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_callbacks( OpenDecoder, CloseDecoder );
100 set_description( N_("Speex audio packetizer") );
101 set_capability( "packetizer", 100 );
102 set_callbacks( OpenPacketizer, CloseDecoder );
105 set_description( N_("Speex audio encoder") );
106 set_capability( "encoder", 100 );
107 set_callbacks( OpenEncoder, CloseEncoder );
109 add_integer( ENC_CFG_PREFIX "mode", 0, NULL, ENC_MODE_TEXT,
110 ENC_MODE_LONGTEXT, false );
111 change_integer_list( pi_enc_mode_values, ppsz_enc_mode_descriptions, NULL );
113 add_integer( ENC_CFG_PREFIX "complexity", 3, NULL, ENC_COMPLEXITY_TEXT,
114 ENC_COMPLEXITY_LONGTEXT, false );
115 change_integer_range( 1, 10 );
117 add_bool( ENC_CFG_PREFIX "cbr", false, NULL, ENC_CBR_TEXT,
118 ENC_CBR_LONGTEXT, false );
120 add_float( ENC_CFG_PREFIX "quality", 8.0, NULL, ENC_QUALITY_TEXT,
121 ENC_QUALITY_LONGTEXT, false );
122 change_float_range( 0.0, 10.0 );
124 add_integer( ENC_CFG_PREFIX "max-bitrate", 0, NULL, ENC_MAXBITRATE_TEXT,
125 ENC_MAXBITRATE_LONGTEXT, false );
127 add_bool( ENC_CFG_PREFIX "vad", true, NULL, ENC_VAD_TEXT,
128 ENC_VAD_LONGTEXT, false );
130 add_bool( ENC_CFG_PREFIX "dtx", false, NULL, ENC_DTX_TEXT,
131 ENC_DTX_LONGTEXT, false );
133 /* TODO agc, noise suppression, */
137 static const char *const ppsz_enc_options[] = {
138 "mode", "complexity", "cbr", "quality", "max-bitrate", "vad", "dtx", NULL
141 /*****************************************************************************
142 * decoder_sys_t : speex decoder descriptor
143 *****************************************************************************/
153 int i_frame_in_packet;
159 SpeexHeader *p_header;
160 SpeexStereoState stereo;
162 unsigned int rtp_rate;
167 audio_date_t end_date;
171 static const int pi_channels_maps[6] =
174 AOUT_CHAN_CENTER, AOUT_CHAN_LEFT | AOUT_CHAN_RIGHT,
175 AOUT_CHAN_CENTER | AOUT_CHAN_LEFT | AOUT_CHAN_RIGHT,
176 AOUT_CHAN_LEFT | AOUT_CHAN_RIGHT | AOUT_CHAN_REARLEFT
177 | AOUT_CHAN_REARRIGHT,
178 AOUT_CHAN_LEFT | AOUT_CHAN_RIGHT | AOUT_CHAN_CENTER
179 | AOUT_CHAN_REARLEFT | AOUT_CHAN_REARRIGHT
182 /****************************************************************************
184 ****************************************************************************/
186 static void *DecodeBlock ( decoder_t *, block_t ** );
187 static aout_buffer_t *DecodeRtpSpeexPacket( decoder_t *, block_t **);
188 static int ProcessHeaders( decoder_t * );
189 static int ProcessInitialHeader ( decoder_t *, ogg_packet * );
190 static void *ProcessPacket( decoder_t *, ogg_packet *, block_t ** );
192 static aout_buffer_t *DecodePacket( decoder_t *, ogg_packet * );
193 static block_t *SendPacket( decoder_t *, block_t * );
195 static void ParseSpeexComments( decoder_t *, ogg_packet * );
197 static block_t *Encode ( encoder_t *, aout_buffer_t * );
199 /*****************************************************************************
200 * OpenDecoder: probe the decoder and return score
201 *****************************************************************************/
202 static int OpenDecoder( vlc_object_t *p_this )
204 decoder_t *p_dec = (decoder_t*)p_this;
205 decoder_sys_t *p_sys = p_dec->p_sys;
207 if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','p','x',' ')
208 && p_dec->fmt_in.i_codec != VLC_FOURCC('s', 'p', 'x', 'r') )
213 /* Allocate the memory needed to store the decoder's structure */
214 if( ( p_dec->p_sys = p_sys =
215 (decoder_sys_t *)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 aout_DateSet( &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 = AOUT_FMT_S16_NE;
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_codec == 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_FOURCC('s','p','x',' ');
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 =
313 realloc( p_dec->fmt_in.p_extra, p_dec->fmt_in.i_extra +
314 oggpacket.bytes + 2 );
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 =
399 realloc( p_dec->fmt_out.p_extra, p_dec->fmt_out.i_extra );
400 memcpy( p_dec->fmt_out.p_extra,
401 p_dec->fmt_in.p_extra, p_dec->fmt_out.i_extra );
407 /*****************************************************************************
408 * ProcessInitialHeader: processes the inital Speex header packet.
