const char *psz_namecodec )
146 decoder_sys_t *p_sys;
148 /* Allocate the memory needed to store the decoder's structure */
149 if( ( p_dec->p_sys = p_sys = malloc(sizeof(*p_sys)) ) == NULL )
154 p_codec->type = AVMEDIA_TYPE_AUDIO;
155 p_context->codec_type = AVMEDIA_TYPE_AUDIO;
156 p_context->codec_id = i_codec_id;
157 p_sys->p_context = p_context;
158 p_sys->p_codec = p_codec;
159 p_sys->i_codec_id = i_codec_id;
160 p_sys->psz_namecodec = psz_namecodec;
161 p_sys->b_delayed_open = true;
163 // Initialize decoder extradata
164 InitDecoderConfig( p_dec, p_context);
166 /* ***** Open the codec ***** */
167 if( ffmpeg_OpenCodec( p_dec ) < 0 )
169 msg_Err( p_dec, "cannot open codec (%s)", p_sys->psz_namecodec );
170 free( p_sys->p_context->extradata );
177 case CODEC_ID_WAVPACK:
178 p_sys->i_output_max = 8 * sizeof(int32_t) * 131072;
181 p_sys->i_output_max = p_sys->p_context->channels * sizeof(int32_t) * p_sys->p_context->sample_rate * 2;
184 p_sys->i_output_max = 8 * sizeof(int32_t) * 65535;
186 #if LIBAVCODEC_VERSION_INT >= AV_VERSION_INT( 52, 35, 0 )
187 case CODEC_ID_WMAPRO:
188 p_sys->i_output_max = 8 * sizeof(float) * 6144; /* (1 << 12) * 3/2 */
192 p_sys->i_output_max = 0;
195 if( p_sys->i_output_max < AVCODEC_MAX_AUDIO_FRAME_SIZE )
196 p_sys->i_output_max = AVCODEC_MAX_AUDIO_FRAME_SIZE;
197 msg_Dbg( p_dec, "Using %d bytes output buffer", p_sys->i_output_max );
198 p_sys->p_output = av_malloc( p_sys->i_output_max );
200 p_sys->p_samples = NULL;
201 p_sys->i_samples = 0;
202 p_sys->i_reject_count = 0;
203 p_sys->b_extract = false;
204 p_sys->i_previous_channels = 0;
205 p_sys->i_previous_layout = 0;
208 p_dec->fmt_out.i_cat = AUDIO_ES;
209 /* Try to set as much information as possible but do not trust it */
210 SetupOutputFormat( p_dec, false );
212 date_Set( &p_sys->end_date, 0 );
213 if( p_dec->fmt_out.audio.i_rate )
214 date_Init( &p_sys->end_date, p_dec->fmt_out.audio.i_rate, 1 );
215 else if( p_dec->fmt_in.audio.i_rate )
216 date_Init( &p_sys->end_date, p_dec->fmt_in.audio.i_rate, 1 );
221 /*****************************************************************************
222 * SplitBuffer: Needed because aout really doesn't like big audio chunk and
223 * wma produces easily > 30000 samples...
224 *****************************************************************************/
225 static block_t *SplitBuffer( decoder_t *p_dec )
227 decoder_sys_t *p_sys = p_dec->p_sys;
228 int i_samples = __MIN( p_sys->i_samples, 4096 );
231 if( i_samples == 0 ) return NULL;
233 if( ( p_buffer = decoder_NewAudioBuffer( p_dec, i_samples ) ) == NULL )
236 p_buffer->i_pts = date_Get( &p_sys->end_date );
237 p_buffer->i_length = date_Increment( &p_sys->end_date, i_samples )
240 if( p_sys->b_extract )
241 aout_ChannelExtract( p_buffer->p_buffer, p_dec->fmt_out.audio.i_channels,
242 p_sys->p_samples, p_sys->p_context->channels, i_samples,
243 p_sys->pi_extraction, p_dec->fmt_out.audio.i_bitspersample );
245 memcpy( p_buffer->p_buffer, p_sys->p_samples, p_buffer->i_buffer );
247 p_sys->p_samples += i_samples * p_sys->p_context->channels * ( p_dec->fmt_out.audio.i_bitspersample / 8 );
248 p_sys->i_samples -= i_samples;
253 /*****************************************************************************
254 * DecodeAudio: Called to decode one frame
255 *****************************************************************************/
256 block_t * DecodeAudio ( decoder_t *p_dec, block_t **pp_block )
258 decoder_sys_t *p_sys = p_dec->p_sys;
259 int i_used, i_output;
