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 *****************************************************************************/
31 #include <vlc_common.h>
32 #include <vlc_plugin.h>
33 #include <vlc_input.h>
34 #include <vlc_codec.h>
35 #include "../demux/xiph.h"
38 #include <speex/speex.h>
39 #include <speex/speex_header.h>
40 #include <speex/speex_stereo.h>
41 #include <speex/speex_callbacks.h>
45 /*****************************************************************************
47 *****************************************************************************/
48 static int OpenDecoder ( vlc_object_t * );
49 static int OpenPacketizer( vlc_object_t * );
50 static void CloseDecoder ( vlc_object_t * );
51 static int OpenEncoder ( vlc_object_t * );
52 static void CloseEncoder ( vlc_object_t * );
54 #define ENC_CFG_PREFIX "sout-speex-"
56 #define ENC_MODE_TEXT N_("Mode" )
57 #define ENC_MODE_LONGTEXT N_( \
58 "Enforce the mode of the encoder." )
60 #define ENC_QUALITY_TEXT N_("Encoding quality")
61 #define ENC_QUALITY_LONGTEXT N_( \
62 "Enforce a quality between 0 (low) and 10 (high)." )
64 #define ENC_COMPLEXITY_TEXT N_("Encoding complexity" )
65 #define ENC_COMPLEXITY_LONGTEXT N_( \
66 "Enforce the complexity of the encoder." )
68 #define ENC_MAXBITRATE_TEXT N_( "Maximal bitrate" )
69 #define ENC_MAXBITRATE_LONGTEXT N_( \
70 "Enforce the maximal VBR bitrate" )
72 #define ENC_CBR_TEXT N_( "CBR encoding" )
73 #define ENC_CBR_LONGTEXT N_( \
74 "Enforce a constant bitrate encoding (CBR) instead of default " \
75 "variable bitrate encoding (VBR)." )
77 #define ENC_VAD_TEXT N_( "Voice activity detection" )
78 #define ENC_VAD_LONGTEXT N_( \
79 "Enable voice activity detection (VAD). It is automatically " \
80 "activated in VBR mode." )
82 #define ENC_DTX_TEXT N_( "Discontinuous Transmission" )
83 #define ENC_DTX_LONGTEXT N_( \
84 "Enable discontinuous transmission (DTX)." )
86 static const int pi_enc_mode_values[] = { 0, 1, 2 };
87 static const char * const ppsz_enc_mode_descriptions[] = {
88 N_("Narrow-band (8kHz)"), N_("Wide-band (16kHz)"), N_("Ultra-wideband (32kHz)"), NULL
92 set_category( CAT_INPUT )
93 set_subcategory( SUBCAT_INPUT_ACODEC )
95 set_description( N_("Speex audio decoder") )
96 set_capability( "decoder", 100 )
97 set_shortname( N_("Speex") )
98 set_callbacks( OpenDecoder, CloseDecoder )
101 set_description( N_("Speex audio packetizer") )
102 set_capability( "packetizer", 100 )
103 set_callbacks( OpenPacketizer, CloseDecoder )
106 set_description( N_("Speex audio encoder") )
107 set_capability( "encoder", 100 )
108 set_callbacks( OpenEncoder, CloseEncoder )
110 add_integer( ENC_CFG_PREFIX "mode", 0, ENC_MODE_TEXT,
111 ENC_MODE_LONGTEXT, false )
112 change_integer_list( pi_enc_mode_values, ppsz_enc_mode_descriptions )
114 add_integer( ENC_CFG_PREFIX "complexity", 3, ENC_COMPLEXITY_TEXT,
115 ENC_COMPLEXITY_LONGTEXT, false )
116 change_integer_range( 1, 10 )
118 add_bool( ENC_CFG_PREFIX "cbr", false, ENC_CBR_TEXT,
119 ENC_CBR_LONGTEXT, false )
121 add_float( ENC_CFG_PREFIX "quality", 8.0, ENC_QUALITY_TEXT,
122 ENC_QUALITY_LONGTEXT, false )
123 change_float_range( 0.0, 10.0 )
125 add_integer( ENC_CFG_PREFIX "max-bitrate", 0, ENC_MAXBITRATE_TEXT,
126 ENC_MAXBITRATE_LONGTEXT, false )
128 add_bool( ENC_CFG_PREFIX "vad", true, ENC_VAD_TEXT,
129 ENC_VAD_LONGTEXT, false )
131 add_bool( ENC_CFG_PREFIX "dtx", false, ENC_DTX_TEXT,
132 ENC_DTX_LONGTEXT, false )