409 *****************************************************************************/
410 static int ProcessInitialHeader( decoder_t *p_dec, ogg_packet *p_oggpacket )
412 decoder_sys_t *p_sys = p_dec->p_sys;
415 SpeexHeader *p_header;
416 const SpeexMode *p_mode;
417 SpeexCallback callback;
419 p_sys->p_header = p_header =
420 speex_packet_to_header( (char *)p_oggpacket->packet,
421 p_oggpacket->bytes );
424 msg_Err( p_dec, "cannot read Speex header" );
427 if( p_header->mode >= SPEEX_NB_MODES || p_header->mode < 0 )
429 msg_Err( p_dec, "mode number %d does not (yet/any longer) exist in "
430 "this version of libspeex.", p_header->mode );
434 p_mode = speex_mode_list[p_header->mode];
438 if( p_header->speex_version_id > 1 )
440 msg_Err( p_dec, "this file was encoded with Speex bit-stream "
441 "version %d which is not supported by this decoder.",
442 p_header->speex_version_id );
446 if( p_mode->bitstream_version < p_header->mode_bitstream_version )
448 msg_Err( p_dec, "file encoded with a newer version of Speex." );
451 if( p_mode->bitstream_version > p_header->mode_bitstream_version )
453 msg_Err( p_dec, "file encoded with an older version of Speex." );
457 msg_Dbg( p_dec, "Speex %d Hz audio using %s mode %s%s",
458 p_header->rate, p_mode->modeName,
459 ( p_header->nb_channels == 1 ) ? " (mono" : " (stereo",
460 p_header->vbr ? ", VBR)" : ")" );
462 /* Take care of speex decoder init */
463 speex_bits_init( &p_sys->bits );
464 p_sys->p_state = p_state = speex_decoder_init( p_mode );
467 msg_Err( p_dec, "decoder initialization failed" );
471 if( p_header->nb_channels == 2 )
473 SpeexStereoState stereo = SPEEX_STEREO_STATE_INIT;
474 p_sys->stereo = stereo;
475 callback.callback_id = SPEEX_INBAND_STEREO;
476 callback.func = speex_std_stereo_request_handler;
477 callback.data = &p_sys->stereo;
478 speex_decoder_ctl( p_state, SPEEX_SET_HANDLER, &callback );
480 if( p_header->nb_channels <= 0 ||
481 p_header->nb_channels > 5 )
483 msg_Err( p_dec, "invalid number of channels (not between 1 and 5): %i",
484 p_header->nb_channels );
488 /* Setup the format */
489 p_dec->fmt_out.audio.i_physical_channels =
490 p_dec->fmt_out.audio.i_original_channels =
491 pi_channels_maps[p_header->nb_channels];
492 p_dec->fmt_out.audio.i_channels = p_header->nb_channels;
493 p_dec->fmt_out.audio.i_rate = p_header->rate;
495 aout_DateInit( &p_sys->end_date, p_header->rate );
500 /*****************************************************************************
501 * ProcessPacket: processes a Speex packet.
