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 void *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
= (aout_buffer_t
*(*)(decoder_t
*, block_t
**))
237 DecodeRtpSpeexPacket
;
241 p_dec
->pf_decode_audio
= (aout_buffer_t
*(*)(decoder_t
*, block_t
**))
244 p_dec
->pf_packetize
= (block_t
*(*)(decoder_t
*, block_t
**))
247 p_sys
->p_state
= NULL
;
248 p_sys
->p_header
= NULL
;
249 p_sys
->i_frame_in_packet
= 0;
254 static int OpenPacketizer( vlc_object_t
*p_this
)
256 decoder_t
*p_dec
= (decoder_t
*)p_this
;
258 int i_ret
= OpenDecoder( p_this
);
260 if( i_ret
== VLC_SUCCESS
)
262 p_dec
->p_sys
->b_packetizer
= true;
263 p_dec
->fmt_out
.i_codec
= VLC_CODEC_SPEEX
;
269 /****************************************************************************
270 * DecodeBlock: the whole thing
271 ****************************************************************************
272 * This function must be fed with ogg packets.
273 ****************************************************************************/
274 static void *DecodeBlock( decoder_t
*p_dec
, block_t
**pp_block
)
276 decoder_sys_t
*p_sys
= p_dec
->p_sys
;
277 ogg_packet oggpacket
;
279 if( !pp_block
) return NULL
;
283 /* Block to Ogg packet */
284 oggpacket
.packet
= (*pp_block
)->p_buffer
;
285 oggpacket
.bytes
= (*pp_block
)->i_buffer
;
289 if( p_sys
->b_packetizer
) return NULL
;
291 /* Block to Ogg packet */
292 oggpacket
.packet
= NULL
;
296 oggpacket
.granulepos
= -1;
299 oggpacket
.packetno
= 0;
301 /* Check for headers */
302 if( !p_sys
->b_has_headers
)
304 if( ProcessHeaders( p_dec
) )
306 block_Release( *pp_block
);
309 p_sys
->b_has_headers
= true;
312 return ProcessPacket( p_dec
, &oggpacket
, pp_block
);
315 /*****************************************************************************
316 * ProcessHeaders: process Speex headers.
317 *****************************************************************************/
318 static int ProcessHeaders( decoder_t
*p_dec
)
320 decoder_sys_t
*p_sys
= p_dec
->p_sys
;
321 ogg_packet oggpacket
;
323 unsigned pi_size
[XIPH_MAX_HEADER_COUNT
];
324 void *pp_data
[XIPH_MAX_HEADER_COUNT
];
326 if( xiph_SplitHeaders( pi_size
, pp_data
, &i_count
,
327 p_dec
->fmt_in
.i_extra
, p_dec
->fmt_in
.p_extra
) )
332 oggpacket
.granulepos
= -1;
334 oggpacket
.packetno
= 0;
336 /* Take care of the initial Vorbis header */
337 oggpacket
.b_o_s
= 1; /* yes this actually is a b_o_s packet :) */
338 oggpacket
.bytes
= pi_size
[0];
339 oggpacket
.packet
= pp_data
[0];
340 if( ProcessInitialHeader( p_dec
, &oggpacket
) != VLC_SUCCESS
)
342 msg_Err( p_dec
, "initial Speex header is corrupted" );
346 /* The next packet in order is the comments header */
348 oggpacket
.bytes
= pi_size
[1];
349 oggpacket
.packet
= pp_data
[1];
350 ParseSpeexComments( p_dec
, &oggpacket
);
352 if( p_sys
->b_packetizer
)
354 p_dec
->fmt_out
.i_extra
= p_dec
->fmt_in
.i_extra
;
355 p_dec
->fmt_out
.p_extra
= xrealloc( p_dec
->fmt_out
.p_extra
,
356 p_dec
->fmt_out
.i_extra
);
357 memcpy( p_dec
->fmt_out
.p_extra
,
358 p_dec
->fmt_in
.p_extra
, p_dec
->fmt_out
.i_extra
);
361 for( unsigned i
= 0; i
< i_count
; i
++ )
366 for( unsigned i
= 0; i
< i_count
; i
++ )
371 /*****************************************************************************
372 * ProcessInitialHeader: processes the inital Speex header packet.