134 /* TODO agc, noise suppression, */
138 static const char *const ppsz_enc_options[] = {
139 "mode", "complexity", "cbr", "quality", "max-bitrate", "vad", "dtx", NULL
142 /*****************************************************************************
143 * decoder_sys_t : speex decoder descriptor
144 *****************************************************************************/
154 int i_frame_in_packet;
160 SpeexHeader *p_header;
161 SpeexStereoState stereo;
163 unsigned int rtp_rate;
172 static const int pi_channels_maps[6] =
175 AOUT_CHAN_CENTER, AOUT_CHAN_LEFT | AOUT_CHAN_RIGHT,
176 AOUT_CHAN_CENTER | AOUT_CHAN_LEFT | AOUT_CHAN_RIGHT,
177 AOUT_CHAN_LEFT | AOUT_CHAN_RIGHT | AOUT_CHAN_REARLEFT
178 | AOUT_CHAN_REARRIGHT,
179 AOUT_CHAN_LEFT | AOUT_CHAN_RIGHT | AOUT_CHAN_CENTER
180 | AOUT_CHAN_REARLEFT | AOUT_CHAN_REARRIGHT
183 /****************************************************************************
185 ****************************************************************************/
187 static block_t *DecodeBlock ( decoder_t *, block_t ** );
188 static block_t *DecodeRtpSpeexPacket( decoder_t *, block_t **);
189 static int ProcessHeaders( decoder_t * );
190 static int ProcessInitialHeader ( decoder_t *, ogg_packet * );
191 static void *ProcessPacket( decoder_t *, ogg_packet *, block_t ** );
193 static block_t *DecodePacket( decoder_t *, ogg_packet * );
194 static block_t *SendPacket( decoder_t *, block_t * );
196 static void ParseSpeexComments( decoder_t *, ogg_packet * );
198 static block_t *Encode ( encoder_t *, block_t * );
200 /*****************************************************************************
201 * OpenDecoder: probe the decoder and return score
202 *****************************************************************************/
203 static int OpenDecoder( vlc_object_t *p_this )
205 decoder_t *p_dec = (decoder_t*)p_this;
206 decoder_sys_t *p_sys;
208 if( p_dec->fmt_in.i_codec != VLC_CODEC_SPEEX )
211 /* Allocate the memory needed to store the decoder's structure */
212 if( ( p_dec->p_sys = p_sys = malloc(sizeof(decoder_sys_t)) ) == NULL )
214 p_dec->p_sys->bits.buf_size = 0;
215 p_dec->p_sys->b_packetizer = false;
216 p_dec->p_sys->rtp_rate = p_dec->fmt_in.audio.i_rate;
217 p_dec->p_sys->b_has_headers = false;
219 date_Set( &p_sys->end_date, 0 );
221 /* Set output properties */
222 p_dec->fmt_out.i_cat = AUDIO_ES;
223 p_dec->fmt_out.i_codec = VLC_CODEC_S16N;
227 If the codec is spxr then this decoder is
228 being invoked on a Speex stream arriving via RTP.
229 A special decoder callback is used.
231 if (p_dec->fmt_in.i_original_fourcc == VLC_FOURCC('s', 'p', 'x', 'r'))
233 msg_Dbg( p_dec, "Using RTP version of Speex decoder @ rate %d.",
234 p_dec->fmt_in.audio.i_rate );
235 p_dec->pf_decode_audio = DecodeRtpSpeexPacket;
239 p_dec->pf_decode_audio = DecodeBlock;
241 p_dec->pf_packetize = DecodeBlock;
243 p_sys->p_state = NULL;
244 p_sys->p_header = NULL;
245 p_sys->i_frame_in_packet = 0;
250 static int OpenPacketizer( vlc_object_t *p_this )
252 decoder_t *p_dec = (decoder_t*)p_this;
254 int i_ret = OpenDecoder( p_this );
256 if( i_ret == VLC_SUCCESS )
258 p_dec->p_sys->b_packetizer = true;
259 p_dec->fmt_out.i_codec = VLC_CODEC_SPEEX;
265 /****************************************************************************
266 * DecodeBlock: the whole thing
267 ****************************************************************************
268 * This function must be fed with ogg packets.