264 if( !pp_block || !*pp_block ) return NULL;
268 if( !p_sys->p_context->extradata_size && p_dec->fmt_in.i_extra &&
269 p_sys->b_delayed_open)
271 InitDecoderConfig( p_dec, p_sys->p_context);
272 if( ffmpeg_OpenCodec( p_dec ) )
273 msg_Err( p_dec, "Cannot open decoder %s", p_sys->psz_namecodec );
275 if( p_sys->b_delayed_open )
277 block_Release( p_block );
281 if( p_block->i_flags & (BLOCK_FLAG_DISCONTINUITY|BLOCK_FLAG_CORRUPTED) )
283 block_Release( p_block );
284 avcodec_flush_buffers( p_sys->p_context );
285 p_sys->i_samples = 0;
286 date_Set( &p_sys->end_date, 0 );
288 if( p_sys->i_codec_id == CODEC_ID_MP2 || p_sys->i_codec_id == CODEC_ID_MP3 )
289 p_sys->i_reject_count = 3;
293 if( p_sys->i_samples > 0 )
296 p_buffer = SplitBuffer( p_dec );
297 if( !p_buffer ) block_Release( p_block );
301 if( !date_Get( &p_sys->end_date ) && !p_block->i_pts )
303 /* We've just started the stream, wait for the first PTS. */
304 block_Release( p_block );
308 if( p_block->i_buffer <= 0 )
310 block_Release( p_block );
314 if( (p_block->i_flags & BLOCK_FLAG_PRIVATE_REALLOCATED) == 0 )
316 *pp_block = p_block = block_Realloc( p_block, 0, p_block->i_buffer + FF_INPUT_BUFFER_PADDING_SIZE );
319 p_block->i_buffer -= FF_INPUT_BUFFER_PADDING_SIZE;
320 memset( &p_block->p_buffer[p_block->i_buffer], 0, FF_INPUT_BUFFER_PADDING_SIZE );
322 p_block->i_flags |= BLOCK_FLAG_PRIVATE_REALLOCATED;
327 i_output = __MAX( p_block->i_buffer, p_sys->i_output_max );
328 if( i_output > p_sys->i_output_max )
330 /* Grow output buffer if necessary (eg. for PCM data) */
331 p_sys->p_output = av_realloc( p_sys->p_output, i_output );
334 av_init_packet( &pkt );
335 pkt.data = p_block->p_buffer;
336 pkt.size = p_block->i_buffer;
337 i_used = avcodec_decode_audio3( p_sys->p_context,
338 (int16_t*)p_sys->p_output, &i_output,
341 if( i_used < 0 || i_output < 0 )
344 msg_Warn( p_dec, "cannot decode one frame (%zu bytes)",
347 block_Release( p_block );
350 else if( (size_t)i_used > p_block->i_buffer )
352 i_used = p_block->i_buffer;
355 p_block->i_buffer -= i_used;
356 p_block->p_buffer += i_used;
358 } while( p_block->i_buffer > 0 && i_output <= 0 );
360 if( p_sys->p_context->channels <= 0 || p_sys->p_context->channels > 8 ||
361 p_sys->p_context->sample_rate <= 0 )
363 msg_Warn( p_dec, "invalid audio properties channels count %d, sample rate %d",
364 p_sys->p_context->channels, p_sys->p_context->sample_rate );
365 block_Release( p_block );
369 if( p_dec->fmt_out.audio.i_rate != (unsigned int)p_sys->p_context->sample_rate )
371 date_Init( &p_sys->end_date, p_sys->p_context->sample_rate, 1 );
372 date_Set( &p_sys->end_date, p_block->i_pts );
375 /* **** Set audio output parameters **** */
376 SetupOutputFormat( p_dec, true );
378 if( p_block->i_pts != 0 &&
379 p_block->i_pts != date_Get( &p_sys->end_date ) )
381 date_Set( &p_sys->end_date, p_block->i_pts );
385 /* **** Now we can output these samples **** */
386 p_sys->i_samples = i_output / (p_dec->fmt_out.audio.i_bitspersample / 8) / p_sys->p_context->channels;
387 p_sys->p_samples = p_sys->p_output;
389 /* Silent unwanted samples */
390 if( p_sys->i_reject_count > 0 )
392 memset( p_sys->p_output, 0, i_output );
393 p_sys->i_reject_count--;
396 p_buffer = SplitBuffer( p_dec );
397 if( !p_buffer ) block_Release( p_block );
401 /*****************************************************************************
402 * EndAudioDec: audio decoder destruction
403 *****************************************************************************/