373 *****************************************************************************/
374 static int ProcessInitialHeader( decoder_t
*p_dec
, ogg_packet
*p_oggpacket
)
376 decoder_sys_t
*p_sys
= p_dec
->p_sys
;
379 SpeexHeader
*p_header
;
380 const SpeexMode
*p_mode
;
381 SpeexCallback callback
;
383 p_sys
->p_header
= p_header
=
384 speex_packet_to_header( (char *)p_oggpacket
->packet
,
385 p_oggpacket
->bytes
);
388 msg_Err( p_dec
, "cannot read Speex header" );
391 if( p_header
->mode
>= SPEEX_NB_MODES
|| p_header
->mode
< 0 )
393 msg_Err( p_dec
, "mode number %d does not (yet/any longer) exist in "
394 "this version of libspeex.", p_header
->mode
);
398 p_mode
= speex_mode_list
[p_header
->mode
];
402 if( p_header
->speex_version_id
> 1 )
404 msg_Err( p_dec
, "this file was encoded with Speex bit-stream "
405 "version %d which is not supported by this decoder.",
406 p_header
->speex_version_id
);
410 if( p_mode
->bitstream_version
< p_header
->mode_bitstream_version
)
412 msg_Err( p_dec
, "file encoded with a newer version of Speex." );
415 if( p_mode
->bitstream_version
> p_header
->mode_bitstream_version
)
417 msg_Err( p_dec
, "file encoded with an older version of Speex." );
421 msg_Dbg( p_dec
, "Speex %d Hz audio using %s mode %s%s",
422 p_header
->rate
, p_mode
->modeName
,
423 ( p_header
->nb_channels
== 1 ) ? " (mono" : " (stereo",
424 p_header
->vbr
? ", VBR)" : ")" );
426 /* Take care of speex decoder init */
427 speex_bits_init( &p_sys
->bits
);
428 p_sys
->p_state
= p_state
= speex_decoder_init( p_mode
);
431 msg_Err( p_dec
, "decoder initialization failed" );
435 if( p_header
->nb_channels
== 2 )
437 SpeexStereoState stereo
= SPEEX_STEREO_STATE_INIT
;
438 p_sys
->stereo
= stereo
;
439 callback
.callback_id
= SPEEX_INBAND_STEREO
;
440 callback
.func
= speex_std_stereo_request_handler
;
441 callback
.data
= &p_sys
->stereo
;
442 speex_decoder_ctl( p_state
, SPEEX_SET_HANDLER
, &callback
);
444 if( p_header
->nb_channels
<= 0 ||
445 p_header
->nb_channels
> 5 )
447 msg_Err( p_dec
, "invalid number of channels (not between 1 and 5): %i",
448 p_header
->nb_channels
);
452 /* Setup the format */
453 p_dec
->fmt_out
.audio
.i_physical_channels
=
454 p_dec
->fmt_out
.audio
.i_original_channels
=
455 pi_channels_maps
[p_header
->nb_channels
];
456 p_dec
->fmt_out
.audio
.i_channels
= p_header
->nb_channels
;
457 p_dec
->fmt_out
.audio
.i_rate
= p_header
->rate
;
459 date_Init( &p_sys
->end_date
, p_header
->rate
, 1 );
464 /*****************************************************************************
465 * ProcessPacket: processes a Speex packet.
466 *****************************************************************************/
467 static void *ProcessPacket( decoder_t
*p_dec
, ogg_packet
*p_oggpacket
,
470 decoder_sys_t
*p_sys
= p_dec
->p_sys
;
471 block_t
*p_block
= *pp_block
;
473 /* Date management */
474 if( p_block
&& p_block
->i_pts
> VLC_TS_INVALID
&&
475 p_block
->i_pts
!= date_Get( &p_sys
->end_date
) )
477 date_Set( &p_sys
->end_date
, p_block
->i_pts
);
480 if( !date_Get( &p_sys
->end_date
) )
482 /* We've just started the stream, wait for the first PTS. */
483 if( p_block
) block_Release( p_block
);
487 *pp_block
= NULL
; /* To avoid being fed the same packet again */
489 if( p_sys
->b_packetizer
)
491 if ( p_sys
->p_header
->frames_per_packet
> 1 )
493 short *p_frame_holder
= NULL
;
494 int i_bits_before
= 0, i_bits_after
= 0, i_bytes_in_speex_frame
= 0,
495 i_pcm_output_size
= 0, i_bits_in_speex_frame
= 0;
496 block_t
*p_new_block
= NULL
;
498 i_pcm_output_size
= p_sys
->p_header
->frame_size
;
499 p_frame_holder
= (short*)xmalloc( sizeof(short)*i_pcm_output_size
);
501 speex_bits_read_from( &p_sys
->bits
, (char*)p_oggpacket
->packet
,
503 i_bits_before
= speex_bits_remaining( &p_sys
->bits
);
504 speex_decode_int(p_sys
->p_state
, &p_sys
->bits
, p_frame_holder
);
505 i_bits_after
= speex_bits_remaining( &p_sys
->bits
);
507 i_bits_in_speex_frame
= i_bits_before
- i_bits_after
;
508 i_bytes_in_speex_frame
= ( i_bits_in_speex_frame
+
509 (8 - (i_bits_in_speex_frame
% 8)) )
512 p_new_block
= block_New( p_dec
, i_bytes_in_speex_frame
);
513 memset( p_new_block
->p_buffer
, 0xff, i_bytes_in_speex_frame
);
516 * Copy the first frame in this packet to a new packet.
