1 /*****************************************************************************
2 * speex.c: speex decoder/packetizer/encoder module making use of libspeex.
3 *****************************************************************************
4 * Copyright (C) 2003-2008 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>
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, NULL, ENC_MODE_TEXT,
111 ENC_MODE_LONGTEXT, false )
112 change_integer_list( pi_enc_mode_values, ppsz_enc_mode_descriptions, NULL )
114 add_integer( ENC_CFG_PREFIX "complexity", 3, NULL, ENC_COMPLEXITY_TEXT,
115 ENC_COMPLEXITY_LONGTEXT, false )
116 change_integer_range( 1, 10 )
118 add_bool( ENC_CFG_PREFIX "cbr", false, NULL, ENC_CBR_TEXT,
119 ENC_CBR_LONGTEXT, false )
121 add_float( ENC_CFG_PREFIX "quality", 8.0, NULL, 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, NULL, ENC_MAXBITRATE_TEXT,
126 ENC_MAXBITRATE_LONGTEXT, false )
128 add_bool( ENC_CFG_PREFIX "vad", true, NULL, ENC_VAD_TEXT,
129 ENC_VAD_LONGTEXT, false )
131 add_bool( ENC_CFG_PREFIX "dtx", false, NULL, 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;
168 audio_date_t end_date;
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 void *DecodeBlock ( decoder_t *, block_t ** );
188 static aout_buffer_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 aout_buffer_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 *, aout_buffer_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 = p_dec->p_sys;
208 if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','p','x',' ')
209 && p_dec->fmt_in.i_codec != VLC_FOURCC('s', 'p', 'x', 'r') )
214 /* Allocate the memory needed to store the decoder's structure */
215 if( ( p_dec->p_sys = p_sys =
216 (decoder_sys_t *)malloc(sizeof(decoder_sys_t)) ) == NULL )
218 p_dec->p_sys->bits.buf_size = 0;
219 p_dec->p_sys->b_packetizer = false;
220 p_dec->p_sys->rtp_rate = p_dec->fmt_in.audio.i_rate;
222 aout_DateSet( &p_sys->end_date, 0 );
224 /* Set output properties */
225 p_dec->fmt_out.i_cat = AUDIO_ES;
226 p_dec->fmt_out.i_codec = AOUT_FMT_S16_NE;
230 If the codec is spxr then this decoder is
231 being invoked on a Speex stream arriving via RTP.
232 A special decoder callback is used.
234 if (p_dec->fmt_in.i_codec == VLC_FOURCC('s', 'p', 'x', 'r'))
236 msg_Dbg( p_dec, "Using RTP version of Speex decoder @ rate %d.",
237 p_dec->fmt_in.audio.i_rate );
238 p_dec->pf_decode_audio = (aout_buffer_t *(*)(decoder_t *, block_t **))
239 DecodeRtpSpeexPacket;
243 p_dec->pf_decode_audio = (aout_buffer_t *(*)(decoder_t *, block_t **))
246 p_dec->pf_packetize = (block_t *(*)(decoder_t *, block_t **))
249 p_sys->i_headers = 0;
250 p_sys->p_state = NULL;
251 p_sys->p_header = NULL;
252 p_sys->i_frame_in_packet = 0;
257 static int OpenPacketizer( vlc_object_t *p_this )
259 decoder_t *p_dec = (decoder_t*)p_this;
261 int i_ret = OpenDecoder( p_this );
263 if( i_ret == VLC_SUCCESS )
265 p_dec->p_sys->b_packetizer = true;
266 p_dec->fmt_out.i_codec = VLC_FOURCC('s','p','x',' ');
272 /****************************************************************************
273 * DecodeBlock: the whole thing
274 ****************************************************************************
275 * This function must be fed with ogg packets.