404 void EndAudioDec( decoder_t *p_dec )
406 decoder_sys_t *p_sys = p_dec->p_sys;
408 av_free( p_sys->p_output );
411 /*****************************************************************************
413 *****************************************************************************/
415 void GetVlcAudioFormat( vlc_fourcc_t *pi_codec, unsigned *pi_bits, int i_sample_fmt )
417 switch( i_sample_fmt )
419 case AV_SAMPLE_FMT_U8:
420 *pi_codec = VLC_CODEC_U8;
423 case AV_SAMPLE_FMT_S32:
424 *pi_codec = VLC_CODEC_S32N;
427 case AV_SAMPLE_FMT_FLT:
428 *pi_codec = VLC_CODEC_FL32;
431 case AV_SAMPLE_FMT_DBL:
432 *pi_codec = VLC_CODEC_FL64;
436 case AV_SAMPLE_FMT_S16:
438 *pi_codec = VLC_CODEC_S16N;
444 static const uint64_t pi_channels_map[][2] =
446 { AV_CH_FRONT_LEFT, AOUT_CHAN_LEFT },
447 { AV_CH_FRONT_RIGHT, AOUT_CHAN_RIGHT },
448 { AV_CH_FRONT_CENTER, AOUT_CHAN_CENTER },
449 { AV_CH_LOW_FREQUENCY, AOUT_CHAN_LFE },
450 { AV_CH_BACK_LEFT, AOUT_CHAN_REARLEFT },
451 { AV_CH_BACK_RIGHT, AOUT_CHAN_REARRIGHT },
452 { AV_CH_FRONT_LEFT_OF_CENTER, 0 },
453 { AV_CH_FRONT_RIGHT_OF_CENTER, 0 },
454 { AV_CH_BACK_CENTER, AOUT_CHAN_REARCENTER },
455 { AV_CH_SIDE_LEFT, AOUT_CHAN_MIDDLELEFT },
456 { AV_CH_SIDE_RIGHT, AOUT_CHAN_MIDDLERIGHT },
457 { AV_CH_TOP_CENTER, 0 },
458 { AV_CH_TOP_FRONT_LEFT, 0 },
459 { AV_CH_TOP_FRONT_CENTER, 0 },
460 { AV_CH_TOP_FRONT_RIGHT, 0 },
461 { AV_CH_TOP_BACK_LEFT, 0 },
462 { AV_CH_TOP_BACK_CENTER, 0 },
463 { AV_CH_TOP_BACK_RIGHT, 0 },
464 { AV_CH_STEREO_LEFT, 0 },
465 { AV_CH_STEREO_RIGHT, 0 },
468 static void SetupOutputFormat( decoder_t *p_dec, bool b_trust )
470 decoder_sys_t *p_sys = p_dec->p_sys;
472 GetVlcAudioFormat( &p_dec->fmt_out.i_codec,
473 &p_dec->fmt_out.audio.i_bitspersample,
474 p_sys->p_context->sample_fmt );
475 p_dec->fmt_out.audio.i_rate = p_sys->p_context->sample_rate;
478 if( p_sys->i_previous_channels == p_sys->p_context->channels &&
479 p_sys->i_previous_layout == p_sys->p_context->channel_layout )
483 p_sys->i_previous_channels = p_sys->p_context->channels;
484 p_sys->i_previous_layout = p_sys->p_context->channel_layout;
488 * FIXME should we use fmt_in.audio.i_physical_channels or not ?
490 const unsigned i_order_max = 8 * sizeof(p_sys->p_context->channel_layout);
491 uint32_t pi_order_src[i_order_max];
492 int i_channels_src = 0;
494 if( p_sys->p_context->channel_layout )
496 for( unsigned i = 0; i < sizeof(pi_channels_map)/sizeof(*pi_channels_map); i++ )
498 if( p_sys->p_context->channel_layout & pi_channels_map[i][0] )
499 pi_order_src[i_channels_src++] = pi_channels_map[i][1];
504 /* Create default order */
506 msg_Warn( p_dec, "Physical channel configuration not set : guessing" );
507 for( unsigned int i = 0; i < __MIN( i_order_max, (unsigned)p_sys->p_context->channels ); i++ )
509 if( i < sizeof(pi_channels_map)/sizeof(*pi_channels_map) )
510 pi_order_src[i_channels_src++] = pi_channels_map[i][1];
513 if( i_channels_src != p_sys->p_context->channels && b_trust )
514 msg_Err( p_dec, "Channel layout not understood" );
516 uint32_t i_layout_dst;
518 p_sys->b_extract = aout_CheckChannelExtraction( p_sys->pi_extraction,
519 &i_layout_dst, &i_channels_dst,
520 NULL, pi_order_src, i_channels_src );
521 if( i_channels_dst != i_channels_src && b_trust )
522 msg_Warn( p_dec, "%d channels are dropped", i_channels_src - i_channels_dst );
524 p_dec->fmt_out.audio.i_physical_channels =
525 p_dec->fmt_out.audio.i_original_channels = i_layout_dst;
526 p_dec->fmt_out.audio.i_channels = i_channels_dst;