1 /*****************************************************************************
2 * speex.c: speex decoder/packetizer/encoder module making use of libspeex.
3 *****************************************************************************
4 * Copyright (C) 2003-2009 the VideoLAN team
7 * Authors: Gildas Bazin <gbazin@videolan.org>
9 * This program is free software; you can redistribute it and/or modify
10 * it under the terms of the GNU General Public License as published by
11 * the Free Software Foundation; either version 2 of the License, or
12 * (at your option) any later version.
14 * This program is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 * GNU General Public License for more details.
19 * You should have received a copy of the GNU General Public License
20 * along with this program; if not, write to the Free Software
21 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
22 *****************************************************************************/
24 /*****************************************************************************
26 *****************************************************************************/
33 #include <vlc_common.h>
34 #include <vlc_plugin.h>
35 #include <vlc_input.h>
36 #include <vlc_codec.h>
38 #include <vlc_memory.h>
41 #include <speex/speex.h>
42 #include <speex/speex_header.h>
43 #include <speex/speex_stereo.h>
44 #include <speex/speex_callbacks.h>
48 /*****************************************************************************
50 *****************************************************************************/
51 static int OpenDecoder ( vlc_object_t * );
52 static int OpenPacketizer( vlc_object_t * );
53 static void CloseDecoder ( vlc_object_t * );
54 static int OpenEncoder ( vlc_object_t * );
55 static void CloseEncoder ( vlc_object_t * );
57 #define ENC_CFG_PREFIX "sout-speex-"
59 #define ENC_MODE_TEXT N_("Mode" )
60 #define ENC_MODE_LONGTEXT N_( \
61 "Enforce the mode of the encoder." )
63 #define ENC_QUALITY_TEXT N_("Encoding quality")
64 #define ENC_QUALITY_LONGTEXT N_( \
65 "Enforce a quality between 0 (low) and 10 (high)." )
67 #define ENC_COMPLEXITY_TEXT N_("Encoding complexity" )
68 #define ENC_COMPLEXITY_LONGTEXT N_( \
69 "Enforce the complexity of the encoder." )
71 #define ENC_MAXBITRATE_TEXT N_( "Maximal bitrate" )
72 #define ENC_MAXBITRATE_LONGTEXT N_( \
73 "Enforce the maximal VBR bitrate" )
75 #define ENC_CBR_TEXT N_( "CBR encoding" )
76 #define ENC_CBR_LONGTEXT N_( \
77 "Enforce a constant bitrate encoding (CBR) instead of default " \
78 "variable bitrate encoding (VBR)." )
80 #define ENC_VAD_TEXT N_( "Voice activity detection" )
81 #define ENC_VAD_LONGTEXT N_( \
82 "Enable voice activity detection (VAD). It is automatically " \
83 "activated in VBR mode." )
85 #define ENC_DTX_TEXT N_( "Discontinuous Transmission" )
86 #define ENC_DTX_LONGTEXT N_( \
87 "Enable discontinuous transmission (DTX)." )
89 static const int pi_enc_mode_values[] = { 0, 1, 2 };
90 static const char * const ppsz_enc_mode_descriptions[] = {
91 N_("Narrow-band (8kHz)"), N_("Wide-band (16kHz)"), N_("Ultra-wideband (32kHz)"), NULL
95 set_category( CAT_INPUT )
96 set_subcategory( SUBCAT_INPUT_ACODEC )
98 set_description( N_("Speex audio decoder") )
99 set_capability( "decoder", 100 )
100 set_shortname( N_("Speex") )
101 set_callbacks( OpenDecoder, CloseDecoder )
104 set_description( N_("Speex audio packetizer") )
105 set_capability( "packetizer", 100 )
106 set_callbacks( OpenPacketizer, CloseDecoder )
109 set_description( N_("Speex audio encoder") )
110 set_capability( "encoder", 100 )
111 set_callbacks( OpenEncoder, CloseEncoder )
113 add_integer( ENC_CFG_PREFIX "mode", 0, NULL, ENC_MODE_TEXT,
114 ENC_MODE_LONGTEXT, false )
115 change_integer_list( pi_enc_mode_values, ppsz_enc_mode_descriptions, NULL )
117 add_integer( ENC_CFG_PREFIX "complexity", 3, NULL, ENC_COMPLEXITY_TEXT,
118 ENC_COMPLEXITY_LONGTEXT, false )
119 change_integer_range( 1, 10 )
121 add_bool( ENC_CFG_PREFIX "cbr", false, NULL, ENC_CBR_TEXT,