269 ****************************************************************************/
270 static block_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
272 decoder_sys_t *p_sys = p_dec->p_sys;
273 ogg_packet oggpacket;
275 if( !pp_block ) return NULL;
279 /* Block to Ogg packet */
280 oggpacket.packet = (*pp_block)->p_buffer;
281 oggpacket.bytes = (*pp_block)->i_buffer;
285 if( p_sys->b_packetizer ) return NULL;
287 /* Block to Ogg packet */
288 oggpacket.packet = NULL;
292 oggpacket.granulepos = -1;
295 oggpacket.packetno = 0;
297 /* Check for headers */
298 if( !p_sys->b_has_headers )
300 if( ProcessHeaders( p_dec ) )
302 block_Release( *pp_block );
305 p_sys->b_has_headers = true;
308 return ProcessPacket( p_dec, &oggpacket, pp_block );
311 /*****************************************************************************
312 * ProcessHeaders: process Speex headers.
313 *****************************************************************************/
314 static int ProcessHeaders( decoder_t *p_dec )
316 decoder_sys_t *p_sys = p_dec->p_sys;
317 ogg_packet oggpacket;
319 unsigned pi_size[XIPH_MAX_HEADER_COUNT];
320 void *pp_data[XIPH_MAX_HEADER_COUNT];
322 if( xiph_SplitHeaders( pi_size, pp_data, &i_count,
323 p_dec->fmt_in.i_extra, p_dec->fmt_in.p_extra) )
328 oggpacket.granulepos = -1;
330 oggpacket.packetno = 0;
332 /* Take care of the initial Vorbis header */
333 oggpacket.b_o_s = 1; /* yes this actually is a b_o_s packet :) */
334 oggpacket.bytes = pi_size[0];
335 oggpacket.packet = pp_data[0];
336 if( ProcessInitialHeader( p_dec, &oggpacket ) != VLC_SUCCESS )
338 msg_Err( p_dec, "initial Speex header is corrupted" );
342 /* The next packet in order is the comments header */
344 oggpacket.bytes = pi_size[1];
345 oggpacket.packet = pp_data[1];
346 ParseSpeexComments( p_dec, &oggpacket );
348 if( p_sys->b_packetizer )
350 p_dec->fmt_out.i_extra = p_dec->fmt_in.i_extra;
351 p_dec->fmt_out.p_extra = xrealloc( p_dec->fmt_out.p_extra,
352 p_dec->fmt_out.i_extra );
353 memcpy( p_dec->fmt_out.p_extra,
354 p_dec->fmt_in.p_extra, p_dec->fmt_out.i_extra );
357 for( unsigned i = 0; i < i_count; i++ )
362 for( unsigned i = 0; i < i_count; i++ )
367 /*****************************************************************************
368 * ProcessInitialHeader: processes the inital Speex header packet.
369 *****************************************************************************/
370 static int ProcessInitialHeader( decoder_t *p_dec, ogg_packet *p_oggpacket )
372 decoder_sys_t *p_sys = p_dec->p_sys;
375 SpeexHeader *p_header;
376 const SpeexMode *p_mode;
377 SpeexCallback callback;
379 p_sys->p_header = p_header =
380 speex_packet_to_header( (char *)p_oggpacket->packet,
381 p_oggpacket->bytes );
384 msg_Err( p_dec, "cannot read Speex header" );
387 if( p_header->mode >= SPEEX_NB_MODES || p_header->mode < 0 )
389 msg_Err( p_dec, "mode number %d does not (yet/any longer) exist in "
390 "this version of libspeex.", p_header->mode );
394 p_mode = speex_mode_list[p_header->mode];
398 if( p_header->speex_version_id > 1 )
400 msg_Err( p_dec, "this file was encoded with Speex bit-stream "
401 "version %d which is not supported by this decoder.",
402 p_header->speex_version_id );
406 if( p_mode->bitstream_version < p_header->mode_bitstream_version )
408 msg_Err( p_dec, "file encoded with a newer version of Speex." );
411 if( p_mode->bitstream_version > p_header->mode_bitstream_version )
413 msg_Err( p_dec, "file encoded with an older version of Speex." );