276 ****************************************************************************/
277 static void *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
279 decoder_sys_t *p_sys = p_dec->p_sys;
280 ogg_packet oggpacket;
282 if( !pp_block ) return NULL;
286 /* Block to Ogg packet */
287 oggpacket.packet = (*pp_block)->p_buffer;
288 oggpacket.bytes = (*pp_block)->i_buffer;
292 if( p_sys->b_packetizer ) return NULL;
294 /* Block to Ogg packet */
295 oggpacket.packet = NULL;
299 oggpacket.granulepos = -1;
302 oggpacket.packetno = 0;
304 /* Check for headers */
305 if( p_sys->i_headers == 0 && p_dec->fmt_in.i_extra )
307 p_sys->i_headers = 2;
309 else if( oggpacket.bytes && p_sys->i_headers < 2 )
313 p_dec->fmt_in.p_extra =
314 realloc( p_dec->fmt_in.p_extra, p_dec->fmt_in.i_extra +
315 oggpacket.bytes + 2 );
316 p_extra = ((uint8_t *)p_dec->fmt_in.p_extra) + p_dec->fmt_in.i_extra;
317 *(p_extra++) = oggpacket.bytes >> 8;
318 *(p_extra++) = oggpacket.bytes & 0xFF;
320 memcpy( p_extra, oggpacket.packet, oggpacket.bytes );
321 p_dec->fmt_in.i_extra += oggpacket.bytes + 2;
323 block_Release( *pp_block );
328 if( p_sys->i_headers == 2 )
330 if( ProcessHeaders( p_dec ) != VLC_SUCCESS )
332 p_sys->i_headers = 0;
333 p_dec->fmt_in.i_extra = 0;
334 block_Release( *pp_block );
337 else p_sys->i_headers++;
340 return ProcessPacket( p_dec, &oggpacket, pp_block );
343 /*****************************************************************************
344 * ProcessHeaders: process Speex headers.
345 *****************************************************************************/
346 static int ProcessHeaders( decoder_t *p_dec )
348 decoder_sys_t *p_sys = p_dec->p_sys;
349 ogg_packet oggpacket;
353 if( !p_dec->fmt_in.i_extra ) return VLC_EGENERIC;
355 oggpacket.granulepos = -1;
356 oggpacket.b_o_s = 1; /* yes this actually is a b_o_s packet :) */
358 oggpacket.packetno = 0;
359 p_extra = p_dec->fmt_in.p_extra;
360 i_extra = p_dec->fmt_in.i_extra;
362 /* Take care of the initial Vorbis header */
363 oggpacket.bytes = *(p_extra++) << 8;
364 oggpacket.bytes |= (*(p_extra++) & 0xFF);
365 oggpacket.packet = p_extra;
366 p_extra += oggpacket.bytes;
367 i_extra -= (oggpacket.bytes + 2);
370 msg_Err( p_dec, "header data corrupted");
374 /* Take care of the initial Speex header */
375 if( ProcessInitialHeader( p_dec, &oggpacket ) != VLC_SUCCESS )
377 msg_Err( p_dec, "initial Speex header is corrupted" );
381 /* The next packet in order is the comments header */
383 oggpacket.bytes = *(p_extra++) << 8;
384 oggpacket.bytes |= (*(p_extra++) & 0xFF);
385 oggpacket.packet = p_extra;
386 p_extra += oggpacket.bytes;
387 i_extra -= (oggpacket.bytes + 2);
390 msg_Err( p_dec, "header data corrupted");
394 ParseSpeexComments( p_dec, &oggpacket );
396 if( p_sys->b_packetizer )
398 p_dec->fmt_out.i_extra = p_dec->fmt_in.i_extra;
399 p_dec->fmt_out.p_extra =
400 realloc( p_dec->fmt_out.p_extra, p_dec->fmt_out.i_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 aout_DateInit( &p_sys->end_date, p_header->rate );
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 != aout_DateGet( &p_sys->end_date ) )
514 aout_DateSet( &p_sys->end_date, p_block->i_pts );
517 if( !aout_DateGet( &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 );
538 speex_bits_read_from( &p_sys->bits, (char*)p_oggpacket->packet,
540 i_bits_before = speex_bits_remaining( &p_sys->bits );
541 speex_decode_int(p_sys->p_state, &p_sys->bits, p_frame_holder);
542 i_bits_after = speex_bits_remaining( &p_sys->bits );
544 i_bits_in_speex_frame = i_bits_before - i_bits_after;
545 i_bytes_in_speex_frame = ( i_bits_in_speex_frame +
546 (8 - (i_bits_in_speex_frame % 8)) )
549 p_new_block = block_New( p_dec, i_bytes_in_speex_frame );
550 memset( p_new_block->p_buffer, 0xff, i_bytes_in_speex_frame );
553 * Copy the first frame in this packet to a new packet.