122 ENC_CBR_LONGTEXT, false )
124 add_float( ENC_CFG_PREFIX "quality", 8.0, NULL, ENC_QUALITY_TEXT,
125 ENC_QUALITY_LONGTEXT, false )
126 change_float_range( 0.0, 10.0 )
128 add_integer( ENC_CFG_PREFIX "max-bitrate", 0, NULL, ENC_MAXBITRATE_TEXT,
129 ENC_MAXBITRATE_LONGTEXT, false )
131 add_bool( ENC_CFG_PREFIX "vad", true, NULL, ENC_VAD_TEXT,
132 ENC_VAD_LONGTEXT, false )
134 add_bool( ENC_CFG_PREFIX "dtx", false, NULL, ENC_DTX_TEXT,
135 ENC_DTX_LONGTEXT, false )
137 /* TODO agc, noise suppression, */
141 static const char *const ppsz_enc_options[] = {
142 "mode", "complexity", "cbr", "quality", "max-bitrate", "vad", "dtx", NULL
145 /*****************************************************************************
146 * decoder_sys_t : speex decoder descriptor
147 *****************************************************************************/
157 int i_frame_in_packet;
163 SpeexHeader *p_header;
164 SpeexStereoState stereo;
166 unsigned int rtp_rate;
175 static const int pi_channels_maps[6] =
178 AOUT_CHAN_CENTER, AOUT_CHAN_LEFT | AOUT_CHAN_RIGHT,
179 AOUT_CHAN_CENTER | AOUT_CHAN_LEFT | AOUT_CHAN_RIGHT,
180 AOUT_CHAN_LEFT | AOUT_CHAN_RIGHT | AOUT_CHAN_REARLEFT
181 | AOUT_CHAN_REARRIGHT,
182 AOUT_CHAN_LEFT | AOUT_CHAN_RIGHT | AOUT_CHAN_CENTER
183 | AOUT_CHAN_REARLEFT | AOUT_CHAN_REARRIGHT
186 /****************************************************************************
188 ****************************************************************************/
190 static void *DecodeBlock ( decoder_t *, block_t ** );
191 static aout_buffer_t *DecodeRtpSpeexPacket( decoder_t *, block_t **);
192 static int ProcessHeaders( decoder_t * );
193 static int ProcessInitialHeader ( decoder_t *, ogg_packet * );
194 static void *ProcessPacket( decoder_t *, ogg_packet *, block_t ** );
196 static aout_buffer_t *DecodePacket( decoder_t *, ogg_packet * );
197 static block_t *SendPacket( decoder_t *, block_t * );
199 static void ParseSpeexComments( decoder_t *, ogg_packet * );
201 static block_t *Encode ( encoder_t *, aout_buffer_t * );
203 /*****************************************************************************
204 * OpenDecoder: probe the decoder and return score
205 *****************************************************************************/
206 static int OpenDecoder( vlc_object_t *p_this )
208 decoder_t *p_dec = (decoder_t*)p_this;
209 decoder_sys_t *p_sys;
211 if( p_dec->fmt_in.i_codec != VLC_CODEC_SPEEX )
214 /* Allocate the memory needed to store the decoder's structure */
215 if( ( p_dec->p_sys = p_sys = malloc(sizeof(decoder_sys_t)) ) == NULL )
217 p_dec->p_sys->bits.buf_size = 0;
218 p_dec->p_sys->b_packetizer = false;
219 p_dec->p_sys->rtp_rate = p_dec->fmt_in.audio.i_rate;
221 date_Set( &p_sys->end_date, 0 );
223 /* Set output properties */
224 p_dec->fmt_out.i_cat = AUDIO_ES;
225 p_dec->fmt_out.i_codec = VLC_CODEC_S16N;
229 If the codec is spxr then this decoder is
230 being invoked on a Speex stream arriving via RTP.
231 A special decoder callback is used.
233 if (p_dec->fmt_in.i_original_fourcc == VLC_FOURCC('s', 'p', 'x', 'r'))
235 msg_Dbg( p_dec, "Using RTP version of Speex decoder @ rate %d.",
236 p_dec->fmt_in.audio.i_rate );
237 p_dec->pf_decode_audio = (aout_buffer_t *(*)(decoder_t *, block_t **))
238 DecodeRtpSpeexPacket;
242 p_dec->pf_decode_audio = (aout_buffer_t *(*)(decoder_t *, block_t **))
245 p_dec->pf_packetize = (block_t *(*)(decoder_t *, block_t **))
248 p_sys->i_headers = 0;
249 p_sys->p_state = NULL;
250 p_sys->p_header = NULL;
251 p_sys->i_frame_in_packet = 0;
256 static int OpenPacketizer( vlc_object_t *p_this )
258 decoder_t *p_dec = (decoder_t*)p_this;
260 int i_ret = OpenDecoder( p_this );
262 if( i_ret == VLC_SUCCESS )
264 p_dec->p_sys->b_packetizer = true;
265 p_dec->fmt_out.i_codec = VLC_CODEC_SPEEX;
271 /****************************************************************************
272 * DecodeBlock: the whole thing
273 ****************************************************************************
274 * This function must be fed with ogg packets.
