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>
36 #include "../demux/xiph.h"
39 #include <speex/speex.h>
40 #include <speex/speex_header.h>
41 #include <speex/speex_stereo.h>
42 #include <speex/speex_callbacks.h>
46 /*****************************************************************************
48 *****************************************************************************/
49 static int OpenDecoder ( vlc_object_t
* );
50 static int OpenPacketizer( vlc_object_t
* );
51 static void CloseDecoder ( vlc_object_t
* );
52 static int OpenEncoder ( vlc_object_t
* );
53 static void CloseEncoder ( vlc_object_t
* );
55 #define ENC_CFG_PREFIX "sout-speex-"
57 #define ENC_MODE_TEXT N_("Mode" )
58 #define ENC_MODE_LONGTEXT N_( \
59 "Enforce the mode of the encoder." )
61 #define ENC_QUALITY_TEXT N_("Encoding quality")
62 #define ENC_QUALITY_LONGTEXT N_( \
63 "Enforce a quality between 0 (low) and 10 (high)." )
65 #define ENC_COMPLEXITY_TEXT N_("Encoding complexity" )
66 #define ENC_COMPLEXITY_LONGTEXT N_( \
67 "Enforce the complexity of the encoder." )
69 #define ENC_MAXBITRATE_TEXT N_( "Maximal bitrate" )
70 #define ENC_MAXBITRATE_LONGTEXT N_( \
71 "Enforce the maximal VBR bitrate" )
73 #define ENC_CBR_TEXT N_( "CBR encoding" )
74 #define ENC_CBR_LONGTEXT N_( \
75 "Enforce a constant bitrate encoding (CBR) instead of default " \
76 "variable bitrate encoding (VBR)." )
78 #define ENC_VAD_TEXT N_( "Voice activity detection" )
79 #define ENC_VAD_LONGTEXT N_( \
80 "Enable voice activity detection (VAD). It is automatically " \
81 "activated in VBR mode." )
83 #define ENC_DTX_TEXT N_( "Discontinuous Transmission" )
84 #define ENC_DTX_LONGTEXT N_( \
85 "Enable discontinuous transmission (DTX)." )
87 static const int pi_enc_mode_values
[] = { 0, 1, 2 };
88 static const char * const ppsz_enc_mode_descriptions
[] = {
89 N_("Narrow-band (8kHz)"), N_("Wide-band (16kHz)"), N_("Ultra-wideband (32kHz)"), NULL
93 set_category( CAT_INPUT
)
94 set_subcategory( SUBCAT_INPUT_ACODEC
)
96 set_description( N_("Speex audio decoder") )
97 set_capability( "decoder", 100 )
98 set_shortname( N_("Speex") )
99 set_callbacks( OpenDecoder
, CloseDecoder
)
102 set_description( N_("Speex audio packetizer") )
103 set_capability( "packetizer", 100 )
104 set_callbacks( OpenPacketizer
, CloseDecoder
)
107 set_description( N_("Speex audio encoder") )
108 set_capability( "encoder", 100 )
109 set_callbacks( OpenEncoder
, CloseEncoder
)
111 add_integer( ENC_CFG_PREFIX
"mode", 0, ENC_MODE_TEXT
,
112 ENC_MODE_LONGTEXT
, false )
113 change_integer_list( pi_enc_mode_values
, ppsz_enc_mode_descriptions
)
115 add_integer( ENC_CFG_PREFIX
"complexity", 3, ENC_COMPLEXITY_TEXT
,
116 ENC_COMPLEXITY_LONGTEXT
, false )
117 change_integer_range( 1, 10 )
119 add_bool( ENC_CFG_PREFIX
"cbr", false, ENC_CBR_TEXT
,
120 ENC_CBR_LONGTEXT
, false )
122 add_float( ENC_CFG_PREFIX
"quality", 8.0, ENC_QUALITY_TEXT
,
123 ENC_QUALITY_LONGTEXT
, false )
124 change_float_range( 0.0, 10.0 )
126 add_integer( ENC_CFG_PREFIX
"max-bitrate", 0, ENC_MAXBITRATE_TEXT
,
127 ENC_MAXBITRATE_LONGTEXT
, false )
129 add_bool( ENC_CFG_PREFIX
"vad", true, ENC_VAD_TEXT
,
130 ENC_VAD_LONGTEXT
, false )
132 add_bool( ENC_CFG_PREFIX
"dtx", false, ENC_DTX_TEXT
,
133 ENC_DTX_LONGTEXT
, false )
135 /* TODO agc, noise suppression, */
139 static const char *const ppsz_enc_options
[] = {
140 "mode", "complexity", "cbr", "quality", "max-bitrate", "vad", "dtx", NULL