502 *****************************************************************************/
503 static void *ProcessPacket( decoder_t *p_dec, ogg_packet *p_oggpacket,
506 decoder_sys_t *p_sys = p_dec->p_sys;
507 block_t *p_block = *pp_block;
509 /* Date management */
510 if( p_block && p_block->i_pts > 0 &&
511 p_block->i_pts != aout_DateGet( &p_sys->end_date ) )
513 aout_DateSet( &p_sys->end_date, p_block->i_pts );
516 if( !aout_DateGet( &p_sys->end_date ) )
518 /* We've just started the stream, wait for the first PTS. */
519 if( p_block ) block_Release( p_block );
523 *pp_block = NULL; /* To avoid being fed the same packet again */
525 if( p_sys->b_packetizer )
527 if ( p_sys->p_header->frames_per_packet > 1 )
529 short *p_frame_holder = NULL;
530 int i_bits_before = 0, i_bits_after = 0, i_bytes_in_speex_frame = 0,
531 i_pcm_output_size = 0, i_bits_in_speex_frame = 0;
532 block_t *p_new_block = NULL;
534 i_pcm_output_size = p_sys->p_header->frame_size;
535 p_frame_holder = (short*)malloc( sizeof(short)*i_pcm_output_size );
537 speex_bits_read_from( &p_sys->bits, (char*)p_oggpacket->packet,
539 i_bits_before = speex_bits_remaining( &p_sys->bits );
540 speex_decode_int(p_sys->p_state, &p_sys->bits, p_frame_holder);
541 i_bits_after = speex_bits_remaining( &p_sys->bits );
543 i_bits_in_speex_frame = i_bits_before - i_bits_after;
544 i_bytes_in_speex_frame = ( i_bits_in_speex_frame +
545 (8 - (i_bits_in_speex_frame % 8)) )
548 p_new_block = block_New( p_dec, i_bytes_in_speex_frame );
549 memset( p_new_block->p_buffer, 0xff, i_bytes_in_speex_frame );
552 * Copy the first frame in this packet to a new packet.
554 speex_bits_rewind( &p_sys->bits );
555 speex_bits_write( &p_sys->bits,
556 (char*)p_new_block->p_buffer,
557 (int)i_bytes_in_speex_frame );
560 * Move the remaining part of the original packet (subsequent
561 * frames, if there are any) into the beginning
562 * of the original packet so
563 * they are preserved following the realloc.
564 * Note: Any bits that
565 * remain in the initial packet
566 * are "filler" if they do not constitute
569 if ( i_bits_after > 7 )
571 /* round-down since we rounded-up earlier (to include
572 * the speex terminator code.
574 i_bytes_in_speex_frame--;
575 speex_bits_write( &p_sys->bits,
576 (char*)p_block->p_buffer,
577 p_block->i_buffer - i_bytes_in_speex_frame );
578 p_block = block_Realloc( p_block,
580 p_block->i_buffer-i_bytes_in_speex_frame );
585 speex_bits_reset( &p_sys->bits );
588 free( p_frame_holder );
589 return SendPacket( p_dec, p_new_block);
593 return SendPacket( p_dec, p_block );
598 aout_buffer_t *p_aout_buffer;
600 if( p_sys->i_headers >= p_sys->p_header->extra_headers + 2 )
601 p_aout_buffer = DecodePacket( p_dec, p_oggpacket );
603 p_aout_buffer = NULL; /* Skip headers */
605 if( p_block ) block_Release( p_block );
606 return p_aout_buffer;
610 static aout_buffer_t *DecodeRtpSpeexPacket( decoder_t *p_dec, block_t **pp_block )
612 block_t *p_speex_bit_block = *pp_block;
613 decoder_sys_t *p_sys = p_dec->p_sys;
614 aout_buffer_t *p_aout_buffer;
616 unsigned int i_speex_frame_size;
618 if ( !p_speex_bit_block || p_speex_bit_block->i_pts == 0 ) return NULL;
621 If the SpeexBits buffer size is 0 (a default value),
622 we know that a proper initialization has not yet been done.