275 ****************************************************************************/
276 static void *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
278 decoder_sys_t *p_sys = p_dec->p_sys;
279 ogg_packet oggpacket;
281 if( !pp_block ) return NULL;
285 /* Block to Ogg packet */
286 oggpacket.packet = (*pp_block)->p_buffer;
287 oggpacket.bytes = (*pp_block)->i_buffer;
291 if( p_sys->b_packetizer ) return NULL;
293 /* Block to Ogg packet */
294 oggpacket.packet = NULL;
298 oggpacket.granulepos = -1;
301 oggpacket.packetno = 0;
303 /* Check for headers */
304 if( p_sys->i_headers == 0 && p_dec->fmt_in.i_extra )
306 p_sys->i_headers = 2;
308 else if( oggpacket.bytes && p_sys->i_headers < 2 )
312 p_dec->fmt_in.p_extra = realloc_or_free( p_dec->fmt_in.p_extra,
313 p_dec->fmt_in.i_extra + oggpacket.bytes + 2 );
314 assert( p_dec->fmt_in.p_extra );
315 p_extra = ((uint8_t *)p_dec->fmt_in.p_extra) + p_dec->fmt_in.i_extra;
316 *(p_extra++) = oggpacket.bytes >> 8;
317 *(p_extra++) = oggpacket.bytes & 0xFF;
319 memcpy( p_extra, oggpacket.packet, oggpacket.bytes );
320 p_dec->fmt_in.i_extra += oggpacket.bytes + 2;
322 block_Release( *pp_block );
327 if( p_sys->i_headers == 2 )
329 if( ProcessHeaders( p_dec ) != VLC_SUCCESS )
331 p_sys->i_headers = 0;
332 p_dec->fmt_in.i_extra = 0;
333 block_Release( *pp_block );
336 else p_sys->i_headers++;
339 return ProcessPacket( p_dec, &oggpacket, pp_block );
342 /*****************************************************************************
343 * ProcessHeaders: process Speex headers.
344 *****************************************************************************/
345 static int ProcessHeaders( decoder_t *p_dec )
347 decoder_sys_t *p_sys = p_dec->p_sys;
348 ogg_packet oggpacket;
352 if( !p_dec->fmt_in.i_extra ) return VLC_EGENERIC;
354 oggpacket.granulepos = -1;
355 oggpacket.b_o_s = 1; /* yes this actually is a b_o_s packet :) */
357 oggpacket.packetno = 0;
358 p_extra = p_dec->fmt_in.p_extra;
359 i_extra = p_dec->fmt_in.i_extra;
361 /* Take care of the initial Vorbis header */
362 oggpacket.bytes = *(p_extra++) << 8;
363 oggpacket.bytes |= (*(p_extra++) & 0xFF);
364 oggpacket.packet = p_extra;
365 p_extra += oggpacket.bytes;
366 i_extra -= (oggpacket.bytes + 2);
369 msg_Err( p_dec, "header data corrupted");
373 /* Take care of the initial Speex header */
374 if( ProcessInitialHeader( p_dec, &oggpacket ) != VLC_SUCCESS )
376 msg_Err( p_dec, "initial Speex header is corrupted" );
380 /* The next packet in order is the comments header */
382 oggpacket.bytes = *(p_extra++) << 8;
383 oggpacket.bytes |= (*(p_extra++) & 0xFF);
384 oggpacket.packet = p_extra;
385 p_extra += oggpacket.bytes;
386 i_extra -= (oggpacket.bytes + 2);
389 msg_Err( p_dec, "header data corrupted");
393 ParseSpeexComments( p_dec, &oggpacket );
395 if( p_sys->b_packetizer )
397 p_dec->fmt_out.i_extra = p_dec->fmt_in.i_extra;
398 p_dec->fmt_out.p_extra = realloc_or_free( p_dec->fmt_out.p_extra,
399 p_dec->fmt_out.i_extra );
400 assert( p_dec->fmt_out.p_extra );
401 memcpy( p_dec->fmt_out.p_extra,
402 p_dec->fmt_in.p_extra, p_dec->fmt_out.i_extra );
408 /*****************************************************************************
409 * ProcessInitialHeader: processes the inital Speex header packet.