417 msg_Dbg( p_dec, "Speex %d Hz audio using %s mode %s%s",
418 p_header->rate, p_mode->modeName,
419 ( p_header->nb_channels == 1 ) ? " (mono" : " (stereo",
420 p_header->vbr ? ", VBR)" : ")" );
422 /* Take care of speex decoder init */
423 speex_bits_init( &p_sys->bits );
424 p_sys->p_state = p_state = speex_decoder_init( p_mode );
427 msg_Err( p_dec, "decoder initialization failed" );
431 if( p_header->nb_channels == 2 )
433 SpeexStereoState stereo = SPEEX_STEREO_STATE_INIT;
434 p_sys->stereo = stereo;
435 callback.callback_id = SPEEX_INBAND_STEREO;
436 callback.func = speex_std_stereo_request_handler;
437 callback.data = &p_sys->stereo;
438 speex_decoder_ctl( p_state, SPEEX_SET_HANDLER, &callback );
440 if( p_header->nb_channels <= 0 ||
441 p_header->nb_channels > 5 )
443 msg_Err( p_dec, "invalid number of channels (not between 1 and 5): %i",
444 p_header->nb_channels );
448 /* Setup the format */
449 p_dec->fmt_out.audio.i_physical_channels =
450 p_dec->fmt_out.audio.i_original_channels =
451 pi_channels_maps[p_header->nb_channels];
452 p_dec->fmt_out.audio.i_channels = p_header->nb_channels;
453 p_dec->fmt_out.audio.i_rate = p_header->rate;
455 date_Init( &p_sys->end_date, p_header->rate, 1 );
460 /*****************************************************************************
461 * ProcessPacket: processes a Speex packet.
462 *****************************************************************************/
463 static void *ProcessPacket( decoder_t *p_dec, ogg_packet *p_oggpacket,
466 decoder_sys_t *p_sys = p_dec->p_sys;
467 block_t *p_block = *pp_block;
469 /* Date management */
470 if( p_block && p_block->i_pts > VLC_TS_INVALID &&
471 p_block->i_pts != date_Get( &p_sys->end_date ) )
473 date_Set( &p_sys->end_date, p_block->i_pts );
476 if( !date_Get( &p_sys->end_date ) )
478 /* We've just started the stream, wait for the first PTS. */
479 if( p_block ) block_Release( p_block );
483 *pp_block = NULL; /* To avoid being fed the same packet again */
485 if( p_sys->b_packetizer )
487 if ( p_sys->p_header->frames_per_packet > 1 )
489 short *p_frame_holder = NULL;
490 int i_bits_before = 0, i_bits_after = 0, i_bytes_in_speex_frame = 0,
491 i_pcm_output_size = 0, i_bits_in_speex_frame = 0;
492 block_t *p_new_block = NULL;
494 i_pcm_output_size = p_sys->p_header->frame_size;
495 p_frame_holder = (short*)xmalloc( sizeof(short)*i_pcm_output_size );
497 speex_bits_read_from( &p_sys->bits, (char*)p_oggpacket->packet,
499 i_bits_before = speex_bits_remaining( &p_sys->bits );
500 speex_decode_int(p_sys->p_state, &p_sys->bits, p_frame_holder);
501 i_bits_after = speex_bits_remaining( &p_sys->bits );
503 i_bits_in_speex_frame = i_bits_before - i_bits_after;
504 i_bytes_in_speex_frame = ( i_bits_in_speex_frame +
505 (8 - (i_bits_in_speex_frame % 8)) )
508 p_new_block = block_New( p_dec, i_bytes_in_speex_frame );
509 memset( p_new_block->p_buffer, 0xff, i_bytes_in_speex_frame );
512 * Copy the first frame in this packet to a new packet.
514 speex_bits_rewind( &p_sys->bits );
515 speex_bits_write( &p_sys->bits,
516 (char*)p_new_block->p_buffer,
517 (int)i_bytes_in_speex_frame );
520 * Move the remaining part of the original packet (subsequent
521 * frames, if there are any) into the beginning
522 * of the original packet so
523 * they are preserved following the realloc.
524 * Note: Any bits that
525 * remain in the initial packet
526 * are "filler" if they do not constitute
529 if ( i_bits_after > 7 )
531 /* round-down since we rounded-up earlier (to include
532 * the speex terminator code.