555 speex_bits_rewind( &p_sys->bits );
556 speex_bits_write( &p_sys->bits,
557 (char*)p_new_block->p_buffer,
558 (int)i_bytes_in_speex_frame );
561 * Move the remaining part of the original packet (subsequent
562 * frames, if there are any) into the beginning
563 * of the original packet so
564 * they are preserved following the realloc.
565 * Note: Any bits that
566 * remain in the initial packet
567 * are "filler" if they do not constitute
570 if ( i_bits_after > 7 )
572 /* round-down since we rounded-up earlier (to include
573 * the speex terminator code.
575 i_bytes_in_speex_frame--;
576 speex_bits_write( &p_sys->bits,
577 (char*)p_block->p_buffer,
578 p_block->i_buffer - i_bytes_in_speex_frame );
579 p_block = block_Realloc( p_block,
581 p_block->i_buffer-i_bytes_in_speex_frame );
586 speex_bits_reset( &p_sys->bits );
589 free( p_frame_holder );
590 return SendPacket( p_dec, p_new_block);
594 return SendPacket( p_dec, p_block );
599 aout_buffer_t *p_aout_buffer;
601 if( p_sys->i_headers >= p_sys->p_header->extra_headers + 2 )
602 p_aout_buffer = DecodePacket( p_dec, p_oggpacket );
604 p_aout_buffer = NULL; /* Skip headers */
606 if( p_block ) block_Release( p_block );
607 return p_aout_buffer;
611 static aout_buffer_t *DecodeRtpSpeexPacket( decoder_t *p_dec, block_t **pp_block )
613 block_t *p_speex_bit_block = *pp_block;
614 decoder_sys_t *p_sys = p_dec->p_sys;
615 aout_buffer_t *p_aout_buffer;
617 unsigned int i_speex_frame_size;
619 if ( !p_speex_bit_block || p_speex_bit_block->i_pts == 0 ) return NULL;
622 If the SpeexBits buffer size is 0 (a default value),
623 we know that a proper initialization has not yet been done.
625 if ( p_sys->bits.buf_size==0 )
627 p_sys->p_header = (SpeexHeader *)malloc(sizeof(SpeexHeader));
628 if ( !p_sys->p_header )
630 msg_Err( p_dec, "Could not allocate a Speex header.");
633 speex_init_header( p_sys->p_header,p_sys->rtp_rate,1,&speex_nb_mode );
634 speex_bits_init( &p_sys->bits );
635 p_sys->p_state = speex_decoder_init( &speex_nb_mode );
636 if ( !p_sys->p_state )
638 msg_Err( p_dec, "Could not allocate a Speex decoder." );
639 free( p_sys->p_header );
644 Assume that variable bit rate is enabled. Also assume
645 that there is only one frame per packet.