518 speex_bits_rewind( &p_sys
->bits
);
519 speex_bits_write( &p_sys
->bits
,
520 (char*)p_new_block
->p_buffer
,
521 (int)i_bytes_in_speex_frame
);
524 * Move the remaining part of the original packet (subsequent
525 * frames, if there are any) into the beginning
526 * of the original packet so
527 * they are preserved following the realloc.
528 * Note: Any bits that
529 * remain in the initial packet
530 * are "filler" if they do not constitute
533 if ( i_bits_after
> 7 )
535 /* round-down since we rounded-up earlier (to include
536 * the speex terminator code.
538 i_bytes_in_speex_frame
--;
539 speex_bits_write( &p_sys
->bits
,
540 (char*)p_block
->p_buffer
,
541 p_block
->i_buffer
- i_bytes_in_speex_frame
);
542 p_block
= block_Realloc( p_block
,
544 p_block
->i_buffer
-i_bytes_in_speex_frame
);
549 speex_bits_reset( &p_sys
->bits
);
552 free( p_frame_holder
);
553 return SendPacket( p_dec
, p_new_block
);
557 return SendPacket( p_dec
, p_block
);
562 aout_buffer_t
*p_aout_buffer
= DecodePacket( p_dec
, p_oggpacket
);
565 block_Release( p_block
);
566 return p_aout_buffer
;
570 static aout_buffer_t
*DecodeRtpSpeexPacket( decoder_t
*p_dec
, block_t
**pp_block
)
572 block_t
*p_speex_bit_block
= *pp_block
;
573 decoder_sys_t
*p_sys
= p_dec
->p_sys
;
574 aout_buffer_t
*p_aout_buffer
;
576 unsigned int i_speex_frame_size
;
578 if ( !p_speex_bit_block
|| p_speex_bit_block
->i_pts
<= VLC_TS_INVALID
)
582 If the SpeexBits buffer size is 0 (a default value),
583 we know that a proper initialization has not yet been done.
585 if ( p_sys
->bits
.buf_size
==0 )
587 p_sys
->p_header
= (SpeexHeader
*)malloc(sizeof(SpeexHeader
));
588 if ( !p_sys
->p_header
)
590 msg_Err( p_dec
, "Could not allocate a Speex header.");
593 speex_init_header( p_sys
->p_header
,p_sys
->rtp_rate
,1,&speex_nb_mode
);
594 speex_bits_init( &p_sys
->bits
);
595 p_sys
->p_state
= speex_decoder_init( &speex_nb_mode
);
596 if ( !p_sys
->p_state
)
598 msg_Err( p_dec
, "Could not allocate a Speex decoder." );
599 free( p_sys
->p_header
);
604 Assume that variable bit rate is enabled. Also assume
605 that there is only one frame per packet.
