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 #ifdef HAVE_LIBAVCODEC_AVCODEC_H
39 # include <libavcodec/avcodec.h>
40 #elif defined(HAVE_FFMPEG_AVCODEC_H)
41 # include <ffmpeg/avcodec.h>
48 /*****************************************************************************
49 * decoder_sys_t : decoder descriptor
50 *****************************************************************************/
55 /* Temporary buffer for libavcodec */
62 audio_sample_format_t aout_format;
76 int pi_extraction[AOUT_CHAN_MAX];
77 int i_previous_channels;
78 int64_t i_previous_layout;
81 #define BLOCK_FLAG_PRIVATE_REALLOCATED (1 << BLOCK_FLAG_PRIVATE_SHIFT)
83 static void SetupOutputFormat( decoder_t *p_dec, bool b_trust );
85 static void InitDecoderConfig( decoder_t *p_dec, AVCodecContext *p_context )
87 if( p_dec->fmt_in.i_extra > 0 )
89 const uint8_t * const p_src = p_dec->fmt_in.p_extra;
93 if( p_dec->fmt_in.i_codec == VLC_CODEC_FLAC )
96 i_size = p_dec->fmt_in.i_extra - 8;
98 else if( p_dec->fmt_in.i_codec == VLC_CODEC_ALAC )
100 static const uint8_t p_pattern[] = { 0, 0, 0, 36, 'a', 'l', 'a', 'c' };
101 /* Find alac atom XXX it is a bit ugly */
102 for( i_offset = 0; i_offset < p_dec->fmt_in.i_extra - sizeof(p_pattern); i_offset++ )
104 if( !memcmp( &p_src[i_offset], p_pattern, sizeof(p_pattern) ) )
107 i_size = __MIN( p_dec->fmt_in.i_extra - i_offset, 36 );
114 i_size = p_dec->fmt_in.i_extra;
119 p_context->extradata =
120 malloc( i_size + FF_INPUT_BUFFER_PADDING_SIZE );
121 if( p_context->extradata )
123 uint8_t *p_dst = p_context->extradata;
125 p_context->extradata_size = i_size;
127 memcpy( &p_dst[0], &p_src[i_offset], i_size );
128 memset( &p_dst[i_size], 0, FF_INPUT_BUFFER_PADDING_SIZE );
134 p_context->extradata_size = 0;
135 p_context->extradata = NULL;
139 /*****************************************************************************
140 * InitAudioDec: initialize audio decoder
141 *****************************************************************************
142 * The ffmpeg codec will be opened, some memory allocated.
143 *****************************************************************************/
144 int InitAudioDec( decoder_t *p_dec, AVCodecContext *p_context,
145 AVCodec *p_codec, int i_codec_id, const char *psz_namecodec )
147 decoder_sys_t *p_sys;
149 /* Allocate the memory needed to store the decoder's structure */
150 if( ( p_dec->p_sys = p_sys = malloc(sizeof(*p_sys)) ) == NULL )
155 p_codec->type = CODEC_TYPE_AUDIO;
156 p_context->codec_type = CODEC_TYPE_AUDIO;
157 p_context->codec_id = i_codec_id;
158 p_sys->p_context = p_context;
159 p_sys->p_codec = p_codec;
160 p_sys->i_codec_id = i_codec_id;
161 p_sys->psz_namecodec = psz_namecodec;
162 p_sys->b_delayed_open = true;
164 // Initialize decoder extradata
165 InitDecoderConfig( p_dec, p_context);
167 /* ***** Open the codec ***** */
168 if( ffmpeg_OpenCodec( p_dec ) < 0 )
170 msg_Err( p_dec, "cannot open codec (%s)", p_sys->psz_namecodec );
171 free( p_sys->p_context->extradata );
178 case CODEC_ID_WAVPACK:
179 p_sys->i_output_max = 8 * sizeof(int32_t) * 131072;
182 p_sys->i_output_max = p_sys->p_context->channels * sizeof(int32_t) * p_sys->p_context->sample_rate * 2;
185 p_sys->i_output_max = 8 * sizeof(int32_t) * 65535;
187 #if LIBAVCODEC_VERSION_INT >= AV_VERSION_INT( 52, 35, 0 )
188 case CODEC_ID_WMAPRO:
189 p_sys->i_output_max = 8 * sizeof(float) * 6144; /* (1 << 12) * 3/2 */
193 p_sys->i_output_max = 0;
196 if( p_sys->i_output_max < AVCODEC_MAX_AUDIO_FRAME_SIZE )
197 p_sys->i_output_max = AVCODEC_MAX_AUDIO_FRAME_SIZE;
198 msg_Dbg( p_dec, "Using %d bytes output buffer", p_sys->i_output_max );
199 p_sys->p_output = av_malloc( p_sys->i_output_max );
201 p_sys->p_samples = NULL;
202 p_sys->i_samples = 0;
203 p_sys->i_reject_count = 0;
204 p_sys->b_extract = false;
205 p_sys->i_previous_channels = 0;
206 p_sys->i_previous_layout = 0;
209 p_dec->fmt_out.i_cat = AUDIO_ES;
210 /* Try to set as much information as possible but do not trust it */
211 SetupOutputFormat( p_dec, false );
213 date_Set( &p_sys->end_date, 0 );
214 if( p_dec->fmt_out.audio.i_rate )
215 date_Init( &p_sys->end_date, p_dec->fmt_out.audio.i_rate, 1 );
216 else if( p_dec->fmt_in.audio.i_rate )
217 date_Init( &p_sys->end_date, p_dec->fmt_in.audio.i_rate, 1 );
222 /*****************************************************************************
223 * SplitBuffer: Needed because aout really doesn't like big audio chunk and
224 * wma produces easily > 30000 samples...