647 p_sys->p_header->vbr = 1;
648 p_sys->p_header->frames_per_packet = 1;
650 p_dec->fmt_out.audio.i_channels = p_sys->p_header->nb_channels;
651 p_dec->fmt_out.audio.i_physical_channels =
652 p_dec->fmt_out.audio.i_original_channels =
653 pi_channels_maps[p_sys->p_header->nb_channels];
654 p_dec->fmt_out.audio.i_rate = p_sys->p_header->rate;
656 if ( speex_mode_query( &speex_nb_mode,
657 SPEEX_MODE_FRAME_SIZE,
658 &i_speex_frame_size ) )
660 msg_Err( p_dec, "Could not determine the frame size." );
661 speex_decoder_destroy( p_sys->p_state );
662 free( p_sys->p_header );
665 p_dec->fmt_out.audio.i_bytes_per_frame = i_speex_frame_size;
667 aout_DateInit(&p_sys->end_date, p_sys->p_header->rate);
671 If the SpeexBits are initialized but there is
672 still no header, an error must be thrown.
674 if ( !p_sys->p_header )
676 msg_Err( p_dec, "There is no valid Speex header found." );
681 if ( !aout_DateGet( &p_sys->end_date ) )
682 aout_DateSet( &p_sys->end_date, p_speex_bit_block->i_dts );
685 Ask for a new audio output buffer and make sure
688 p_aout_buffer = decoder_NewAudioBuffer( p_dec,
689 p_sys->p_header->frame_size );
690 if ( !p_aout_buffer || p_aout_buffer->i_nb_bytes == 0 )
692 msg_Err(p_dec, "Oops: No new buffer was returned!");
697 Read the Speex payload into the SpeexBits buffer.
699 speex_bits_read_from( &p_sys->bits,
700 (char*)p_speex_bit_block->p_buffer,
701 p_speex_bit_block->i_buffer );
704 Decode the input and ensure that no errors
707 i_decode_ret = speex_decode_int( p_sys->p_state, &p_sys->bits,
708 (int16_t*)p_aout_buffer->p_buffer );
709 if ( i_decode_ret < 0 )
711 msg_Err( p_dec, "Decoding failed. Perhaps we have a bad stream?" );
716 Handle date management on the audio output buffer.
718 p_aout_buffer->start_date = aout_DateGet( &p_sys->end_date );
719 p_aout_buffer->end_date = aout_DateIncrement( &p_sys->end_date,
720 p_sys->p_header->frame_size );
723 p_sys->i_frame_in_packet++;
724 block_Release( p_speex_bit_block );
726 return p_aout_buffer;
729 /*****************************************************************************
730 * DecodePacket: decodes a Speex packet.
731 *****************************************************************************/
732 static aout_buffer_t *DecodePacket( decoder_t *p_dec, ogg_packet *p_oggpacket )
734 decoder_sys_t *p_sys = p_dec->p_sys;
736 if( p_oggpacket->bytes )
738 /* Copy Ogg packet to Speex bitstream */
739 speex_bits_read_from( &p_sys->bits, (char *)p_oggpacket->packet,
740 p_oggpacket->bytes );
741 p_sys->i_frame_in_packet = 0;
744 /* Decode one frame at a time */
745 if( p_sys->i_frame_in_packet < p_sys->p_header->frames_per_packet )
747 aout_buffer_t *p_aout_buffer;
748 if( p_sys->p_header->frame_size == 0 )
752 decoder_NewAudioBuffer( p_dec, p_sys->p_header->frame_size );
758 switch( speex_decode_int( p_sys->p_state, &p_sys->bits,
759 (int16_t *)p_aout_buffer->p_buffer ) )
762 msg_Err( p_dec, "decoding error: corrupted stream?" );
763 case -1: /* End of stream */
767 if( speex_bits_remaining( &p_sys->bits ) < 0 )
769 msg_Err( p_dec, "decoding overflow: corrupted stream?" );
772 if( p_sys->p_header->nb_channels == 2 )
773 speex_decode_stereo_int( (int16_t *)p_aout_buffer->p_buffer,
774 p_sys->p_header->frame_size,
777 /* Date management */
778 p_aout_buffer->start_date = aout_DateGet( &p_sys->end_date );
779 p_aout_buffer->end_date =
780 aout_DateIncrement( &p_sys->end_date, p_sys->p_header->frame_size );
782 p_sys->i_frame_in_packet++;
784 return p_aout_buffer;
792 /*****************************************************************************
793 * SendPacket: send an ogg packet to the stream output.