607 p_sys
->p_header
->vbr
= 1;
608 p_sys
->p_header
->frames_per_packet
= 1;
610 p_dec
->fmt_out
.audio
.i_channels
= p_sys
->p_header
->nb_channels
;
611 p_dec
->fmt_out
.audio
.i_physical_channels
=
612 p_dec
->fmt_out
.audio
.i_original_channels
=
613 pi_channels_maps
[p_sys
->p_header
->nb_channels
];
614 p_dec
->fmt_out
.audio
.i_rate
= p_sys
->p_header
->rate
;
616 if ( speex_mode_query( &speex_nb_mode
,
617 SPEEX_MODE_FRAME_SIZE
,
618 &i_speex_frame_size
) )
620 msg_Err( p_dec
, "Could not determine the frame size." );
621 speex_decoder_destroy( p_sys
->p_state
);
622 free( p_sys
->p_header
);
625 p_dec
->fmt_out
.audio
.i_bytes_per_frame
= i_speex_frame_size
;
627 date_Init(&p_sys
->end_date
, p_sys
->p_header
->rate
, 1);
631 If the SpeexBits are initialized but there is
632 still no header, an error must be thrown.
634 if ( !p_sys
->p_header
)
636 msg_Err( p_dec
, "There is no valid Speex header found." );
641 if ( !date_Get( &p_sys
->end_date
) )
642 date_Set( &p_sys
->end_date
, p_speex_bit_block
->i_dts
);
645 Ask for a new audio output buffer and make sure
648 p_aout_buffer
= decoder_NewAudioBuffer( p_dec
,
649 p_sys
->p_header
->frame_size
);
650 if ( !p_aout_buffer
|| p_aout_buffer
->i_buffer
== 0 )
652 msg_Err(p_dec
, "Oops: No new buffer was returned!");
657 Read the Speex payload into the SpeexBits buffer.
659 speex_bits_read_from( &p_sys
->bits
,
660 (char*)p_speex_bit_block
->p_buffer
,
661 p_speex_bit_block
->i_buffer
);
664 Decode the input and ensure that no errors
667 i_decode_ret
= speex_decode_int( p_sys
->p_state
, &p_sys
->bits
,
668 (int16_t*)p_aout_buffer
->p_buffer
);
669 if ( i_decode_ret
< 0 )
671 msg_Err( p_dec
, "Decoding failed. Perhaps we have a bad stream?" );
676 Handle date management on the audio output buffer.
678 p_aout_buffer
->i_pts
= date_Get( &p_sys
->end_date
);
679 p_aout_buffer
->i_length
= date_Increment( &p_sys
->end_date
,
680 p_sys
->p_header
->frame_size
) - p_aout_buffer
->i_pts
;
683 p_sys
->i_frame_in_packet
++;
684 block_Release( p_speex_bit_block
);
686 return p_aout_buffer
;
689 /*****************************************************************************
690 * DecodePacket: decodes a Speex packet.
691 *****************************************************************************/
692 static aout_buffer_t
*DecodePacket( decoder_t
*p_dec
, ogg_packet
*p_oggpacket
)
694 decoder_sys_t
*p_sys
= p_dec
->p_sys
;
696 if( p_oggpacket
->bytes
)
698 /* Copy Ogg packet to Speex bitstream */
699 speex_bits_read_from( &p_sys
->bits
, (char *)p_oggpacket
->packet
,
700 p_oggpacket
->bytes
);
701 p_sys
->i_frame_in_packet
= 0;
704 /* Decode one frame at a time */
705 if( p_sys
->i_frame_in_packet
< p_sys
->p_header
->frames_per_packet
)
707 aout_buffer_t
*p_aout_buffer
;
708 if( p_sys
->p_header
->frame_size
== 0 )
712 decoder_NewAudioBuffer( p_dec
, p_sys
->p_header
->frame_size
);
718 switch( speex_decode_int( p_sys
->p_state
, &p_sys
->bits
,
719 (int16_t *)p_aout_buffer
->p_buffer
) )
722 msg_Err( p_dec
, "decoding error: corrupted stream?" );
723 case -1: /* End of stream */
727 if( speex_bits_remaining( &p_sys
->bits
) < 0 )
729 msg_Err( p_dec
, "decoding overflow: corrupted stream?" );
732 if( p_sys
->p_header
->nb_channels
== 2 )
733 speex_decode_stereo_int( (int16_t *)p_aout_buffer
->p_buffer
,
734 p_sys
->p_header
->frame_size
,
737 /* Date management */
738 p_aout_buffer
->i_pts
= date_Get( &p_sys
->end_date
);
739 p_aout_buffer
->i_length
=
740 date_Increment( &p_sys
->end_date
, p_sys
->p_header
->frame_size
)
741 - p_aout_buffer
->i_pts
;
743 p_sys
->i_frame_in_packet
++;
745 return p_aout_buffer
;
753 /*****************************************************************************
754 * SendPacket: send an ogg packet to the stream output.