410 *****************************************************************************/
411 static int ProcessInitialHeader( decoder_t *p_dec, ogg_packet *p_oggpacket )
413 decoder_sys_t *p_sys = p_dec->p_sys;
416 SpeexHeader *p_header;
417 const SpeexMode *p_mode;
418 SpeexCallback callback;
420 p_sys->p_header = p_header =
421 speex_packet_to_header( (char *)p_oggpacket->packet,
422 p_oggpacket->bytes );
425 msg_Err( p_dec, "cannot read Speex header" );
428 if( p_header->mode >= SPEEX_NB_MODES || p_header->mode < 0 )
430 msg_Err( p_dec, "mode number %d does not (yet/any longer) exist in "
431 "this version of libspeex.", p_header->mode );
435 p_mode = speex_mode_list[p_header->mode];
439 if( p_header->speex_version_id > 1 )
441 msg_Err( p_dec, "this file was encoded with Speex bit-stream "
442 "version %d which is not supported by this decoder.",
443 p_header->speex_version_id );
447 if( p_mode->bitstream_version < p_header->mode_bitstream_version )
449 msg_Err( p_dec, "file encoded with a newer version of Speex." );
452 if( p_mode->bitstream_version > p_header->mode_bitstream_version )
454 msg_Err( p_dec, "file encoded with an older version of Speex." );
458 msg_Dbg( p_dec, "Speex %d Hz audio using %s mode %s%s",
459 p_header->rate, p_mode->modeName,
460 ( p_header->nb_channels == 1 ) ? " (mono" : " (stereo",
461 p_header->vbr ? ", VBR)" : ")" );
463 /* Take care of speex decoder init */
464 speex_bits_init( &p_sys->bits );
465 p_sys->p_state = p_state = speex_decoder_init( p_mode );
468 msg_Err( p_dec, "decoder initialization failed" );
472 if( p_header->nb_channels == 2 )
474 SpeexStereoState stereo = SPEEX_STEREO_STATE_INIT;
475 p_sys->stereo = stereo;
476 callback.callback_id = SPEEX_INBAND_STEREO;
477 callback.func = speex_std_stereo_request_handler;
478 callback.data = &p_sys->stereo;
479 speex_decoder_ctl( p_state, SPEEX_SET_HANDLER, &callback );
481 if( p_header->nb_channels <= 0 ||
482 p_header->nb_channels > 5 )
484 msg_Err( p_dec, "invalid number of channels (not between 1 and 5): %i",
485 p_header->nb_channels );
489 /* Setup the format */
490 p_dec->fmt_out.audio.i_physical_channels =
491 p_dec->fmt_out.audio.i_original_channels =
492 pi_channels_maps[p_header->nb_channels];
493 p_dec->fmt_out.audio.i_channels = p_header->nb_channels;
494 p_dec->fmt_out.audio.i_rate = p_header->rate;
496 date_Init( &p_sys->end_date, p_header->rate, 1 );
501 /*****************************************************************************
502 * ProcessPacket: processes a Speex packet.
503 *****************************************************************************/
504 static void *ProcessPacket( decoder_t *p_dec, ogg_packet *p_oggpacket,
507 decoder_sys_t *p_sys = p_dec->p_sys;
508 block_t *p_block = *pp_block;
510 /* Date management */
511 if( p_block && p_block->i_pts > 0 &&
512 p_block->i_pts != date_Get( &p_sys->end_date ) )
514 date_Set( &p_sys->end_date, p_block->i_pts );
517 if( !date_Get( &p_sys->end_date ) )
519 /* We've just started the stream, wait for the first PTS. */
520 if( p_block ) block_Release( p_block );
524 *pp_block = NULL; /* To avoid being fed the same packet again */
526 if( p_sys->b_packetizer )
528 if ( p_sys->p_header->frames_per_packet > 1 )
530 short *p_frame_holder = NULL;
531 int i_bits_before = 0, i_bits_after = 0, i_bytes_in_speex_frame = 0,
532 i_pcm_output_size = 0, i_bits_in_speex_frame = 0;
533 block_t *p_new_block = NULL;
535 i_pcm_output_size = p_sys->p_header->frame_size;
536 p_frame_holder = (short*)malloc( sizeof(short)*i_pcm_output_size );
537 assert( p_frame_holder );
539 speex_bits_read_from( &p_sys->bits, (char*)p_oggpacket->packet,
541 i_bits_before = speex_bits_remaining( &p_sys->bits );
542 speex_decode_int(p_sys->p_state, &p_sys->bits, p_frame_holder);
543 i_bits_after = speex_bits_remaining( &p_sys->bits );
545 i_bits_in_speex_frame = i_bits_before - i_bits_after;
546 i_bytes_in_speex_frame = ( i_bits_in_speex_frame +
547 (8 - (i_bits_in_speex_frame % 8)) )
550 p_new_block = block_New( p_dec, i_bytes_in_speex_frame );
551 memset( p_new_block->p_buffer, 0xff, i_bytes_in_speex_frame );
554 * Copy the first frame in this packet to a new packet.