624 if ( p_sys->bits.buf_size==0 )
626 p_sys->p_header = (SpeexHeader *)malloc(sizeof(SpeexHeader));
627 if ( !p_sys->p_header )
629 msg_Err( p_dec, "Could not allocate a Speex header.");
632 speex_init_header( p_sys->p_header,p_sys->rtp_rate,1,&speex_nb_mode );
633 speex_bits_init( &p_sys->bits );
634 p_sys->p_state = speex_decoder_init( &speex_nb_mode );
635 if ( !p_sys->p_state )
637 msg_Err( p_dec, "Could not allocate a Speex decoder." );
638 free( p_sys->p_header );
643 Assume that variable bit rate is enabled. Also assume
644 that there is only one frame per packet.
646 p_sys->p_header->vbr = 1;
647 p_sys->p_header->frames_per_packet = 1;
649 p_dec->fmt_out.audio.i_channels = p_sys->p_header->nb_channels;
650 p_dec->fmt_out.audio.i_physical_channels =
651 p_dec->fmt_out.audio.i_original_channels =
652 pi_channels_maps[p_sys->p_header->nb_channels];
653 p_dec->fmt_out.audio.i_rate = p_sys->p_header->rate;
655 if ( speex_mode_query( &speex_nb_mode,
656 SPEEX_MODE_FRAME_SIZE,
657 &i_speex_frame_size ) )
659 msg_Err( p_dec, "Could not determine the frame size." );
660 speex_decoder_destroy( p_sys->p_state );
661 free( p_sys->p_header );
664 p_dec->fmt_out.audio.i_bytes_per_frame = i_speex_frame_size;
666 aout_DateInit(&p_sys->end_date, p_sys->p_header->rate);
670 If the SpeexBits are initialized but there is
671 still no header, an error must be thrown.
673 if ( !p_sys->p_header )
675 msg_Err( p_dec, "There is no valid Speex header found." );
680 if ( !aout_DateGet( &p_sys->end_date ) )
681 aout_DateSet( &p_sys->end_date, p_speex_bit_block->i_dts );
684 Ask for a new audio output buffer and make sure
687 p_aout_buffer = p_dec->pf_aout_buffer_new( p_dec,
688 p_sys->p_header->frame_size );
689 if ( !p_aout_buffer || p_aout_buffer->i_nb_bytes == 0 )
691 msg_Err(p_dec, "Oops: No new buffer was returned!");
696 Read the Speex payload into the SpeexBits buffer.
698 speex_bits_read_from( &p_sys->bits,
699 (char*)p_speex_bit_block->p_buffer,
700 p_speex_bit_block->i_buffer );
703 Decode the input and ensure that no errors
706 i_decode_ret = speex_decode_int( p_sys->p_state, &p_sys->bits,
707 (int16_t*)p_aout_buffer->p_buffer );
708 if ( i_decode_ret < 0 )
710 msg_Err( p_dec, "Decoding failed. Perhaps we have a bad stream?" );
715 Handle date management on the audio output buffer.
717 p_aout_buffer->start_date = aout_DateGet( &p_sys->end_date );
718 p_aout_buffer->end_date = aout_DateIncrement( &p_sys->end_date,
719 p_sys->p_header->frame_size );
722 p_sys->i_frame_in_packet++;
723 block_Release( p_speex_bit_block );
725 return p_aout_buffer;
728 /*****************************************************************************
729 * DecodePacket: decodes a Speex packet.
730 *****************************************************************************/
731 static aout_buffer_t *DecodePacket( decoder_t *p_dec, ogg_packet *p_oggpacket )
733 decoder_sys_t *p_sys = p_dec->p_sys;
735 if( p_oggpacket->bytes )
737 /* Copy Ogg packet to Speex bitstream */
738 speex_bits_read_from( &p_sys->bits, (char *)p_oggpacket->packet,
739 p_oggpacket->bytes );
740 p_sys->i_frame_in_packet = 0;
743 /* Decode one frame at a time */
744 if( p_sys->i_frame_in_packet < p_sys->p_header->frames_per_packet )
746 aout_buffer_t *p_aout_buffer;
747 if( p_sys->p_header->frame_size == 0 )
751 p_dec->pf_aout_buffer_new( p_dec, p_sys->p_header->frame_size );
757 switch( speex_decode_int( p_sys->p_state, &p_sys->bits,
758 (int16_t *)p_aout_buffer->p_buffer ) )
761 msg_Err( p_dec, "decoding error: corrupted stream?" );
762 case -1: /* End of stream */
766 if( speex_bits_remaining( &p_sys->bits ) < 0 )
768 msg_Err( p_dec, "decoding overflow: corrupted stream?" );
771 if( p_sys->p_header->nb_channels == 2 )
772 speex_decode_stereo_int( (int16_t *)p_aout_buffer->p_buffer,
773 p_sys->p_header->frame_size,
776 /* Date management */
777 p_aout_buffer->start_date = aout_DateGet( &p_sys->end_date );
778 p_aout_buffer->end_date =
779 aout_DateIncrement( &p_sys->end_date, p_sys->p_header->frame_size );
781 p_sys->i_frame_in_packet++;
783 return p_aout_buffer;
791 /*****************************************************************************
792 * SendPacket: send an ogg packet to the stream output.