534 i_bytes_in_speex_frame--;
535 speex_bits_write( &p_sys->bits,
536 (char*)p_block->p_buffer,
537 p_block->i_buffer - i_bytes_in_speex_frame );
538 p_block = block_Realloc( p_block,
540 p_block->i_buffer-i_bytes_in_speex_frame );
545 speex_bits_reset( &p_sys->bits );
548 free( p_frame_holder );
549 return SendPacket( p_dec, p_new_block);
553 return SendPacket( p_dec, p_block );
558 block_t *p_aout_buffer = DecodePacket( p_dec, p_oggpacket );
561 block_Release( p_block );
562 return p_aout_buffer;
566 static block_t *DecodeRtpSpeexPacket( decoder_t *p_dec, block_t **pp_block )
568 block_t *p_speex_bit_block = *pp_block;
569 decoder_sys_t *p_sys = p_dec->p_sys;
570 block_t *p_aout_buffer;
572 unsigned int i_speex_frame_size;
574 if ( !p_speex_bit_block || p_speex_bit_block->i_pts <= VLC_TS_INVALID )
578 If the SpeexBits buffer size is 0 (a default value),
579 we know that a proper initialization has not yet been done.
581 if ( p_sys->bits.buf_size==0 )
583 p_sys->p_header = (SpeexHeader *)malloc(sizeof(SpeexHeader));
584 if ( !p_sys->p_header )
586 msg_Err( p_dec, "Could not allocate a Speex header.");
589 speex_init_header( p_sys->p_header,p_sys->rtp_rate,1,&speex_nb_mode );
590 speex_bits_init( &p_sys->bits );
591 p_sys->p_state = speex_decoder_init( &speex_nb_mode );
592 if ( !p_sys->p_state )
594 msg_Err( p_dec, "Could not allocate a Speex decoder." );
595 free( p_sys->p_header );
600 Assume that variable bit rate is enabled. Also assume
601 that there is only one frame per packet.
603 p_sys->p_header->vbr = 1;
604 p_sys->p_header->frames_per_packet = 1;
606 p_dec->fmt_out.audio.i_channels = p_sys->p_header->nb_channels;
607 p_dec->fmt_out.audio.i_physical_channels =
608 p_dec->fmt_out.audio.i_original_channels =
609 pi_channels_maps[p_sys->p_header->nb_channels];
610 p_dec->fmt_out.audio.i_rate = p_sys->p_header->rate;
612 if ( speex_mode_query( &speex_nb_mode,
613 SPEEX_MODE_FRAME_SIZE,
614 &i_speex_frame_size ) )
616 msg_Err( p_dec, "Could not determine the frame size." );
617 speex_decoder_destroy( p_sys->p_state );
618 free( p_sys->p_header );
621 p_dec->fmt_out.audio.i_bytes_per_frame = i_speex_frame_size;
623 date_Init(&p_sys->end_date, p_sys->p_header->rate, 1);
627 If the SpeexBits are initialized but there is
628 still no header, an error must be thrown.
630 if ( !p_sys->p_header )
632 msg_Err( p_dec, "There is no valid Speex header found." );
637 if ( !date_Get( &p_sys->end_date ) )
638 date_Set( &p_sys->end_date, p_speex_bit_block->i_dts );
641 Ask for a new audio output buffer and make sure
644 p_aout_buffer = decoder_NewAudioBuffer( p_dec,
645 p_sys->p_header->frame_size );
646 if ( !p_aout_buffer || p_aout_buffer->i_buffer == 0 )
648 msg_Err(p_dec, "Oops: No new buffer was returned!");
653 Read the Speex payload into the SpeexBits buffer.
655 speex_bits_read_from( &p_sys->bits,
656 (char*)p_speex_bit_block->p_buffer,
657 p_speex_bit_block->i_buffer );
660 Decode the input and ensure that no errors
663 i_decode_ret = speex_decode_int( p_sys->p_state, &p_sys->bits,
664 (int16_t*)p_aout_buffer->p_buffer );
665 if ( i_decode_ret < 0 )
667 msg_Err( p_dec, "Decoding failed. Perhaps we have a bad stream?" );
672 Handle date management on the audio output buffer.
674 p_aout_buffer->i_pts = date_Get( &p_sys->end_date );
675 p_aout_buffer->i_length = date_Increment( &p_sys->end_date,
676 p_sys->p_header->frame_size ) - p_aout_buffer->i_pts;
679 p_sys->i_frame_in_packet++;
680 block_Release( p_speex_bit_block );
682 return p_aout_buffer;
685 /*****************************************************************************
686 * DecodePacket: decodes a Speex packet.