794 *****************************************************************************/
795 static block_t *SendPacket( decoder_t *p_dec, block_t *p_block )
797 decoder_sys_t *p_sys = p_dec->p_sys;
799 /* Date management */
800 p_block->i_dts = p_block->i_pts = aout_DateGet( &p_sys->end_date );
802 if( p_sys->i_headers >= p_sys->p_header->extra_headers + 2 )
805 aout_DateIncrement( &p_sys->end_date,
806 p_sys->p_header->frame_size ) -
810 p_block->i_length = 0;
815 /*****************************************************************************
816 * ParseSpeexComments:
817 *****************************************************************************/
818 #define readint(buf, base) (((buf[base+3]<<24)&0xff000000)| \
819 ((buf[base+2]<<16)&0xff0000)| \
820 ((buf[base+1]<<8)&0xff00)| \
823 static void ParseSpeexComments( decoder_t *p_dec, ogg_packet *p_oggpacket )
825 decoder_sys_t *p_sys = p_dec->p_sys;
826 const SpeexMode *p_mode;
828 assert( p_sys->p_header->mode < SPEEX_NB_MODES );
830 p_mode = speex_mode_list[p_sys->p_header->mode];
831 assert( p_mode != NULL );
833 if( !p_dec->p_description )
835 p_dec->p_description = vlc_meta_New();
836 if( !p_dec->p_description )
842 if( asprintf( &psz_mode, "%s%s", p_mode->modeName, p_sys->p_header->vbr ? " VBR" : "" ) >= 0 )
844 vlc_meta_AddExtra( p_dec->p_description, _("Mode"), psz_mode );
848 /* TODO: finish comments parsing */
849 VLC_UNUSED( p_oggpacket );
852 /*****************************************************************************
853 * CloseDecoder: speex decoder destruction
854 *****************************************************************************/
855 static void CloseDecoder( vlc_object_t *p_this )
857 decoder_t * p_dec = (decoder_t *)p_this;
858 decoder_sys_t *p_sys = p_dec->p_sys;
862 speex_decoder_destroy( p_sys->p_state );
863 speex_bits_destroy( &p_sys->bits );
866 free( p_sys->p_header );
870 /*****************************************************************************
871 * encoder_sys_t: encoder descriptor
872 *****************************************************************************/
873 #define MAX_FRAME_SIZE 2000
874 #define MAX_FRAME_BYTES 2000
882 char p_buffer_out[MAX_FRAME_BYTES];
889 SpeexStereoState stereo;
892 int i_frames_per_packet;
893 int i_frames_in_packet;
905 /*****************************************************************************
906 * OpenEncoder: probe the encoder and return score
907 *****************************************************************************/
908 static int OpenEncoder( vlc_object_t *p_this )
910 encoder_t *p_enc = (encoder_t *)p_this;
911 encoder_sys_t *p_sys;
912 const SpeexMode *p_speex_mode = &speex_nb_mode;
914 const char *pp_header[2];
918 if( p_enc->fmt_out.i_codec != VLC_FOURCC('s','p','x',' ') &&
924 config_ChainParse( p_enc, ENC_CFG_PREFIX, ppsz_enc_options, p_enc->p_cfg );
925 switch( var_GetInteger( p_enc, ENC_CFG_PREFIX "mode" ) )
928 msg_Dbg( p_enc, "Using wideband" );
929 p_speex_mode = &speex_wb_mode;
932 msg_Dbg( p_enc, "Using ultra-wideband" );
933 p_speex_mode = &speex_uwb_mode;
936 msg_Dbg( p_enc, "Using narrowband" );
937 p_speex_mode = &speex_nb_mode;