225 *****************************************************************************/
226 static aout_buffer_t *SplitBuffer( decoder_t *p_dec )
228 decoder_sys_t *p_sys = p_dec->p_sys;
229 int i_samples = __MIN( p_sys->i_samples, 4096 );
230 aout_buffer_t *p_buffer;
232 if( i_samples == 0 ) return NULL;
234 if( ( p_buffer = decoder_NewAudioBuffer( p_dec, i_samples ) ) == NULL )
237 p_buffer->i_pts = date_Get( &p_sys->end_date );
238 p_buffer->i_length = date_Increment( &p_sys->end_date, i_samples )
241 if( p_sys->b_extract )
242 aout_ChannelExtract( p_buffer->p_buffer, p_dec->fmt_out.audio.i_channels,
243 p_sys->p_samples, p_sys->p_context->channels, i_samples,
244 p_sys->pi_extraction, p_dec->fmt_out.audio.i_bitspersample );
246 memcpy( p_buffer->p_buffer, p_sys->p_samples, p_buffer->i_buffer );
248 p_sys->p_samples += i_samples * p_sys->p_context->channels * ( p_dec->fmt_out.audio.i_bitspersample / 8 );
249 p_sys->i_samples -= i_samples;
254 /*****************************************************************************
255 * DecodeAudio: Called to decode one frame
256 *****************************************************************************/
257 aout_buffer_t * DecodeAudio ( decoder_t *p_dec, block_t **pp_block )
259 decoder_sys_t *p_sys = p_dec->p_sys;
260 int i_used, i_output;
261 aout_buffer_t *p_buffer;
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 i_used = avcodec_decode_audio2( p_sys->p_context,
335 (int16_t*)p_sys->p_output, &i_output,
336 p_block->p_buffer, p_block->i_buffer );
338 if( i_used < 0 || i_output < 0 )
341 msg_Warn( p_dec, "cannot decode one frame (%zu bytes)",
344 block_Release( p_block );
347 else if( (size_t)i_used > p_block->i_buffer )
349 i_used = p_block->i_buffer;
352 p_block->i_buffer -= i_used;
353 p_block->p_buffer += i_used;
355 } while( p_block->i_buffer > 0 && i_output <= 0 );
357 if( p_sys->p_context->channels <= 0 || p_sys->p_context->channels > 8 ||
358 p_sys->p_context->sample_rate <= 0 )
360 msg_Warn( p_dec, "invalid audio properties channels count %d, sample rate %d",
361 p_sys->p_context->channels, p_sys->p_context->sample_rate );
362 block_Release( p_block );
366 if( p_dec->fmt_out.audio.i_rate != (unsigned int)p_sys->p_context->sample_rate )
368 date_Init( &p_sys->end_date, p_sys->p_context->sample_rate, 1 );
369 date_Set( &p_sys->end_date, p_block->i_pts );
372 /* **** Set audio output parameters **** */
373 SetupOutputFormat( p_dec, true );
375 if( p_block->i_pts != 0 &&
376 p_block->i_pts != date_Get( &p_sys->end_date ) )
378 date_Set( &p_sys->end_date, p_block->i_pts );
382 /* **** Now we can output these samples **** */
383 p_sys->i_samples = i_output / (p_dec->fmt_out.audio.i_bitspersample / 8) / p_sys->p_context->channels;
384 p_sys->p_samples = p_sys->p_output;
386 /* Silent unwanted samples */
387 if( p_sys->i_reject_count > 0 )
389 memset( p_sys->p_output, 0, i_output );
390 p_sys->i_reject_count--;
393 p_buffer = SplitBuffer( p_dec );
394 if( !p_buffer ) block_Release( p_block );
398 /*****************************************************************************
399 * EndAudioDec: audio decoder destruction
400 *****************************************************************************/
401 void EndAudioDec( decoder_t *p_dec )