687 *****************************************************************************/
688 static block_t *DecodePacket( decoder_t *p_dec, ogg_packet *p_oggpacket )
690 decoder_sys_t *p_sys = p_dec->p_sys;
692 if( p_oggpacket->bytes )
694 /* Copy Ogg packet to Speex bitstream */
695 speex_bits_read_from( &p_sys->bits, (char *)p_oggpacket->packet,
696 p_oggpacket->bytes );
697 p_sys->i_frame_in_packet = 0;
700 /* Decode one frame at a time */
701 if( p_sys->i_frame_in_packet < p_sys->p_header->frames_per_packet )
703 block_t *p_aout_buffer;
704 if( p_sys->p_header->frame_size == 0 )
708 decoder_NewAudioBuffer( p_dec, p_sys->p_header->frame_size );
714 switch( speex_decode_int( p_sys->p_state, &p_sys->bits,
715 (int16_t *)p_aout_buffer->p_buffer ) )
718 msg_Err( p_dec, "decoding error: corrupted stream?" );
719 case -1: /* End of stream */
723 if( speex_bits_remaining( &p_sys->bits ) < 0 )
725 msg_Err( p_dec, "decoding overflow: corrupted stream?" );
728 if( p_sys->p_header->nb_channels == 2 )
729 speex_decode_stereo_int( (int16_t *)p_aout_buffer->p_buffer,
730 p_sys->p_header->frame_size,
733 /* Date management */
734 p_aout_buffer->i_pts = date_Get( &p_sys->end_date );
735 p_aout_buffer->i_length =
736 date_Increment( &p_sys->end_date, p_sys->p_header->frame_size )
737 - p_aout_buffer->i_pts;
739 p_sys->i_frame_in_packet++;
741 return p_aout_buffer;
749 /*****************************************************************************
750 * SendPacket: send an ogg packet to the stream output.
751 *****************************************************************************/
752 static block_t *SendPacket( decoder_t *p_dec, block_t *p_block )
754 decoder_sys_t *p_sys = p_dec->p_sys;
756 /* Date management */
757 p_block->i_dts = p_block->i_pts = date_Get( &p_sys->end_date );
760 date_Increment( &p_sys->end_date,
761 p_sys->p_header->frame_size ) -
767 /*****************************************************************************
768 * ParseSpeexComments:
769 *****************************************************************************/
770 #define readint(buf, base) (((buf[base+3]<<24)&0xff000000)| \
771 ((buf[base+2]<<16)&0xff0000)| \
772 ((buf[base+1]<<8)&0xff00)| \
775 static void ParseSpeexComments( decoder_t *p_dec, ogg_packet *p_oggpacket )
777 decoder_sys_t *p_sys = p_dec->p_sys;
778 const SpeexMode *p_mode;
780 assert( p_sys->p_header->mode < SPEEX_NB_MODES );
782 p_mode = speex_mode_list[p_sys->p_header->mode];
783 assert( p_mode != NULL );
785 if( !p_dec->p_description )
787 p_dec->p_description = vlc_meta_New();
788 if( !p_dec->p_description )
794 if( asprintf( &psz_mode, "%s%s", p_mode->modeName, p_sys->p_header->vbr ? " VBR" : "" ) >= 0 )
796 vlc_meta_AddExtra( p_dec->p_description, _("Mode"), psz_mode );
800 /* TODO: finish comments parsing */
801 VLC_UNUSED( p_oggpacket );
804 /*****************************************************************************
805 * CloseDecoder: speex decoder destruction
806 *****************************************************************************/
807 static void CloseDecoder( vlc_object_t *p_this )
809 decoder_t * p_dec = (decoder_t *)p_this;
810 decoder_sys_t *p_sys = p_dec->p_sys;
814 speex_decoder_destroy( p_sys->p_state );
815 speex_bits_destroy( &p_sys->bits );
818 free( p_sys->p_header );
822 /*****************************************************************************
823 * encoder_sys_t: encoder descriptor
824 *****************************************************************************/
825 #define MAX_FRAME_SIZE 2000
826 #define MAX_FRAME_BYTES 2000
834 char p_buffer_out[MAX_FRAME_BYTES];
841 SpeexStereoState stereo;
844 int i_frames_per_packet;
845 int i_frames_in_packet;
852 /*****************************************************************************
853 * OpenEncoder: probe the encoder and return score