556 speex_bits_rewind( &p_sys->bits );
557 speex_bits_write( &p_sys->bits,
558 (char*)p_new_block->p_buffer,
559 (int)i_bytes_in_speex_frame );
562 * Move the remaining part of the original packet (subsequent
563 * frames, if there are any) into the beginning
564 * of the original packet so
565 * they are preserved following the realloc.
566 * Note: Any bits that
567 * remain in the initial packet
568 * are "filler" if they do not constitute
571 if ( i_bits_after > 7 )
573 /* round-down since we rounded-up earlier (to include
574 * the speex terminator code.
576 i_bytes_in_speex_frame--;
577 speex_bits_write( &p_sys->bits,
578 (char*)p_block->p_buffer,
579 p_block->i_buffer - i_bytes_in_speex_frame );
580 p_block = block_Realloc( p_block,
582 p_block->i_buffer-i_bytes_in_speex_frame );
587 speex_bits_reset( &p_sys->bits );
590 free( p_frame_holder );
591 return SendPacket( p_dec, p_new_block);
595 return SendPacket( p_dec, p_block );
600 aout_buffer_t *p_aout_buffer;
602 if( p_sys->i_headers >= p_sys->p_header->extra_headers + 2 )
603 p_aout_buffer = DecodePacket( p_dec, p_oggpacket );
605 p_aout_buffer = NULL; /* Skip headers */
607 if( p_block ) block_Release( p_block );
608 return p_aout_buffer;
612 static aout_buffer_t *DecodeRtpSpeexPacket( decoder_t *p_dec, block_t **pp_block )
614 block_t *p_speex_bit_block = *pp_block;
615 decoder_sys_t *p_sys = p_dec->p_sys;
616 aout_buffer_t *p_aout_buffer;
618 unsigned int i_speex_frame_size;
620 if ( !p_speex_bit_block || p_speex_bit_block->i_pts == 0 ) return NULL;
623 If the SpeexBits buffer size is 0 (a default value),
624 we know that a proper initialization has not yet been done.
626 if ( p_sys->bits.buf_size==0 )
628 p_sys->p_header = (SpeexHeader *)malloc(sizeof(SpeexHeader));
629 if ( !p_sys->p_header )
631 msg_Err( p_dec, "Could not allocate a Speex header.");
634 speex_init_header( p_sys->p_header,p_sys->rtp_rate,1,&speex_nb_mode );
635 speex_bits_init( &p_sys->bits );
636 p_sys->p_state = speex_decoder_init( &speex_nb_mode );
637 if ( !p_sys->p_state )
639 msg_Err( p_dec, "Could not allocate a Speex decoder." );
640 free( p_sys->p_header );
645 Assume that variable bit rate is enabled. Also assume
646 that there is only one frame per packet.
648 p_sys->p_header->vbr = 1;
649 p_sys->p_header->frames_per_packet = 1;
651 p_dec->fmt_out.audio.i_channels = p_sys->p_header->nb_channels;
652 p_dec->fmt_out.audio.i_physical_channels =
653 p_dec->fmt_out.audio.i_original_channels =
654 pi_channels_maps[p_sys->p_header->nb_channels];
655 p_dec->fmt_out.audio.i_rate = p_sys->p_header->rate;
657 if ( speex_mode_query( &speex_nb_mode,
658 SPEEX_MODE_FRAME_SIZE,
659 &i_speex_frame_size ) )
661 msg_Err( p_dec, "Could not determine the frame size." );
662 speex_decoder_destroy( p_sys->p_state );
663 free( p_sys->p_header );
666 p_dec->fmt_out.audio.i_bytes_per_frame = i_speex_frame_size;
668 date_Init(&p_sys->end_date, p_sys->p_header->rate, 1);
672 If the SpeexBits are initialized but there is
673 still no header, an error must be thrown.
675 if ( !p_sys->p_header )
677 msg_Err( p_dec, "There is no valid Speex header found." );
682 if ( !date_Get( &p_sys->end_date ) )
683 date_Set( &p_sys->end_date, p_speex_bit_block->i_dts );
686 Ask for a new audio output buffer and make sure
689 p_aout_buffer = decoder_NewAudioBuffer( p_dec,
690 p_sys->p_header->frame_size );
691 if ( !p_aout_buffer || p_aout_buffer->i_buffer == 0 )
693 msg_Err(p_dec, "Oops: No new buffer was returned!");