403 decoder_sys_t *p_sys = p_dec->p_sys;
405 av_free( p_sys->p_output );
408 /*****************************************************************************
410 *****************************************************************************/
412 void GetVlcAudioFormat( vlc_fourcc_t *pi_codec, unsigned *pi_bits, int i_sample_fmt )
414 switch( i_sample_fmt )
417 *pi_codec = VLC_CODEC_U8;
421 *pi_codec = VLC_CODEC_S32N;
425 *pi_codec = VLC_CODEC_FL32;
429 *pi_codec = VLC_CODEC_FL64;
435 *pi_codec = VLC_CODEC_S16N;
441 static const uint64_t pi_channels_map[][2] =
443 { CH_FRONT_LEFT, AOUT_CHAN_LEFT },
444 { CH_FRONT_RIGHT, AOUT_CHAN_RIGHT },
445 { CH_FRONT_CENTER, AOUT_CHAN_CENTER },
446 { CH_LOW_FREQUENCY, AOUT_CHAN_LFE },
447 { CH_BACK_LEFT, AOUT_CHAN_REARLEFT },
448 { CH_BACK_RIGHT, AOUT_CHAN_REARRIGHT },
449 { CH_FRONT_LEFT_OF_CENTER, 0 },
450 { CH_FRONT_RIGHT_OF_CENTER, 0 },
451 { CH_BACK_CENTER, AOUT_CHAN_REARCENTER },
452 { CH_SIDE_LEFT, AOUT_CHAN_MIDDLELEFT },
453 { CH_SIDE_RIGHT, AOUT_CHAN_MIDDLERIGHT },
454 { CH_TOP_CENTER, 0 },
455 { CH_TOP_FRONT_LEFT, 0 },
456 { CH_TOP_FRONT_CENTER, 0 },
457 { CH_TOP_FRONT_RIGHT, 0 },
458 { CH_TOP_BACK_LEFT, 0 },
459 { CH_TOP_BACK_CENTER, 0 },
460 { CH_TOP_BACK_RIGHT, 0 },
461 { CH_STEREO_LEFT, 0 },
462 { CH_STEREO_RIGHT, 0 },
465 static void SetupOutputFormat( decoder_t *p_dec, bool b_trust )
467 decoder_sys_t *p_sys = p_dec->p_sys;
469 GetVlcAudioFormat( &p_dec->fmt_out.i_codec,
470 &p_dec->fmt_out.audio.i_bitspersample,
471 p_sys->p_context->sample_fmt );
472 p_dec->fmt_out.audio.i_rate = p_sys->p_context->sample_rate;
475 if( p_sys->i_previous_channels == p_sys->p_context->channels &&
476 p_sys->i_previous_layout == p_sys->p_context->channel_layout )
480 p_sys->i_previous_channels = p_sys->p_context->channels;
481 p_sys->i_previous_layout = p_sys->p_context->channel_layout;
485 * FIXME should we use fmt_in.audio.i_physical_channels or not ?
487 const unsigned i_order_max = 8 * sizeof(p_sys->p_context->channel_layout);
488 uint32_t pi_order_src[i_order_max];
489 int i_channels_src = 0;
491 if( p_sys->p_context->channel_layout )
493 for( unsigned i = 0; i < sizeof(pi_channels_map)/sizeof(*pi_channels_map); i++ )
495 if( p_sys->p_context->channel_layout & pi_channels_map[i][0] )
496 pi_order_src[i_channels_src++] = pi_channels_map[i][1];
501 /* Create default order */
503 msg_Warn( p_dec, "Physical channel configuration not set : guessing" );
504 for( unsigned int i = 0; i < __MIN( i_order_max, (unsigned)p_sys->p_context->channels ); i++ )
506 if( i < sizeof(pi_channels_map)/sizeof(*pi_channels_map) )
507 pi_order_src[i_channels_src++] = pi_channels_map[i][1];
510 if( i_channels_src != p_sys->p_context->channels && b_trust )
511 msg_Err( p_dec, "Channel layout not understood" );
513 uint32_t i_layout_dst;
515 p_sys->b_extract = aout_CheckChannelExtraction( p_sys->pi_extraction,
516 &i_layout_dst, &i_channels_dst,
517 NULL, pi_order_src, i_channels_src );
518 if( i_channels_dst != i_channels_src && b_trust )
519 msg_Warn( p_dec, "%d channels are dropped", i_channels_src - i_channels_dst );
521 p_dec->fmt_out.audio.i_physical_channels =
522 p_dec->fmt_out.audio.i_original_channels = i_layout_dst;
523 p_dec->fmt_out.audio.i_channels = i_channels_dst;