854 *****************************************************************************/
855 static int OpenEncoder( vlc_object_t *p_this )
857 encoder_t *p_enc = (encoder_t *)p_this;
858 encoder_sys_t *p_sys;
859 const SpeexMode *p_speex_mode = &speex_nb_mode;
861 const char *pp_header[2];
865 if( p_enc->fmt_out.i_codec != VLC_CODEC_SPEEX &&
871 config_ChainParse( p_enc, ENC_CFG_PREFIX, ppsz_enc_options, p_enc->p_cfg );
872 switch( var_GetInteger( p_enc, ENC_CFG_PREFIX "mode" ) )
875 msg_Dbg( p_enc, "Using wideband" );
876 p_speex_mode = &speex_wb_mode;
879 msg_Dbg( p_enc, "Using ultra-wideband" );
880 p_speex_mode = &speex_uwb_mode;
883 msg_Dbg( p_enc, "Using narrowband" );
884 p_speex_mode = &speex_nb_mode;
888 /* Allocate the memory needed to store the decoder's structure */
889 if( ( p_sys = (encoder_sys_t *)malloc(sizeof(encoder_sys_t)) ) == NULL )
891 p_enc->p_sys = p_sys;
892 p_enc->pf_encode_audio = Encode;
893 p_enc->fmt_in.i_codec = VLC_CODEC_S16N;
894 p_enc->fmt_out.i_codec = VLC_CODEC_SPEEX;
896 speex_init_header( &p_sys->header, p_enc->fmt_in.audio.i_rate,
899 p_sys->header.frames_per_packet = 1;
900 p_sys->header.vbr = var_GetBool( p_enc, ENC_CFG_PREFIX "cbr" ) ? 0 : 1;
901 p_sys->header.nb_channels = p_enc->fmt_in.audio.i_channels;
903 /* Create a new encoder state in narrowband mode */
904 p_sys->p_state = speex_encoder_init( p_speex_mode );
907 i_tmp = var_GetInteger( p_enc, ENC_CFG_PREFIX "complexity" );
908 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_COMPLEXITY, &i_tmp );
910 i_tmp = var_GetBool( p_enc, ENC_CFG_PREFIX "cbr" ) ? 0 : 1;
911 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_VBR, &i_tmp );
913 if( i_tmp == 0 ) /* CBR */
915 i_tmp = var_GetFloat( p_enc, ENC_CFG_PREFIX "quality" );
916 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_QUALITY, &i_tmp );
918 i_tmp = var_GetBool( p_enc, ENC_CFG_PREFIX "vad" ) ? 1 : 0;
919 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_VAD, &i_tmp );
925 f_tmp = var_GetFloat( p_enc, ENC_CFG_PREFIX "quality" );
926 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_VBR_QUALITY, &f_tmp );
928 i_tmp = var_GetInteger( p_enc, ENC_CFG_PREFIX "max-bitrate" );
930 #ifdef SPEEX_SET_VBR_MAX_BITRATE
931 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_VBR_MAX_BITRATE, &i_tmp );
933 msg_Dbg( p_enc, "max-bitrate cannot be set in this version of libspeex");
937 i_tmp = var_GetBool( p_enc, ENC_CFG_PREFIX "dtx" ) ? 1 : 0;
938 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_DTX, &i_tmp );
941 /*Initialization of the structure that holds the bits*/
942 speex_bits_init( &p_sys->bits );
944 p_sys->i_frames_in_packet = 0;
945 p_sys->i_samples_delay = 0;
947 speex_encoder_ctl( p_sys->p_state, SPEEX_GET_FRAME_SIZE,
948 &p_sys->i_frame_length );
950 p_sys->i_frame_size = p_sys->i_frame_length *
951 sizeof(int16_t) * p_enc->fmt_in.audio.i_channels;
952 p_sys->p_buffer = xmalloc( p_sys->i_frame_size );
954 /* Create and store headers */
955 pp_header[0] = speex_header_to_packet( &p_sys->header, &pi_header[0] );
956 pp_header[1] = "ENCODER=VLC media player";
957 pi_header[1] = sizeof("ENCODER=VLC media player");
959 p_enc->fmt_out.i_extra = 3 * 2 + pi_header[0] + pi_header[1];
960 p_extra = p_enc->fmt_out.p_extra = xmalloc( p_enc->fmt_out.i_extra );
961 for( i = 0; i < 2; i++ )
963 *(p_extra++) = pi_header[i] >> 8;
964 *(p_extra++) = pi_header[i] & 0xFF;
965 memcpy( p_extra, pp_header[i], pi_header[i] );
966 p_extra += pi_header[i];
969 msg_Dbg( p_enc, "encoding: frame size:%d, channels:%d, samplerate:%d",
970 p_sys->i_frame_size, p_enc->fmt_in.audio.i_channels,
971 p_enc->fmt_in.audio.i_rate );
976 /****************************************************************************
977 * Encode: the whole thing
978 ****************************************************************************
979 * This function spits out ogg packets.