143 /*****************************************************************************
144 * decoder_sys_t : speex decoder descriptor
145 *****************************************************************************/
155 int i_frame_in_packet
;
161 SpeexHeader
*p_header
;
162 SpeexStereoState stereo
;
164 unsigned int rtp_rate
;
173 static const int pi_channels_maps
[6] =
176 AOUT_CHAN_CENTER
, AOUT_CHAN_LEFT
| AOUT_CHAN_RIGHT
,
177 AOUT_CHAN_CENTER
| AOUT_CHAN_LEFT
| AOUT_CHAN_RIGHT
,
178 AOUT_CHAN_LEFT
| AOUT_CHAN_RIGHT
| AOUT_CHAN_REARLEFT
179 | AOUT_CHAN_REARRIGHT
,
180 AOUT_CHAN_LEFT
| AOUT_CHAN_RIGHT
| AOUT_CHAN_CENTER
181 | AOUT_CHAN_REARLEFT
| AOUT_CHAN_REARRIGHT
184 /****************************************************************************
186 ****************************************************************************/
188 static block_t
*DecodeBlock ( decoder_t
*, block_t
** );
189 static aout_buffer_t
*DecodeRtpSpeexPacket( decoder_t
*, block_t
**);
190 static int ProcessHeaders( decoder_t
* );
191 static int ProcessInitialHeader ( decoder_t
*, ogg_packet
* );
192 static void *ProcessPacket( decoder_t
*, ogg_packet
*, block_t
** );
194 static aout_buffer_t
*DecodePacket( decoder_t
*, ogg_packet
* );
195 static block_t
*SendPacket( decoder_t
*, block_t
* );
197 static void ParseSpeexComments( decoder_t
*, ogg_packet
* );
199 static block_t
*Encode ( encoder_t
*, aout_buffer_t
* );
201 /*****************************************************************************
202 * OpenDecoder: probe the decoder and return score
203 *****************************************************************************/
204 static int OpenDecoder( vlc_object_t
*p_this
)
206 decoder_t
*p_dec
= (decoder_t
*)p_this
;
207 decoder_sys_t
*p_sys
;
209 if( p_dec
->fmt_in
.i_codec
!= VLC_CODEC_SPEEX
)
212 /* Allocate the memory needed to store the decoder's structure */
213 if( ( p_dec
->p_sys
= p_sys
= malloc(sizeof(decoder_sys_t
)) ) == NULL
)
215 p_dec
->p_sys
->bits
.buf_size
= 0;
216 p_dec
->p_sys
->b_packetizer
= false;
217 p_dec
->p_sys
->rtp_rate
= p_dec
->fmt_in
.audio
.i_rate
;
218 p_dec
->p_sys
->b_has_headers
= false;
220 date_Set( &p_sys
->end_date
, 0 );
222 /* Set output properties */
223 p_dec
->fmt_out
.i_cat
= AUDIO_ES
;
224 p_dec
->fmt_out
.i_codec
= VLC_CODEC_S16N
;
228 If the codec is spxr then this decoder is
229 being invoked on a Speex stream arriving via RTP.
230 A special decoder callback is used.
232 if (p_dec
->fmt_in
.i_original_fourcc
== VLC_FOURCC('s', 'p', 'x', 'r'))
234 msg_Dbg( p_dec
, "Using RTP version of Speex decoder @ rate %d.",
235 p_dec
->fmt_in
.audio
.i_rate
);
236 p_dec
->pf_decode_audio
= DecodeRtpSpeexPacket
;
240 p_dec
->pf_decode_audio
= DecodeBlock
;
242 p_dec
->pf_packetize
= DecodeBlock
;
244 p_sys
->p_state
= NULL
;
245 p_sys
->p_header
= NULL
;
246 p_sys
->i_frame_in_packet
= 0;
251 static int OpenPacketizer( vlc_object_t
*p_this
)
253 decoder_t
*p_dec
= (decoder_t
*)p_this
;
255 int i_ret
= OpenDecoder( p_this
);
257 if( i_ret
== VLC_SUCCESS
)
259 p_dec
->p_sys
->b_packetizer
= true;
260 p_dec
->fmt_out
.i_codec
= VLC_CODEC_SPEEX
;
266 /****************************************************************************
267 * DecodeBlock: the whole thing
268 ****************************************************************************
269 * This function must be fed with ogg packets.