698 Read the Speex payload into the SpeexBits buffer.
700 speex_bits_read_from( &p_sys->bits,
701 (char*)p_speex_bit_block->p_buffer,
702 p_speex_bit_block->i_buffer );
705 Decode the input and ensure that no errors
708 i_decode_ret = speex_decode_int( p_sys->p_state, &p_sys->bits,
709 (int16_t*)p_aout_buffer->p_buffer );
710 if ( i_decode_ret < 0 )
712 msg_Err( p_dec, "Decoding failed. Perhaps we have a bad stream?" );
717 Handle date management on the audio output buffer.
719 p_aout_buffer->i_pts = date_Get( &p_sys->end_date );
720 p_aout_buffer->i_length = date_Increment( &p_sys->end_date,
721 p_sys->p_header->frame_size ) - p_aout_buffer->i_pts;
724 p_sys->i_frame_in_packet++;
725 block_Release( p_speex_bit_block );
727 return p_aout_buffer;
730 /*****************************************************************************
731 * DecodePacket: decodes a Speex packet.
732 *****************************************************************************/
733 static aout_buffer_t *DecodePacket( decoder_t *p_dec, ogg_packet *p_oggpacket )
735 decoder_sys_t *p_sys = p_dec->p_sys;
737 if( p_oggpacket->bytes )
739 /* Copy Ogg packet to Speex bitstream */
740 speex_bits_read_from( &p_sys->bits, (char *)p_oggpacket->packet,
741 p_oggpacket->bytes );
742 p_sys->i_frame_in_packet = 0;
745 /* Decode one frame at a time */
746 if( p_sys->i_frame_in_packet < p_sys->p_header->frames_per_packet )
748 aout_buffer_t *p_aout_buffer;
749 if( p_sys->p_header->frame_size == 0 )
753 decoder_NewAudioBuffer( p_dec, p_sys->p_header->frame_size );
759 switch( speex_decode_int( p_sys->p_state, &p_sys->bits,
760 (int16_t *)p_aout_buffer->p_buffer ) )
763 msg_Err( p_dec, "decoding error: corrupted stream?" );
764 case -1: /* End of stream */
768 if( speex_bits_remaining( &p_sys->bits ) < 0 )
770 msg_Err( p_dec, "decoding overflow: corrupted stream?" );
773 if( p_sys->p_header->nb_channels == 2 )
774 speex_decode_stereo_int( (int16_t *)p_aout_buffer->p_buffer,
775 p_sys->p_header->frame_size,
778 /* Date management */
779 p_aout_buffer->i_pts = date_Get( &p_sys->end_date );
780 p_aout_buffer->i_length =
781 date_Increment( &p_sys->end_date, p_sys->p_header->frame_size )
782 - p_aout_buffer->i_pts;
784 p_sys->i_frame_in_packet++;
786 return p_aout_buffer;
794 /*****************************************************************************
795 * SendPacket: send an ogg packet to the stream output.
796 *****************************************************************************/
797 static block_t *SendPacket( decoder_t *p_dec, block_t *p_block )
799 decoder_sys_t *p_sys = p_dec->p_sys;
801 /* Date management */
802 p_block->i_dts = p_block->i_pts = date_Get( &p_sys->end_date );
804 if( p_sys->i_headers >= p_sys->p_header->extra_headers + 2 )
807 date_Increment( &p_sys->end_date,
808 p_sys->p_header->frame_size ) -
812 p_block->i_length = 0;
817 /*****************************************************************************
818 * ParseSpeexComments:
819 *****************************************************************************/
820 #define readint(buf, base) (((buf[base+3]<<24)&0xff000000)| \
821 ((buf[base+2]<<16)&0xff0000)| \
822 ((buf[base+1]<<8)&0xff00)| \
825 static void ParseSpeexComments( decoder_t *p_dec, ogg_packet *p_oggpacket )
827 decoder_sys_t *p_sys = p_dec->p_sys;
828 const SpeexMode *p_mode;
830 assert( p_sys->p_header->mode < SPEEX_NB_MODES );
832 p_mode = speex_mode_list[p_sys->p_header->mode];
833 assert( p_mode != NULL );
835 if( !p_dec->p_description )
837 p_dec->p_description = vlc_meta_New();
838 if( !p_dec->p_description )
844 if( asprintf( &psz_mode, "%s%s", p_mode->modeName, p_sys->p_header->vbr ? " VBR" : "" ) >= 0 )
846 vlc_meta_AddExtra( p_dec->p_description, _("Mode"), psz_mode );
850 /* TODO: finish comments parsing */
851 VLC_UNUSED( p_oggpacket );
854 /*****************************************************************************
855 * CloseDecoder: speex decoder destruction
856 *****************************************************************************/
857 static void CloseDecoder( vlc_object_t *p_this )
859 decoder_t * p_dec = (decoder_t *)p_this;
860 decoder_sys_t *p_sys = p_dec->p_sys;
864 speex_decoder_destroy( p_sys->p_state );
865 speex_bits_destroy( &p_sys->bits );
868 free( p_sys->p_header );
872 /*****************************************************************************