941 /* Allocate the memory needed to store the decoder's structure */
942 if( ( p_sys = (encoder_sys_t *)malloc(sizeof(encoder_sys_t)) ) == NULL )
944 p_enc->p_sys = p_sys;
945 p_enc->pf_encode_audio = Encode;
946 p_enc->fmt_in.i_codec = AOUT_FMT_S16_NE;
947 p_enc->fmt_out.i_codec = VLC_FOURCC('s','p','x',' ');
949 speex_init_header( &p_sys->header, p_enc->fmt_in.audio.i_rate,
952 p_sys->header.frames_per_packet = 1;
953 p_sys->header.vbr = var_GetBool( p_enc, ENC_CFG_PREFIX "cbr" ) ? 0 : 1;
954 p_sys->header.nb_channels = p_enc->fmt_in.audio.i_channels;
956 /* Create a new encoder state in narrowband mode */
957 p_sys->p_state = speex_encoder_init( p_speex_mode );
960 i_tmp = var_GetInteger( p_enc, ENC_CFG_PREFIX "complexity" );
961 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_COMPLEXITY, &i_tmp );
963 i_tmp = var_GetBool( p_enc, ENC_CFG_PREFIX "cbr" ) ? 0 : 1;
964 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_VBR, &i_tmp );
966 if( i_tmp == 0 ) /* CBR */
968 i_tmp = var_GetFloat( p_enc, ENC_CFG_PREFIX "quality" );
969 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_QUALITY, &i_tmp );
971 i_tmp = var_GetBool( p_enc, ENC_CFG_PREFIX "vad" ) ? 1 : 0;
972 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_VAD, &i_tmp );
978 f_tmp = var_GetFloat( p_enc, ENC_CFG_PREFIX "quality" );
979 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_VBR_QUALITY, &f_tmp );
981 i_tmp = var_GetInteger( p_enc, ENC_CFG_PREFIX "max-bitrate" );
983 #ifdef SPEEX_SET_VBR_MAX_BITRATE
984 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_VBR_MAX_BITRATE, &i_tmp );
986 msg_Dbg( p_enc, "max-bitrate cannot be set in this version of libspeex");
990 i_tmp = var_GetBool( p_enc, ENC_CFG_PREFIX "dtx" ) ? 1 : 0;
991 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_DTX, &i_tmp );
994 /*Initialization of the structure that holds the bits*/
995 speex_bits_init( &p_sys->bits );
997 p_sys->i_frames_in_packet = 0;
998 p_sys->i_samples_delay = 0;
1001 speex_encoder_ctl( p_sys->p_state, SPEEX_GET_FRAME_SIZE,
1002 &p_sys->i_frame_length );
1004 p_sys->i_frame_size = p_sys->i_frame_length *
1005 sizeof(int16_t) * p_enc->fmt_in.audio.i_channels;
1006 p_sys->p_buffer = malloc( p_sys->i_frame_size );
1008 /* Create and store headers */
1009 pp_header[0] = speex_header_to_packet( &p_sys->header, &pi_header[0] );
1010 pp_header[1] = "ENCODER=VLC media player";
1011 pi_header[1] = sizeof("ENCODER=VLC media player");
1013 p_enc->fmt_out.i_extra = 3 * 2 + pi_header[0] + pi_header[1];
1014 p_extra = p_enc->fmt_out.p_extra = malloc( p_enc->fmt_out.i_extra );
1015 for( i = 0; i < 2; i++ )
1017 *(p_extra++) = pi_header[i] >> 8;
1018 *(p_extra++) = pi_header[i] & 0xFF;
1019 memcpy( p_extra, pp_header[i], pi_header[i] );
1020 p_extra += pi_header[i];
1023 msg_Dbg( p_enc, "encoding: frame size:%d, channels:%d, samplerate:%d",
1024 p_sys->i_frame_size, p_enc->fmt_in.audio.i_channels,
1025 p_enc->fmt_in.audio.i_rate );
1030 /****************************************************************************
1031 * Encode: the whole thing
1032 ****************************************************************************
1033 * This function spits out ogg packets.