270 ****************************************************************************/
271 static block_t
*DecodeBlock( decoder_t
*p_dec
, block_t
**pp_block
)
273 decoder_sys_t
*p_sys
= p_dec
->p_sys
;
274 ogg_packet oggpacket
;
276 if( !pp_block
) return NULL
;
280 /* Block to Ogg packet */
281 oggpacket
.packet
= (*pp_block
)->p_buffer
;
282 oggpacket
.bytes
= (*pp_block
)->i_buffer
;
286 if( p_sys
->b_packetizer
) return NULL
;
288 /* Block to Ogg packet */
289 oggpacket
.packet
= NULL
;
293 oggpacket
.granulepos
= -1;
296 oggpacket
.packetno
= 0;
298 /* Check for headers */
299 if( !p_sys
->b_has_headers
)
301 if( ProcessHeaders( p_dec
) )
303 block_Release( *pp_block
);
306 p_sys
->b_has_headers
= true;
309 return ProcessPacket( p_dec
, &oggpacket
, pp_block
);
312 /*****************************************************************************
313 * ProcessHeaders: process Speex headers.
314 *****************************************************************************/
315 static int ProcessHeaders( decoder_t
*p_dec
)
317 decoder_sys_t
*p_sys
= p_dec
->p_sys
;
318 ogg_packet oggpacket
;
320 unsigned pi_size
[XIPH_MAX_HEADER_COUNT
];
321 void *pp_data
[XIPH_MAX_HEADER_COUNT
];
323 if( xiph_SplitHeaders( pi_size
, pp_data
, &i_count
,
324 p_dec
->fmt_in
.i_extra
, p_dec
->fmt_in
.p_extra
) )
329 oggpacket
.granulepos
= -1;
331 oggpacket
.packetno
= 0;
333 /* Take care of the initial Vorbis header */
334 oggpacket
.b_o_s
= 1; /* yes this actually is a b_o_s packet :) */
335 oggpacket
.bytes
= pi_size
[0];
336 oggpacket
.packet
= pp_data
[0];
337 if( ProcessInitialHeader( p_dec
, &oggpacket
) != VLC_SUCCESS
)
339 msg_Err( p_dec
, "initial Speex header is corrupted" );
343 /* The next packet in order is the comments header */
345 oggpacket
.bytes
= pi_size
[1];
346 oggpacket
.packet
= pp_data
[1];
347 ParseSpeexComments( p_dec
, &oggpacket
);
349 if( p_sys
->b_packetizer
)
351 p_dec
->fmt_out
.i_extra
= p_dec
->fmt_in
.i_extra
;
352 p_dec
->fmt_out
.p_extra
= xrealloc( p_dec
->fmt_out
.p_extra
,
353 p_dec
->fmt_out
.i_extra
);
354 memcpy( p_dec
->fmt_out
.p_extra
,
355 p_dec
->fmt_in
.p_extra
, p_dec
->fmt_out
.i_extra
);
358 for( unsigned i
= 0; i
< i_count
; i
++ )
363 for( unsigned i
= 0; i
< i_count
; i
++ )
368 /*****************************************************************************
369 * ProcessInitialHeader: processes the inital Speex header packet.
370 *****************************************************************************/
371 static int ProcessInitialHeader( decoder_t
*p_dec
, ogg_packet
*p_oggpacket
)
373 decoder_sys_t
*p_sys
= p_dec
->p_sys
;
376 SpeexHeader
*p_header
;
377 const SpeexMode
*p_mode
;
378 SpeexCallback callback
;
380 p_sys
->p_header
= p_header
=
381 speex_packet_to_header( (char *)p_oggpacket
->packet
,
382 p_oggpacket
->bytes
);
385 msg_Err( p_dec
, "cannot read Speex header" );
388 if( p_header
->mode
>= SPEEX_NB_MODES
|| p_header
->mode
< 0 )
390 msg_Err( p_dec
, "mode number %d does not (yet/any longer) exist in "
391 "this version of libspeex.", p_header
->mode
);
395 p_mode
= speex_mode_list
[p_header
->mode
];
399 if( p_header
->speex_version_id
> 1 )
401 msg_Err( p_dec
, "this file was encoded with Speex bit-stream "
402 "version %d which is not supported by this decoder.",
403 p_header
->speex_version_id
);
407 if( p_mode
->bitstream_version
< p_header
->mode_bitstream_version
)
409 msg_Err( p_dec
, "file encoded with a newer version of Speex." );
412 if( p_mode
->bitstream_version
> p_header
->mode_bitstream_version
)
414 msg_Err( p_dec
, "file encoded with an older version of Speex." );
418 msg_Dbg( p_dec
, "Speex %d Hz audio using %s mode %s%s",
419 p_header
->rate
, p_mode
->modeName
,
420 ( p_header
->nb_channels
== 1 ) ? " (mono" : " (stereo",
421 p_header
->vbr
? ", VBR)" : ")" );
423 /* Take care of speex decoder init */
424 speex_bits_init( &p_sys
->bits
);
425 p_sys
->p_state
= p_state
= speex_decoder_init( p_mode
);
428 msg_Err( p_dec
, "decoder initialization failed" );
432 if( p_header
->nb_channels
== 2 )
434 SpeexStereoState stereo
= SPEEX_STEREO_STATE_INIT
;
435 p_sys
->stereo
= stereo
;
436 callback
.callback_id
= SPEEX_INBAND_STEREO
;
437 callback
.func
= speex_std_stereo_request_handler
;
438 callback
.data
= &p_sys
->stereo
;
439 speex_decoder_ctl( p_state
, SPEEX_SET_HANDLER
, &callback
);
441 if( p_header
->nb_channels
<= 0 ||
442 p_header
->nb_channels
> 5 )
444 msg_Err( p_dec
, "invalid number of channels (not between 1 and 5): %i",
445 p_header
->nb_channels
);
449 /* Setup the format */
450 p_dec
->fmt_out
.audio
.i_physical_channels
=
451 p_dec
->fmt_out
.audio
.i_original_channels
=
452 pi_channels_maps
[p_header
->nb_channels
];
453 p_dec
->fmt_out
.audio
.i_channels
= p_header
->nb_channels
;
454 p_dec
->fmt_out
.audio
.i_rate
= p_header
->rate
;
456 date_Init( &p_sys
->end_date
, p_header
->rate
, 1 );
461 /*****************************************************************************
462 * ProcessPacket: processes a Speex packet.