793 *****************************************************************************/
794 static block_t *SendPacket( decoder_t *p_dec, block_t *p_block )
796 decoder_sys_t *p_sys = p_dec->p_sys;
798 /* Date management */
799 p_block->i_dts = p_block->i_pts = aout_DateGet( &p_sys->end_date );
801 if( p_sys->i_headers >= p_sys->p_header->extra_headers + 2 )
804 aout_DateIncrement( &p_sys->end_date,
805 p_sys->p_header->frame_size ) -
809 p_block->i_length = 0;
814 /*****************************************************************************
815 * ParseSpeexComments: FIXME should be done in demuxer
816 *****************************************************************************/
817 #define readint(buf, base) (((buf[base+3]<<24)&0xff000000)| \
818 ((buf[base+2]<<16)&0xff0000)| \
819 ((buf[base+1]<<8)&0xff00)| \
822 static void ParseSpeexComments( decoder_t *p_dec, ogg_packet *p_oggpacket )
824 input_thread_t *p_input = (input_thread_t *)p_dec->p_parent;
825 decoder_sys_t *p_sys = p_dec->p_sys;
827 char *p_buf = (char *)p_oggpacket->packet;
828 const SpeexMode *p_mode;
831 if( p_input->i_object_type != VLC_OBJECT_INPUT ) return;
833 assert( p_sys->p_header->mode < SPEEX_NB_MODES );
835 p_mode = speex_mode_list[p_sys->p_header->mode];
836 assert( p_mode != NULL );
838 input_Control( p_input, INPUT_ADD_INFO, _("Speex comment"), _("Mode"),
839 "%s%s", p_mode->modeName,
840 p_sys->p_header->vbr ? " VBR" : "" );
842 if( p_oggpacket->bytes < 8 )
844 msg_Err( p_dec, "invalid/corrupted comments" );
848 i_len = readint( p_buf, 0 ); p_buf += 4;
849 if( i_len > p_oggpacket->bytes - 4 )
851 msg_Err( p_dec, "invalid/corrupted comments" );
855 input_Control( p_input, INPUT_ADD_INFO, _("Speex comment"), p_buf, "" );
857 /* TODO: finish comments parsing */
860 /*****************************************************************************
861 * CloseDecoder: speex decoder destruction
862 *****************************************************************************/
863 static void CloseDecoder( vlc_object_t *p_this )
865 decoder_t * p_dec = (decoder_t *)p_this;
866 decoder_sys_t *p_sys = p_dec->p_sys;
870 speex_decoder_destroy( p_sys->p_state );
871 speex_bits_destroy( &p_sys->bits );
874 free( p_sys->p_header );
878 /*****************************************************************************
879 * encoder_sys_t: encoder descriptor
880 *****************************************************************************/
881 #define MAX_FRAME_SIZE 2000
882 #define MAX_FRAME_BYTES 2000
890 char p_buffer_out[MAX_FRAME_BYTES];
897 SpeexStereoState stereo;
900 int i_frames_per_packet;
901 int i_frames_in_packet;
913 /*****************************************************************************
914 * OpenEncoder: probe the encoder and return score
915 *****************************************************************************/
916 static int OpenEncoder( vlc_object_t *p_this )
918 encoder_t *p_enc = (encoder_t *)p_this;
919 encoder_sys_t *p_sys;
920 const SpeexMode *p_speex_mode = &speex_nb_mode;