755 *****************************************************************************/
756 static block_t
*SendPacket( decoder_t
*p_dec
, block_t
*p_block
)
758 decoder_sys_t
*p_sys
= p_dec
->p_sys
;
760 /* Date management */
761 p_block
->i_dts
= p_block
->i_pts
= date_Get( &p_sys
->end_date
);
764 date_Increment( &p_sys
->end_date
,
765 p_sys
->p_header
->frame_size
) -
771 /*****************************************************************************
772 * ParseSpeexComments:
773 *****************************************************************************/
774 #define readint(buf, base) (((buf[base+3]<<24)&0xff000000)| \
775 ((buf[base+2]<<16)&0xff0000)| \
776 ((buf[base+1]<<8)&0xff00)| \
779 static void ParseSpeexComments( decoder_t
*p_dec
, ogg_packet
*p_oggpacket
)
781 decoder_sys_t
*p_sys
= p_dec
->p_sys
;
782 const SpeexMode
*p_mode
;
784 assert( p_sys
->p_header
->mode
< SPEEX_NB_MODES
);
786 p_mode
= speex_mode_list
[p_sys
->p_header
->mode
];
787 assert( p_mode
!= NULL
);
789 if( !p_dec
->p_description
)
791 p_dec
->p_description
= vlc_meta_New();
792 if( !p_dec
->p_description
)
798 if( asprintf( &psz_mode
, "%s%s", p_mode
->modeName
, p_sys
->p_header
->vbr
? " VBR" : "" ) >= 0 )
800 vlc_meta_AddExtra( p_dec
->p_description
, _("Mode"), psz_mode
);
804 /* TODO: finish comments parsing */
805 VLC_UNUSED( p_oggpacket
);
808 /*****************************************************************************
809 * CloseDecoder: speex decoder destruction
810 *****************************************************************************/
811 static void CloseDecoder( vlc_object_t
*p_this
)
813 decoder_t
* p_dec
= (decoder_t
*)p_this
;
814 decoder_sys_t
*p_sys
= p_dec
->p_sys
;
818 speex_decoder_destroy( p_sys
->p_state
);
819 speex_bits_destroy( &p_sys
->bits
);
822 free( p_sys
->p_header
);
826 /*****************************************************************************
827 * encoder_sys_t: encoder descriptor
828 *****************************************************************************/
829 #define MAX_FRAME_SIZE 2000
830 #define MAX_FRAME_BYTES 2000
838 char p_buffer_out
[MAX_FRAME_BYTES
];
845 SpeexStereoState stereo
;
848 int i_frames_per_packet
;
849 int i_frames_in_packet
;
856 /*****************************************************************************
857 * OpenEncoder: probe the encoder and return score
858 *****************************************************************************/
859 static int OpenEncoder( vlc_object_t
*p_this
)
861 encoder_t
*p_enc
= (encoder_t
*)p_this
;
862 encoder_sys_t
*p_sys
;
863 const SpeexMode
*p_speex_mode
= &speex_nb_mode
;
865 const char *pp_header
[2];
869 if( p_enc
->fmt_out
.i_codec
!= VLC_CODEC_SPEEX
&&
875 config_ChainParse( p_enc
, ENC_CFG_PREFIX
, ppsz_enc_options
, p_enc
->p_cfg
);
876 switch( var_GetInteger( p_enc
, ENC_CFG_PREFIX
"mode" ) )
879 msg_Dbg( p_enc
, "Using wideband" );
880 p_speex_mode
= &speex_wb_mode
;
883 msg_Dbg( p_enc
, "Using ultra-wideband" );
884 p_speex_mode
= &speex_uwb_mode
;
887 msg_Dbg( p_enc
, "Using narrowband" );
888 p_speex_mode
= &speex_nb_mode
;
892 /* Allocate the memory needed to store the decoder's structure */