463 *****************************************************************************/
464 static void *ProcessPacket( decoder_t
*p_dec
, ogg_packet
*p_oggpacket
,
467 decoder_sys_t
*p_sys
= p_dec
->p_sys
;
468 block_t
*p_block
= *pp_block
;
470 /* Date management */
471 if( p_block
&& p_block
->i_pts
> VLC_TS_INVALID
&&
472 p_block
->i_pts
!= date_Get( &p_sys
->end_date
) )
474 date_Set( &p_sys
->end_date
, p_block
->i_pts
);
477 if( !date_Get( &p_sys
->end_date
) )
479 /* We've just started the stream, wait for the first PTS. */
480 if( p_block
) block_Release( p_block
);
484 *pp_block
= NULL
; /* To avoid being fed the same packet again */
486 if( p_sys
->b_packetizer
)
488 if ( p_sys
->p_header
->frames_per_packet
> 1 )
490 short *p_frame_holder
= NULL
;
491 int i_bits_before
= 0, i_bits_after
= 0, i_bytes_in_speex_frame
= 0,
492 i_pcm_output_size
= 0, i_bits_in_speex_frame
= 0;
493 block_t
*p_new_block
= NULL
;
495 i_pcm_output_size
= p_sys
->p_header
->frame_size
;
496 p_frame_holder
= (short*)xmalloc( sizeof(short)*i_pcm_output_size
);
498 speex_bits_read_from( &p_sys
->bits
, (char*)p_oggpacket
->packet
,
500 i_bits_before
= speex_bits_remaining( &p_sys
->bits
);
501 speex_decode_int(p_sys
->p_state
, &p_sys
->bits
, p_frame_holder
);
502 i_bits_after
= speex_bits_remaining( &p_sys
->bits
);
504 i_bits_in_speex_frame
= i_bits_before
- i_bits_after
;
505 i_bytes_in_speex_frame
= ( i_bits_in_speex_frame
+
506 (8 - (i_bits_in_speex_frame
% 8)) )
509 p_new_block
= block_New( p_dec
, i_bytes_in_speex_frame
);
510 memset( p_new_block
->p_buffer
, 0xff, i_bytes_in_speex_frame
);
513 * Copy the first frame in this packet to a new packet.
515 speex_bits_rewind( &p_sys
->bits
);
516 speex_bits_write( &p_sys
->bits
,
517 (char*)p_new_block
->p_buffer
,
518 (int)i_bytes_in_speex_frame
);
521 * Move the remaining part of the original packet (subsequent
522 * frames, if there are any) into the beginning
523 * of the original packet so
524 * they are preserved following the realloc.
525 * Note: Any bits that
526 * remain in the initial packet
527 * are "filler" if they do not constitute
530 if ( i_bits_after
> 7 )
532 /* round-down since we rounded-up earlier (to include
533 * the speex terminator code.
535 i_bytes_in_speex_frame
--;
536 speex_bits_write( &p_sys
->bits
,
537 (char*)p_block
->p_buffer
,
538 p_block
->i_buffer
- i_bytes_in_speex_frame
);
539 p_block
= block_Realloc( p_block
,
541 p_block
->i_buffer
-i_bytes_in_speex_frame
);
546 speex_bits_reset( &p_sys
->bits
);
549 free( p_frame_holder
);
550 return SendPacket( p_dec
, p_new_block
);
554 return SendPacket( p_dec
, p_block
);
559 aout_buffer_t
*p_aout_buffer
= DecodePacket( p_dec
, p_oggpacket
);
562 block_Release( p_block
);
563 return p_aout_buffer
;
567 static aout_buffer_t
*DecodeRtpSpeexPacket( decoder_t
*p_dec
, block_t
**pp_block
)
569 block_t
*p_speex_bit_block
= *pp_block
;
570 decoder_sys_t
*p_sys
= p_dec
->p_sys
;
571 aout_buffer_t
*p_aout_buffer
;
573 unsigned int i_speex_frame_size
;
575 if ( !p_speex_bit_block
|| p_speex_bit_block
->i_pts
<= VLC_TS_INVALID
)
579 If the SpeexBits buffer size is 0 (a default value),
580 we know that a proper initialization has not yet been done.