873 * encoder_sys_t: encoder descriptor
874 *****************************************************************************/
875 #define MAX_FRAME_SIZE 2000
876 #define MAX_FRAME_BYTES 2000
884 char p_buffer_out[MAX_FRAME_BYTES];
891 SpeexStereoState stereo;
894 int i_frames_per_packet;
895 int i_frames_in_packet;
907 /*****************************************************************************
908 * OpenEncoder: probe the encoder and return score
909 *****************************************************************************/
910 static int OpenEncoder( vlc_object_t *p_this )
912 encoder_t *p_enc = (encoder_t *)p_this;
913 encoder_sys_t *p_sys;
914 const SpeexMode *p_speex_mode = &speex_nb_mode;
916 const char *pp_header[2];
920 if( p_enc->fmt_out.i_codec != VLC_CODEC_SPEEX &&
926 config_ChainParse( p_enc, ENC_CFG_PREFIX, ppsz_enc_options, p_enc->p_cfg );
927 switch( var_GetInteger( p_enc, ENC_CFG_PREFIX "mode" ) )
930 msg_Dbg( p_enc, "Using wideband" );
931 p_speex_mode = &speex_wb_mode;
934 msg_Dbg( p_enc, "Using ultra-wideband" );
935 p_speex_mode = &speex_uwb_mode;
938 msg_Dbg( p_enc, "Using narrowband" );
939 p_speex_mode = &speex_nb_mode;
943 /* Allocate the memory needed to store the decoder's structure */
944 if( ( p_sys = (encoder_sys_t *)malloc(sizeof(encoder_sys_t)) ) == NULL )
946 p_enc->p_sys = p_sys;
947 p_enc->pf_encode_audio = Encode;
948 p_enc->fmt_in.i_codec = VLC_CODEC_S16N;
949 p_enc->fmt_out.i_codec = VLC_CODEC_SPEEX;
951 speex_init_header( &p_sys->header, p_enc->fmt_in.audio.i_rate,
954 p_sys->header.frames_per_packet = 1;
955 p_sys->header.vbr = var_GetBool( p_enc, ENC_CFG_PREFIX "cbr" ) ? 0 : 1;
956 p_sys->header.nb_channels = p_enc->fmt_in.audio.i_channels;
958 /* Create a new encoder state in narrowband mode */
959 p_sys->p_state = speex_encoder_init( p_speex_mode );
962 i_tmp = var_GetInteger( p_enc, ENC_CFG_PREFIX "complexity" );
963 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_COMPLEXITY, &i_tmp );
965 i_tmp = var_GetBool( p_enc, ENC_CFG_PREFIX "cbr" ) ? 0 : 1;
966 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_VBR, &i_tmp );
968 if( i_tmp == 0 ) /* CBR */
970 i_tmp = var_GetFloat( p_enc, ENC_CFG_PREFIX "quality" );
971 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_QUALITY, &i_tmp );
973 i_tmp = var_GetBool( p_enc, ENC_CFG_PREFIX "vad" ) ? 1 : 0;
974 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_VAD, &i_tmp );
980 f_tmp = var_GetFloat( p_enc, ENC_CFG_PREFIX "quality" );
981 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_VBR_QUALITY, &f_tmp );
983 i_tmp = var_GetInteger( p_enc, ENC_CFG_PREFIX "max-bitrate" );
985 #ifdef SPEEX_SET_VBR_MAX_BITRATE
986 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_VBR_MAX_BITRATE, &i_tmp );
988 msg_Dbg( p_enc, "max-bitrate cannot be set in this version of libspeex");
992 i_tmp = var_GetBool( p_enc, ENC_CFG_PREFIX "dtx" ) ? 1 : 0;
993 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_DTX, &i_tmp );
996 /*Initialization of the structure that holds the bits*/
997 speex_bits_init( &p_sys->bits );
999 p_sys->i_frames_in_packet = 0;
1000 p_sys->i_samples_delay = 0;
1003 speex_encoder_ctl( p_sys->p_state, SPEEX_GET_FRAME_SIZE,
1004 &p_sys->i_frame_length );
1006 p_sys->i_frame_size = p_sys->i_frame_length *
1007 sizeof(int16_t) * p_enc->fmt_in.audio.i_channels;
1008 p_sys->p_buffer = malloc( p_sys->i_frame_size );
1009 assert( p_sys->p_buffer );
1011 /* Create and store headers */
1012 pp_header[0] = speex_header_to_packet( &p_sys->header, &pi_header[0] );
1013 pp_header[1] = "ENCODER=VLC media player";
1014 pi_header[1] = sizeof("ENCODER=VLC media player");
1016 p_enc->fmt_out.i_extra = 3 * 2 + pi_header[0] + pi_header[1];
1017 p_extra = p_enc->fmt_out.p_extra = malloc( p_enc->fmt_out.i_extra );
1019 for( i = 0; i < 2; i++ )
1021 *(p_extra++) = pi_header[i] >> 8;
1022 *(p_extra++) = pi_header[i] & 0xFF;
1023 memcpy( p_extra, pp_header[i], pi_header[i] );
1024 p_extra += pi_header[i];
1027 msg_Dbg( p_enc, "encoding: frame size:%d, channels:%d, samplerate:%d",
1028 p_sys->i_frame_size, p_enc->fmt_in.audio.i_channels,
1029 p_enc->fmt_in.audio.i_rate );
1034 /****************************************************************************
1035 * Encode: the whole thing
1036 ****************************************************************************
1037 * This function spits out ogg packets.