893 if( ( p_sys
= (encoder_sys_t
*)malloc(sizeof(encoder_sys_t
)) ) == NULL
)
895 p_enc
->p_sys
= p_sys
;
896 p_enc
->pf_encode_audio
= Encode
;
897 p_enc
->fmt_in
.i_codec
= VLC_CODEC_S16N
;
898 p_enc
->fmt_out
.i_codec
= VLC_CODEC_SPEEX
;
900 speex_init_header( &p_sys
->header
, p_enc
->fmt_in
.audio
.i_rate
,
903 p_sys
->header
.frames_per_packet
= 1;
904 p_sys
->header
.vbr
= var_GetBool( p_enc
, ENC_CFG_PREFIX
"cbr" ) ? 0 : 1;
905 p_sys
->header
.nb_channels
= p_enc
->fmt_in
.audio
.i_channels
;
907 /* Create a new encoder state in narrowband mode */
908 p_sys
->p_state
= speex_encoder_init( p_speex_mode
);
911 i_tmp
= var_GetInteger( p_enc
, ENC_CFG_PREFIX
"complexity" );
912 speex_encoder_ctl( p_sys
->p_state
, SPEEX_SET_COMPLEXITY
, &i_tmp
);
914 i_tmp
= var_GetBool( p_enc
, ENC_CFG_PREFIX
"cbr" ) ? 0 : 1;
915 speex_encoder_ctl( p_sys
->p_state
, SPEEX_SET_VBR
, &i_tmp
);
917 if( i_tmp
== 0 ) /* CBR */
919 i_tmp
= var_GetFloat( p_enc
, ENC_CFG_PREFIX
"quality" );
920 speex_encoder_ctl( p_sys
->p_state
, SPEEX_SET_QUALITY
, &i_tmp
);
922 i_tmp
= var_GetBool( p_enc
, ENC_CFG_PREFIX
"vad" ) ? 1 : 0;
923 speex_encoder_ctl( p_sys
->p_state
, SPEEX_SET_VAD
, &i_tmp
);
929 f_tmp
= var_GetFloat( p_enc
, ENC_CFG_PREFIX
"quality" );
930 speex_encoder_ctl( p_sys
->p_state
, SPEEX_SET_VBR_QUALITY
, &f_tmp
);
932 i_tmp
= var_GetInteger( p_enc
, ENC_CFG_PREFIX
"max-bitrate" );
934 #ifdef SPEEX_SET_VBR_MAX_BITRATE
935 speex_encoder_ctl( p_sys
->p_state
, SPEEX_SET_VBR_MAX_BITRATE
, &i_tmp
);
937 msg_Dbg( p_enc
, "max-bitrate cannot be set in this version of libspeex");
941 i_tmp
= var_GetBool( p_enc
, ENC_CFG_PREFIX
"dtx" ) ? 1 : 0;
942 speex_encoder_ctl( p_sys
->p_state
, SPEEX_SET_DTX
, &i_tmp
);
945 /*Initialization of the structure that holds the bits*/
946 speex_bits_init( &p_sys
->bits
);
948 p_sys
->i_frames_in_packet
= 0;
949 p_sys
->i_samples_delay
= 0;
951 speex_encoder_ctl( p_sys
->p_state
, SPEEX_GET_FRAME_SIZE
,
952 &p_sys
->i_frame_length
);
954 p_sys
->i_frame_size
= p_sys
->i_frame_length
*
955 sizeof(int16_t) * p_enc
->fmt_in
.audio
.i_channels
;
956 p_sys
->p_buffer
= xmalloc( p_sys
->i_frame_size
);
958 /* Create and store headers */
959 pp_header
[0] = speex_header_to_packet( &p_sys
->header
, &pi_header
[0] );
960 pp_header
[1] = "ENCODER=VLC media player";
961 pi_header
[1] = sizeof("ENCODER=VLC media player");
963 p_enc
->fmt_out
.i_extra
= 3 * 2 + pi_header
[0] + pi_header
[1];
964 p_extra
= p_enc
->fmt_out
.p_extra
= xmalloc( p_enc
->fmt_out
.i_extra
);
965 for( i
= 0; i
< 2; i
++ )
967 *(p_extra
++) = pi_header
[i
] >> 8;
968 *(p_extra
++) = pi_header
[i
] & 0xFF;
969 memcpy( p_extra
, pp_header
[i
], pi_header
[i
] );
970 p_extra
+= pi_header
[i
];
973 msg_Dbg( p_enc
, "encoding: frame size:%d, channels:%d, samplerate:%d",
974 p_sys
->i_frame_size
, p_enc
->fmt_in
.audio
.i_channels
,
975 p_enc
->fmt_in
.audio
.i_rate
);
980 /****************************************************************************
981 * Encode: the whole thing
982 ****************************************************************************
983 * This function spits out ogg packets.