582 if ( p_sys
->bits
.buf_size
==0 )
584 p_sys
->p_header
= (SpeexHeader
*)malloc(sizeof(SpeexHeader
));
585 if ( !p_sys
->p_header
)
587 msg_Err( p_dec
, "Could not allocate a Speex header.");
590 speex_init_header( p_sys
->p_header
,p_sys
->rtp_rate
,1,&speex_nb_mode
);
591 speex_bits_init( &p_sys
->bits
);
592 p_sys
->p_state
= speex_decoder_init( &speex_nb_mode
);
593 if ( !p_sys
->p_state
)
595 msg_Err( p_dec
, "Could not allocate a Speex decoder." );
596 free( p_sys
->p_header
);
601 Assume that variable bit rate is enabled. Also assume
602 that there is only one frame per packet.
604 p_sys
->p_header
->vbr
= 1;
605 p_sys
->p_header
->frames_per_packet
= 1;
607 p_dec
->fmt_out
.audio
.i_channels
= p_sys
->p_header
->nb_channels
;
608 p_dec
->fmt_out
.audio
.i_physical_channels
=
609 p_dec
->fmt_out
.audio
.i_original_channels
=
610 pi_channels_maps
[p_sys
->p_header
->nb_channels
];
611 p_dec
->fmt_out
.audio
.i_rate
= p_sys
->p_header
->rate
;
613 if ( speex_mode_query( &speex_nb_mode
,
614 SPEEX_MODE_FRAME_SIZE
,
615 &i_speex_frame_size
) )
617 msg_Err( p_dec
, "Could not determine the frame size." );
618 speex_decoder_destroy( p_sys
->p_state
);
619 free( p_sys
->p_header
);
622 p_dec
->fmt_out
.audio
.i_bytes_per_frame
= i_speex_frame_size
;
624 date_Init(&p_sys
->end_date
, p_sys
->p_header
->rate
, 1);
628 If the SpeexBits are initialized but there is
629 still no header, an error must be thrown.
631 if ( !p_sys
->p_header
)
633 msg_Err( p_dec
, "There is no valid Speex header found." );
638 if ( !date_Get( &p_sys
->end_date
) )
639 date_Set( &p_sys
->end_date
, p_speex_bit_block
->i_dts
);
642 Ask for a new audio output buffer and make sure
645 p_aout_buffer
= decoder_NewAudioBuffer( p_dec
,
646 p_sys
->p_header
->frame_size
);
647 if ( !p_aout_buffer
|| p_aout_buffer
->i_buffer
== 0 )
649 msg_Err(p_dec
, "Oops: No new buffer was returned!");
654 Read the Speex payload into the SpeexBits buffer.
656 speex_bits_read_from( &p_sys
->bits
,
657 (char*)p_speex_bit_block
->p_buffer
,
658 p_speex_bit_block
->i_buffer
);
661 Decode the input and ensure that no errors
664 i_decode_ret
= speex_decode_int( p_sys
->p_state
, &p_sys
->bits
,
665 (int16_t*)p_aout_buffer
->p_buffer
);
666 if ( i_decode_ret
< 0 )
668 msg_Err( p_dec
, "Decoding failed. Perhaps we have a bad stream?" );
673 Handle date management on the audio output buffer.
675 p_aout_buffer
->i_pts
= date_Get( &p_sys
->end_date
);
676 p_aout_buffer
->i_length
= date_Increment( &p_sys
->end_date
,
677 p_sys
->p_header
->frame_size
) - p_aout_buffer
->i_pts
;
680 p_sys
->i_frame_in_packet
++;
681 block_Release( p_speex_bit_block
);
683 return p_aout_buffer
;
686 /*****************************************************************************
687 * DecodePacket: decodes a Speex packet.