1038 ****************************************************************************/
1039 static block_t *Encode( encoder_t *p_enc, aout_buffer_t *p_aout_buf )
1041 encoder_sys_t *p_sys = p_enc->p_sys;
1042 block_t *p_block, *p_chain = NULL;
1044 unsigned char *p_buffer = p_aout_buf->p_buffer;
1045 int i_samples = p_aout_buf->i_nb_samples;
1046 int i_samples_delay = p_sys->i_samples_delay;
1048 p_sys->i_pts = p_aout_buf->i_pts -
1049 (mtime_t)1000000 * (mtime_t)p_sys->i_samples_delay /
1050 (mtime_t)p_enc->fmt_in.audio.i_rate;
1052 p_sys->i_samples_delay += i_samples;
1054 while( p_sys->i_samples_delay >= p_sys->i_frame_length )
1059 if( i_samples_delay )
1061 /* Take care of the left-over from last time */
1062 int i_delay_size = i_samples_delay * 2 *
1063 p_enc->fmt_in.audio.i_channels;
1064 int i_size = p_sys->i_frame_size - i_delay_size;
1066 p_samples = (int16_t *)p_sys->p_buffer;
1067 memcpy( p_sys->p_buffer + i_delay_size, p_buffer, i_size );
1068 p_buffer -= i_delay_size;
1069 i_samples += i_samples_delay;
1070 i_samples_delay = 0;
1074 p_samples = (int16_t *)p_buffer;
1077 /* Encode current frame */
1078 if( p_enc->fmt_in.audio.i_channels == 2 )
1079 speex_encode_stereo_int( p_samples, p_sys->i_frame_length,
1083 if( p_sys->preprocess )
1084 speex_preprocess( p_sys->preprocess, p_samples, NULL );
1087 speex_encode_int( p_sys->p_state, p_samples, &p_sys->bits );
1089 p_buffer += p_sys->i_frame_size;
1090 p_sys->i_samples_delay -= p_sys->i_frame_length;
1091 i_samples -= p_sys->i_frame_length;
1093 p_sys->i_frames_in_packet++;
1095 if( p_sys->i_frames_in_packet < p_sys->header.frames_per_packet )
1098 p_sys->i_frames_in_packet = 0;
1100 speex_bits_insert_terminator( &p_sys->bits );
1101 i_out = speex_bits_write( &p_sys->bits, p_sys->p_buffer_out,
1103 speex_bits_reset( &p_sys->bits );
1105 p_block = block_New( p_enc, i_out );
1106 memcpy( p_block->p_buffer, p_sys->p_buffer_out, i_out );
1108 p_block->i_length = (mtime_t)1000000 *
1109 (mtime_t)p_sys->i_frame_length * p_sys->header.frames_per_packet /
1110 (mtime_t)p_enc->fmt_in.audio.i_rate;
1112 p_block->i_dts = p_block->i_pts = p_sys->i_pts;
1115 p_sys->i_pts += p_block->i_length;
1116 block_ChainAppend( &p_chain, p_block );
1120 /* Backup the remaining raw samples */
1123 memcpy( p_sys->p_buffer + i_samples_delay * 2 *
1124 p_enc->fmt_in.audio.i_channels, p_buffer,
1125 i_samples * 2 * p_enc->fmt_in.audio.i_channels );
1131 /*****************************************************************************
1132 * CloseEncoder: encoder destruction
1133 *****************************************************************************/
1134 static void CloseEncoder( vlc_object_t *p_this )
1136 encoder_t *p_enc = (encoder_t *)p_this;
1137 encoder_sys_t *p_sys = p_enc->p_sys;
1139 speex_encoder_destroy( p_sys->p_state );
1140 speex_bits_destroy( &p_sys->bits );
1142 free( p_sys->p_buffer );