980 ****************************************************************************/
981 static block_t *Encode( encoder_t *p_enc, block_t *p_aout_buf )
983 encoder_sys_t *p_sys = p_enc->p_sys;
984 block_t *p_block, *p_chain = NULL;
986 unsigned char *p_buffer = p_aout_buf->p_buffer;
987 int i_samples = p_aout_buf->i_nb_samples;
988 int i_samples_delay = p_sys->i_samples_delay;
990 mtime_t i_pts = p_aout_buf->i_pts -
991 (mtime_t)1000000 * (mtime_t)p_sys->i_samples_delay /
992 (mtime_t)p_enc->fmt_in.audio.i_rate;
994 p_sys->i_samples_delay += i_samples;
996 while( p_sys->i_samples_delay >= p_sys->i_frame_length )
1001 if( i_samples_delay )
1003 /* Take care of the left-over from last time */
1004 int i_delay_size = i_samples_delay * 2 *
1005 p_enc->fmt_in.audio.i_channels;
1006 int i_size = p_sys->i_frame_size - i_delay_size;
1008 p_samples = (int16_t *)p_sys->p_buffer;
1009 memcpy( p_sys->p_buffer + i_delay_size, p_buffer, i_size );
1010 p_buffer -= i_delay_size;
1011 i_samples += i_samples_delay;
1012 i_samples_delay = 0;
1016 p_samples = (int16_t *)p_buffer;
1019 /* Encode current frame */
1020 if( p_enc->fmt_in.audio.i_channels == 2 )
1021 speex_encode_stereo_int( p_samples, p_sys->i_frame_length,
1025 if( p_sys->preprocess )
1026 speex_preprocess( p_sys->preprocess, p_samples, NULL );
1029 speex_encode_int( p_sys->p_state, p_samples, &p_sys->bits );
1031 p_buffer += p_sys->i_frame_size;
1032 p_sys->i_samples_delay -= p_sys->i_frame_length;
1033 i_samples -= p_sys->i_frame_length;
1035 p_sys->i_frames_in_packet++;
1037 if( p_sys->i_frames_in_packet < p_sys->header.frames_per_packet )
1040 p_sys->i_frames_in_packet = 0;
1042 speex_bits_insert_terminator( &p_sys->bits );
1043 i_out = speex_bits_write( &p_sys->bits, p_sys->p_buffer_out,
1045 speex_bits_reset( &p_sys->bits );
1047 p_block = block_New( p_enc, i_out );
1048 memcpy( p_block->p_buffer, p_sys->p_buffer_out, i_out );
1050 p_block->i_length = (mtime_t)1000000 *
1051 (mtime_t)p_sys->i_frame_length * p_sys->header.frames_per_packet /
1052 (mtime_t)p_enc->fmt_in.audio.i_rate;
1054 p_block->i_dts = p_block->i_pts = i_pts;
1057 i_pts += p_block->i_length;
1058 block_ChainAppend( &p_chain, p_block );
1062 /* Backup the remaining raw samples */
1065 memcpy( p_sys->p_buffer + i_samples_delay * 2 *
1066 p_enc->fmt_in.audio.i_channels, p_buffer,
1067 i_samples * 2 * p_enc->fmt_in.audio.i_channels );
1073 /*****************************************************************************
1074 * CloseEncoder: encoder destruction
1075 *****************************************************************************/
1076 static void CloseEncoder( vlc_object_t *p_this )
1078 encoder_t *p_enc = (encoder_t *)p_this;
1079 encoder_sys_t *p_sys = p_enc->p_sys;
1081 speex_encoder_destroy( p_sys->p_state );
1082 speex_bits_destroy( &p_sys->bits );
1084 free( p_sys->p_buffer );