688 *****************************************************************************/
689 static aout_buffer_t
*DecodePacket( decoder_t
*p_dec
, ogg_packet
*p_oggpacket
)
691 decoder_sys_t
*p_sys
= p_dec
->p_sys
;
693 if( p_oggpacket
->bytes
)
695 /* Copy Ogg packet to Speex bitstream */
696 speex_bits_read_from( &p_sys
->bits
, (char *)p_oggpacket
->packet
,
697 p_oggpacket
->bytes
);
698 p_sys
->i_frame_in_packet
= 0;
701 /* Decode one frame at a time */
702 if( p_sys
->i_frame_in_packet
< p_sys
->p_header
->frames_per_packet
)
704 aout_buffer_t
*p_aout_buffer
;
705 if( p_sys
->p_header
->frame_size
== 0 )
709 decoder_NewAudioBuffer( p_dec
, p_sys
->p_header
->frame_size
);
715 switch( speex_decode_int( p_sys
->p_state
, &p_sys
->bits
,
716 (int16_t *)p_aout_buffer
->p_buffer
) )
719 msg_Err( p_dec
, "decoding error: corrupted stream?" );
720 case -1: /* End of stream */
724 if( speex_bits_remaining( &p_sys
->bits
) < 0 )
726 msg_Err( p_dec
, "decoding overflow: corrupted stream?" );
729 if( p_sys
->p_header
->nb_channels
== 2 )
730 speex_decode_stereo_int( (int16_t *)p_aout_buffer
->p_buffer
,
731 p_sys
->p_header
->frame_size
,
734 /* Date management */
735 p_aout_buffer
->i_pts
= date_Get( &p_sys
->end_date
);
736 p_aout_buffer
->i_length
=
737 date_Increment( &p_sys
->end_date
, p_sys
->p_header
->frame_size
)
738 - p_aout_buffer
->i_pts
;
740 p_sys
->i_frame_in_packet
++;
742 return p_aout_buffer
;
750 /*****************************************************************************
751 * SendPacket: send an ogg packet to the stream output.
752 *****************************************************************************/
753 static block_t
*SendPacket( decoder_t
*p_dec
, block_t
*p_block
)
755 decoder_sys_t
*p_sys
= p_dec
->p_sys
;
757 /* Date management */
758 p_block
->i_dts
= p_block
->i_pts
= date_Get( &p_sys
->end_date
);
761 date_Increment( &p_sys
->end_date
,
762 p_sys
->p_header
->frame_size
) -
768 /*****************************************************************************
769 * ParseSpeexComments:
770 *****************************************************************************/
771 #define readint(buf, base) (((buf[base+3]<<24)&0xff000000)| \
772 ((buf[base+2]<<16)&0xff0000)| \
773 ((buf[base+1]<<8)&0xff00)| \
776 static void ParseSpeexComments( decoder_t
*p_dec
, ogg_packet
*p_oggpacket
)
778 decoder_sys_t
*p_sys
= p_dec
->p_sys
;
779 const SpeexMode
*p_mode
;
781 assert( p_sys
->p_header
->mode
< SPEEX_NB_MODES
);
783 p_mode
= speex_mode_list
[p_sys
->p_header
->mode
];
784 assert( p_mode
!= NULL
);
786 if( !p_dec
->p_description
)
788 p_dec
->p_description
= vlc_meta_New();
789 if( !p_dec
->p_description
)
795 if( asprintf( &psz_mode
, "%s%s", p_mode
->modeName
, p_sys
->p_header
->vbr
? " VBR" : "" ) >= 0 )
797 vlc_meta_AddExtra( p_dec
->p_description
, _("Mode"), psz_mode
);
801 /* TODO: finish comments parsing */
802 VLC_UNUSED( p_oggpacket
);
805 /*****************************************************************************
806 * CloseDecoder: speex decoder destruction
807 *****************************************************************************/
808 static void CloseDecoder( vlc_object_t
*p_this
)
810 decoder_t
* p_dec
= (decoder_t
*)p_this
;
811 decoder_sys_t
*p_sys
= p_dec
->p_sys
;
815 speex_decoder_destroy( p_sys
->p_state
);
816 speex_bits_destroy( &p_sys
->bits
);
819 free( p_sys
->p_header
);
823 /*****************************************************************************
824 * encoder_sys_t: encoder descriptor
825 *****************************************************************************/
826 #define MAX_FRAME_SIZE 2000
827 #define MAX_FRAME_BYTES 2000
835 char p_buffer_out
[MAX_FRAME_BYTES
];
842 SpeexStereoState stereo
;
845 int i_frames_per_packet
;
846 int i_frames_in_packet
;
853 /*****************************************************************************
854 * OpenEncoder: probe the encoder and return score