984 ****************************************************************************/
985 static block_t
*Encode( encoder_t
*p_enc
, aout_buffer_t
*p_aout_buf
)
987 encoder_sys_t
*p_sys
= p_enc
->p_sys
;
988 block_t
*p_block
, *p_chain
= NULL
;
990 unsigned char *p_buffer
= p_aout_buf
->p_buffer
;
991 int i_samples
= p_aout_buf
->i_nb_samples
;
992 int i_samples_delay
= p_sys
->i_samples_delay
;
994 mtime_t i_pts
= p_aout_buf
->i_pts
-
995 (mtime_t
)1000000 * (mtime_t
)p_sys
->i_samples_delay
/
996 (mtime_t
)p_enc
->fmt_in
.audio
.i_rate
;
998 p_sys
->i_samples_delay
+= i_samples
;
1000 while( p_sys
->i_samples_delay
>= p_sys
->i_frame_length
)
1005 if( i_samples_delay
)
1007 /* Take care of the left-over from last time */
1008 int i_delay_size
= i_samples_delay
* 2 *
1009 p_enc
->fmt_in
.audio
.i_channels
;
1010 int i_size
= p_sys
->i_frame_size
- i_delay_size
;
1012 p_samples
= (int16_t *)p_sys
->p_buffer
;
1013 memcpy( p_sys
->p_buffer
+ i_delay_size
, p_buffer
, i_size
);
1014 p_buffer
-= i_delay_size
;
1015 i_samples
+= i_samples_delay
;
1016 i_samples_delay
= 0;
1020 p_samples
= (int16_t *)p_buffer
;
1023 /* Encode current frame */
1024 if( p_enc
->fmt_in
.audio
.i_channels
== 2 )
1025 speex_encode_stereo_int( p_samples
, p_sys
->i_frame_length
,
1029 if( p_sys
->preprocess
)
1030 speex_preprocess( p_sys
->preprocess
, p_samples
, NULL
);
1033 speex_encode_int( p_sys
->p_state
, p_samples
, &p_sys
->bits
);
1035 p_buffer
+= p_sys
->i_frame_size
;
1036 p_sys
->i_samples_delay
-= p_sys
->i_frame_length
;
1037 i_samples
-= p_sys
->i_frame_length
;
1039 p_sys
->i_frames_in_packet
++;
1041 if( p_sys
->i_frames_in_packet
< p_sys
->header
.frames_per_packet
)
1044 p_sys
->i_frames_in_packet
= 0;
1046 speex_bits_insert_terminator( &p_sys
->bits
);
1047 i_out
= speex_bits_write( &p_sys
->bits
, p_sys
->p_buffer_out
,
1049 speex_bits_reset( &p_sys
->bits
);
1051 p_block
= block_New( p_enc
, i_out
);
1052 memcpy( p_block
->p_buffer
, p_sys
->p_buffer_out
, i_out
);
1054 p_block
->i_length
= (mtime_t
)1000000 *
1055 (mtime_t
)p_sys
->i_frame_length
* p_sys
->header
.frames_per_packet
/
1056 (mtime_t
)p_enc
->fmt_in
.audio
.i_rate
;
1058 p_block
->i_dts
= p_block
->i_pts
= i_pts
;
1061 i_pts
+= p_block
->i_length
;
1062 block_ChainAppend( &p_chain
, p_block
);
1066 /* Backup the remaining raw samples */
1069 memcpy( p_sys
->p_buffer
+ i_samples_delay
* 2 *
1070 p_enc
->fmt_in
.audio
.i_channels
, p_buffer
,
1071 i_samples
* 2 * p_enc
->fmt_in
.audio
.i_channels
);
1077 /*****************************************************************************
1078 * CloseEncoder: encoder destruction
1079 *****************************************************************************/
1080 static void CloseEncoder( vlc_object_t
*p_this
)
1082 encoder_t
*p_enc
= (encoder_t
*)p_this
;
1083 encoder_sys_t
*p_sys
= p_enc
->p_sys
;
1085 speex_encoder_destroy( p_sys
->p_state
);
1086 speex_bits_destroy( &p_sys
->bits
);
1088 free( p_sys
->p_buffer
);