922 const char *pp_header[2];
926 if( p_enc->fmt_out.i_codec != VLC_FOURCC('s','p','x',' ') &&
932 config_ChainParse( p_enc, ENC_CFG_PREFIX, ppsz_enc_options, p_enc->p_cfg );
933 switch( var_GetInteger( p_enc, ENC_CFG_PREFIX "mode" ) )
936 msg_Dbg( p_enc, "Using wideband" );
937 p_speex_mode = &speex_wb_mode;
940 msg_Dbg( p_enc, "Using ultra-wideband" );
941 p_speex_mode = &speex_uwb_mode;
944 msg_Dbg( p_enc, "Using narrowband" );
945 p_speex_mode = &speex_nb_mode;
949 /* Allocate the memory needed to store the decoder's structure */
950 if( ( p_sys = (encoder_sys_t *)malloc(sizeof(encoder_sys_t)) ) == NULL )
952 p_enc->p_sys = p_sys;
953 p_enc->pf_encode_audio = Encode;
954 p_enc->fmt_in.i_codec = AOUT_FMT_S16_NE;
955 p_enc->fmt_out.i_codec = VLC_FOURCC('s','p','x',' ');
957 speex_init_header( &p_sys->header, p_enc->fmt_in.audio.i_rate,
960 p_sys->header.frames_per_packet = 1;
961 p_sys->header.vbr = var_GetBool( p_enc, ENC_CFG_PREFIX "cbr" ) ? 0 : 1;
962 p_sys->header.nb_channels = p_enc->fmt_in.audio.i_channels;
964 /* Create a new encoder state in narrowband mode */
965 p_sys->p_state = speex_encoder_init( p_speex_mode );
968 i_tmp = var_GetInteger( p_enc, ENC_CFG_PREFIX "complexity" );
969 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_COMPLEXITY, &i_tmp );
971 i_tmp = var_GetBool( p_enc, ENC_CFG_PREFIX "cbr" ) ? 0 : 1;
972 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_VBR, &i_tmp );
974 if( i_tmp == 0 ) /* CBR */
976 i_tmp = var_GetFloat( p_enc, ENC_CFG_PREFIX "quality" );
977 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_QUALITY, &i_tmp );
979 i_tmp = var_GetBool( p_enc, ENC_CFG_PREFIX "vad" ) ? 1 : 0;
980 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_VAD, &i_tmp );
986 f_tmp = var_GetFloat( p_enc, ENC_CFG_PREFIX "quality" );
987 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_VBR_QUALITY, &f_tmp );
989 i_tmp = var_GetInteger( p_enc, ENC_CFG_PREFIX "max-bitrate" );
991 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_VBR_MAX_BITRATE, &i_tmp );
994 i_tmp = var_GetBool( p_enc, ENC_CFG_PREFIX "dtx" ) ? 1 : 0;
995 speex_encoder_ctl( p_sys->p_state, SPEEX_SET_DTX, &i_tmp );
998 /*Initialization of the structure that holds the bits*/
999 speex_bits_init( &p_sys->bits );
1001 p_sys->i_frames_in_packet = 0;
1002 p_sys->i_samples_delay = 0;
1005 speex_encoder_ctl( p_sys->p_state, SPEEX_GET_FRAME_SIZE,
1006 &p_sys->i_frame_length );
1008 p_sys->i_frame_size = p_sys->i_frame_length *
1009 sizeof(int16_t) * p_enc->fmt_in.audio.i_channels;
1010 p_sys->p_buffer = malloc( p_sys->i_frame_size );
1012 /* Create and store headers */
1013 pp_header[0] = speex_header_to_packet( &p_sys->header, &pi_header[0] );
1014 pp_header[1] = "ENCODER=VLC media player";
1015 pi_header[1] = sizeof("ENCODER=VLC media player");
1017 p_enc->fmt_out.i_extra = 3 * 2 + pi_header[0] + pi_header[1];
1018 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->start_date -
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 );