855 *****************************************************************************/
856 static int OpenEncoder( vlc_object_t
*p_this
)
858 encoder_t
*p_enc
= (encoder_t
*)p_this
;
859 encoder_sys_t
*p_sys
;
860 const SpeexMode
*p_speex_mode
= &speex_nb_mode
;
862 const char *pp_header
[2];
866 if( p_enc
->fmt_out
.i_codec
!= VLC_CODEC_SPEEX
&&
872 config_ChainParse( p_enc
, ENC_CFG_PREFIX
, ppsz_enc_options
, p_enc
->p_cfg
);
873 switch( var_GetInteger( p_enc
, ENC_CFG_PREFIX
"mode" ) )
876 msg_Dbg( p_enc
, "Using wideband" );
877 p_speex_mode
= &speex_wb_mode
;
880 msg_Dbg( p_enc
, "Using ultra-wideband" );
881 p_speex_mode
= &speex_uwb_mode
;
884 msg_Dbg( p_enc
, "Using narrowband" );
885 p_speex_mode
= &speex_nb_mode
;
889 /* Allocate the memory needed to store the decoder's structure */
890 if( ( p_sys
= (encoder_sys_t
*)malloc(sizeof(encoder_sys_t
)) ) == NULL
)
892 p_enc
->p_sys
= p_sys
;
893 p_enc
->pf_encode_audio
= Encode
;
894 p_enc
->fmt_in
.i_codec
= VLC_CODEC_S16N
;
895 p_enc
->fmt_out
.i_codec
= VLC_CODEC_SPEEX
;
897 speex_init_header( &p_sys
->header
, p_enc
->fmt_in
.audio
.i_rate
,
900 p_sys
->header
.frames_per_packet
= 1;
901 p_sys
->header
.vbr
= var_GetBool( p_enc
, ENC_CFG_PREFIX
"cbr" ) ? 0 : 1;
902 p_sys
->header
.nb_channels
= p_enc
->fmt_in
.audio
.i_channels
;
904 /* Create a new encoder state in narrowband mode */
905 p_sys
->p_state
= speex_encoder_init( p_speex_mode
);
908 i_tmp
= var_GetInteger( p_enc
, ENC_CFG_PREFIX
"complexity" );
909 speex_encoder_ctl( p_sys
->p_state
, SPEEX_SET_COMPLEXITY
, &i_tmp
);
911 i_tmp
= var_GetBool( p_enc
, ENC_CFG_PREFIX
"cbr" ) ? 0 : 1;
912 speex_encoder_ctl( p_sys
->p_state
, SPEEX_SET_VBR
, &i_tmp
);
914 if( i_tmp
== 0 ) /* CBR */
916 i_tmp
= var_GetFloat( p_enc
, ENC_CFG_PREFIX
"quality" );
917 speex_encoder_ctl( p_sys
->p_state
, SPEEX_SET_QUALITY
, &i_tmp
);
919 i_tmp
= var_GetBool( p_enc
, ENC_CFG_PREFIX
"vad" ) ? 1 : 0;
920 speex_encoder_ctl( p_sys
->p_state
, SPEEX_SET_VAD
, &i_tmp
);
926 f_tmp
= var_GetFloat( p_enc
, ENC_CFG_PREFIX
"quality" );
927 speex_encoder_ctl( p_sys
->p_state
, SPEEX_SET_VBR_QUALITY
, &f_tmp
);
929 i_tmp
= var_GetInteger( p_enc
, ENC_CFG_PREFIX
"max-bitrate" );
931 #ifdef SPEEX_SET_VBR_MAX_BITRATE
932 speex_encoder_ctl( p_sys
->p_state
, SPEEX_SET_VBR_MAX_BITRATE
, &i_tmp
);
934 msg_Dbg( p_enc
, "max-bitrate cannot be set in this version of libspeex");
938 i_tmp
= var_GetBool( p_enc
, ENC_CFG_PREFIX
"dtx" ) ? 1 : 0;
939 speex_encoder_ctl( p_sys
->p_state
, SPEEX_SET_DTX
, &i_tmp
);
942 /*Initialization of the structure that holds the bits*/
943 speex_bits_init( &p_sys
->bits
);
945 p_sys
->i_frames_in_packet
= 0;
946 p_sys
->i_samples_delay
= 0;
948 speex_encoder_ctl( p_sys
->p_state
, SPEEX_GET_FRAME_SIZE
,
949 &p_sys
->i_frame_length
);
951 p_sys
->i_frame_size
= p_sys
->i_frame_length
*
952 sizeof(int16_t) * p_enc
->fmt_in
.audio
.i_channels
;
953 p_sys
->p_buffer
= xmalloc( p_sys
->i_frame_size
);
955 /* Create and store headers */
956 pp_header
[0] = speex_header_to_packet( &p_sys
->header
, &pi_header
[0] );
957 pp_header
[1] = "ENCODER=VLC media player";
958 pi_header
[1] = sizeof("ENCODER=VLC media player");
960 p_enc
->fmt_out
.i_extra
= 3 * 2 + pi_header
[0] + pi_header
[1];
961 p_extra
= p_enc
->fmt_out
.p_extra
= xmalloc( p_enc
->fmt_out
.i_extra
);
962 for( i
= 0; i
< 2; i
++ )
964 *(p_extra
++) = pi_header
[i
] >> 8;
965 *(p_extra
++) = pi_header
[i
] & 0xFF;
966 memcpy( p_extra
, pp_header
[i
], pi_header
[i
] );
967 p_extra
+= pi_header
[i
];
970 msg_Dbg( p_enc
, "encoding: frame size:%d, channels:%d, samplerate:%d",
971 p_sys
->i_frame_size
, p_enc
->fmt_in
.audio
.i_channels
,
972 p_enc
->fmt_in
.audio
.i_rate
);
977 /****************************************************************************
978 * Encode: the whole thing
979 ****************************************************************************
980 * This function spits out ogg packets.