1034 ****************************************************************************/
1035 static block_t *Encode( encoder_t *p_enc, aout_buffer_t *p_aout_buf )
1037 encoder_sys_t *p_sys = p_enc->p_sys;
1038 block_t *p_block, *p_chain = NULL;
1040 unsigned char *p_buffer = p_aout_buf->p_buffer;
1041 int i_samples = p_aout_buf->i_nb_samples;
1042 int i_samples_delay = p_sys->i_samples_delay;
1044 p_sys->i_pts = p_aout_buf->start_date -
1045 (mtime_t)1000000 * (mtime_t)p_sys->i_samples_delay /
1046 (mtime_t)p_enc->fmt_in.audio.i_rate;
1048 p_sys->i_samples_delay += i_samples;
1050 while( p_sys->i_samples_delay >= p_sys->i_frame_length )
1055 if( i_samples_delay )
1057 /* Take care of the left-over from last time */
1058 int i_delay_size = i_samples_delay * 2 *
1059 p_enc->fmt_in.audio.i_channels;
1060 int i_size = p_sys->i_frame_size - i_delay_size;
1062 p_samples = (int16_t *)p_sys->p_buffer;
1063 memcpy( p_sys->p_buffer + i_delay_size, p_buffer, i_size );
1064 p_buffer -= i_delay_size;
1065 i_samples += i_samples_delay;
1066 i_samples_delay = 0;
1070 p_samples = (int16_t *)p_buffer;
1073 /* Encode current frame */
1074 if( p_enc->fmt_in.audio.i_channels == 2 )
1075 speex_encode_stereo_int( p_samples, p_sys->i_frame_length,
1079 if( p_sys->preprocess )
1080 speex_preprocess( p_sys->preprocess, p_samples, NULL );
1083 speex_encode_int( p_sys->p_state, p_samples, &p_sys->bits );
1085 p_buffer += p_sys->i_frame_size;
1086 p_sys->i_samples_delay -= p_sys->i_frame_length;
1087 i_samples -= p_sys->i_frame_length;
1089 p_sys->i_frames_in_packet++;
1091 if( p_sys->i_frames_in_packet < p_sys->header.frames_per_packet )
1094 p_sys->i_frames_in_packet = 0;
1096 speex_bits_insert_terminator( &p_sys->bits );
1097 i_out = speex_bits_write( &p_sys->bits, p_sys->p_buffer_out,
1099 speex_bits_reset( &p_sys->bits );
1101 p_block = block_New( p_enc, i_out );
1102 memcpy( p_block->p_buffer, p_sys->p_buffer_out, i_out );
1104 p_block->i_length = (mtime_t)1000000 *
1105 (mtime_t)p_sys->i_frame_length * p_sys->header.frames_per_packet /
1106 (mtime_t)p_enc->fmt_in.audio.i_rate;
1108 p_block->i_dts = p_block->i_pts = p_sys->i_pts;
1111 p_sys->i_pts += p_block->i_length;
1112 block_ChainAppend( &p_chain, p_block );
1116 /* Backup the remaining raw samples */
1119 memcpy( p_sys->p_buffer + i_samples_delay * 2 *
1120 p_enc->fmt_in.audio.i_channels, p_buffer,
1121 i_samples * 2 * p_enc->fmt_in.audio.i_channels );
1127 /*****************************************************************************
1128 * CloseEncoder: encoder destruction
1129 *****************************************************************************/
1130 static void CloseEncoder( vlc_object_t *p_this )
1132 encoder_t *p_enc = (encoder_t *)p_this;
1133 encoder_sys_t *p_sys = p_enc->p_sys;
1135 speex_encoder_destroy( p_sys->p_state );
1136 speex_bits_destroy( &p_sys->bits );
1138 free( p_sys->p_buffer );