981 ****************************************************************************/
982 static block_t
*Encode( encoder_t
*p_enc
, aout_buffer_t
*p_aout_buf
)
984 encoder_sys_t
*p_sys
= p_enc
->p_sys
;
985 block_t
*p_block
, *p_chain
= NULL
;
987 unsigned char *p_buffer
= p_aout_buf
->p_buffer
;
988 int i_samples
= p_aout_buf
->i_nb_samples
;
989 int i_samples_delay
= p_sys
->i_samples_delay
;
991 mtime_t i_pts
= p_aout_buf
->i_pts
-
992 (mtime_t
)1000000 * (mtime_t
)p_sys
->i_samples_delay
/
993 (mtime_t
)p_enc
->fmt_in
.audio
.i_rate
;
995 p_sys
->i_samples_delay
+= i_samples
;
997 while( p_sys
->i_samples_delay
>= p_sys
->i_frame_length
)
1002 if( i_samples_delay
)
1004 /* Take care of the left-over from last time */
1005 int i_delay_size
= i_samples_delay
* 2 *
1006 p_enc
->fmt_in
.audio
.i_channels
;
1007 int i_size
= p_sys
->i_frame_size
- i_delay_size
;
1009 p_samples
= (int16_t *)p_sys
->p_buffer
;
1010 memcpy( p_sys
->p_buffer
+ i_delay_size
, p_buffer
, i_size
);
1011 p_buffer
-= i_delay_size
;
1012 i_samples
+= i_samples_delay
;
1013 i_samples_delay
= 0;
1017 p_samples
= (int16_t *)p_buffer
;
1020 /* Encode current frame */
1021 if( p_enc
->fmt_in
.audio
.i_channels
== 2 )
1022 speex_encode_stereo_int( p_samples
, p_sys
->i_frame_length
,
1026 if( p_sys
->preprocess
)
1027 speex_preprocess( p_sys
->preprocess
, p_samples
, NULL
);
1030 speex_encode_int( p_sys
->p_state
, p_samples
, &p_sys
->bits
);
1032 p_buffer
+= p_sys
->i_frame_size
;
1033 p_sys
->i_samples_delay
-= p_sys
->i_frame_length
;
1034 i_samples
-= p_sys
->i_frame_length
;
1036 p_sys
->i_frames_in_packet
++;
1038 if( p_sys
->i_frames_in_packet
< p_sys
->header
.frames_per_packet
)
1041 p_sys
->i_frames_in_packet
= 0;
1043 speex_bits_insert_terminator( &p_sys
->bits
);
1044 i_out
= speex_bits_write( &p_sys
->bits
, p_sys
->p_buffer_out
,
1046 speex_bits_reset( &p_sys
->bits
);
1048 p_block
= block_New( p_enc
, i_out
);
1049 memcpy( p_block
->p_buffer
, p_sys
->p_buffer_out
, i_out
);
1051 p_block
->i_length
= (mtime_t
)1000000 *
1052 (mtime_t
)p_sys
->i_frame_length
* p_sys
->header
.frames_per_packet
/
1053 (mtime_t
)p_enc
->fmt_in
.audio
.i_rate
;
1055 p_block
->i_dts
= p_block
->i_pts
= i_pts
;
1058 i_pts
+= p_block
->i_length
;
1059 block_ChainAppend( &p_chain
, p_block
);
1063 /* Backup the remaining raw samples */
1066 memcpy( p_sys
->p_buffer
+ i_samples_delay
* 2 *
1067 p_enc
->fmt_in
.audio
.i_channels
, p_buffer
,
1068 i_samples
* 2 * p_enc
->fmt_in
.audio
.i_channels
);
1074 /*****************************************************************************
1075 * CloseEncoder: encoder destruction
1076 *****************************************************************************/
1077 static void CloseEncoder( vlc_object_t
*p_this
)
1079 encoder_t
*p_enc
= (encoder_t
*)p_this
;
1080 encoder_sys_t
*p_sys
= p_enc
->p_sys
;
1082 speex_encoder_destroy( p_sys
->p_state
);
1083 speex_bits_destroy( &p_sys
->bits
);
1085 free( p_sys
->p_buffer
);