1 /* GStreamer
2 * Copyright (C) 2004 Wim Taymans <wim@fluendo.com>
3 * Copyright (C) 2006 Tim-Philipp Müller <tim centricular net>
4 *
5 * This library is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU Library General Public
7 * License as published by the Free Software Foundation; either
8 * version 2 of the License, or (at your option) any later version.
9 *
10 * This library is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * Library General Public License for more details.
14 *
15 * You should have received a copy of the GNU Library General Public
16 * License along with this library; if not, write to the
17 * Free Software Foundation, Inc., 51 Franklin St, Fifth Floor,
18 * Boston, MA 02110-1301, USA.
19 */
20
21 /**
22 * SECTION:element-speexdec
23 * @title: speexdec
24 * @see_also: speexenc, oggdemux
25 *
26 * This element decodes a Speex stream to raw integer audio.
27 * [Speex](http://www.speex.org/) is a royalty-free
28 * audio codec maintained by the [Xiph.org Foundation](http://www.xiph.org/).
29 *
30 * ## Example pipelines
31 * |[
32 * gst-launch-1.0 -v filesrc location=speex.ogg ! oggdemux ! speexdec ! audioconvert ! audioresample ! alsasink
33 * ]| Decode an Ogg/Speex file. To create an Ogg/Speex file refer to the
34 * documentation of speexenc.
35 *
36 */
37
38 #ifdef HAVE_CONFIG_H
39 # include "config.h"
40 #endif
41
42 #include "gstspeexelements.h"
43 #include "gstspeexdec.h"
44 #include <stdlib.h>
45 #include <string.h>
46 #include <gst/tag/tag.h>
47 #include <gst/audio/audio.h>
48
49 GST_DEBUG_CATEGORY_STATIC (speexdec_debug);
50 #define GST_CAT_DEFAULT speexdec_debug
51
52 #define DEFAULT_ENH TRUE
53
54 enum
55 {
56 ARG_0,
57 ARG_ENH
58 };
59
60 #define FORMAT_STR GST_AUDIO_NE(S16)
61
62 static GstStaticPadTemplate speex_dec_src_factory =
63 GST_STATIC_PAD_TEMPLATE ("src",
64 GST_PAD_SRC,
65 GST_PAD_ALWAYS,
66 GST_STATIC_CAPS ("audio/x-raw, "
67 "format = (string) " FORMAT_STR ", "
68 "layout = (string) interleaved, "
69 "rate = (int) [ 6000, 48000 ], " "channels = (int) [ 1, 2 ]")
70 );
71
72 static GstStaticPadTemplate speex_dec_sink_factory =
73 GST_STATIC_PAD_TEMPLATE ("sink",
74 GST_PAD_SINK,
75 GST_PAD_ALWAYS,
76 GST_STATIC_CAPS ("audio/x-speex")
77 );
78
79 #define gst_speex_dec_parent_class parent_class
80 G_DEFINE_TYPE (GstSpeexDec, gst_speex_dec, GST_TYPE_AUDIO_DECODER);
81 GST_ELEMENT_REGISTER_DEFINE_WITH_CODE (speexdec, "speexdec",
82 GST_RANK_PRIMARY, GST_TYPE_SPEEX_DEC, speex_element_init (plugin));
83
84 static gboolean gst_speex_dec_start (GstAudioDecoder * dec);
85 static gboolean gst_speex_dec_stop (GstAudioDecoder * dec);
86 static gboolean gst_speex_dec_set_format (GstAudioDecoder * bdec,
87 GstCaps * caps);
88 static GstFlowReturn gst_speex_dec_handle_frame (GstAudioDecoder * dec,
89 GstBuffer * buffer);
90
91 static void gst_speex_dec_get_property (GObject * object, guint prop_id,
92 GValue * value, GParamSpec * pspec);
93 static void gst_speex_dec_set_property (GObject * object, guint prop_id,
94 const GValue * value, GParamSpec * pspec);
95
96 static void
gst_speex_dec_class_init(GstSpeexDecClass * klass)97 gst_speex_dec_class_init (GstSpeexDecClass * klass)
98 {
99 GObjectClass *gobject_class;
100 GstElementClass *gstelement_class;
101 GstAudioDecoderClass *base_class;
102
103 gobject_class = (GObjectClass *) klass;
104 gstelement_class = (GstElementClass *) klass;
105 base_class = (GstAudioDecoderClass *) klass;
106
107 gobject_class->set_property = gst_speex_dec_set_property;
108 gobject_class->get_property = gst_speex_dec_get_property;
109
110 base_class->start = GST_DEBUG_FUNCPTR (gst_speex_dec_start);
111 base_class->stop = GST_DEBUG_FUNCPTR (gst_speex_dec_stop);
112 base_class->set_format = GST_DEBUG_FUNCPTR (gst_speex_dec_set_format);
113 base_class->handle_frame = GST_DEBUG_FUNCPTR (gst_speex_dec_handle_frame);
114
115 g_object_class_install_property (G_OBJECT_CLASS (klass), ARG_ENH,
116 g_param_spec_boolean ("enh", "Enh", "Enable perceptual enhancement",
117 DEFAULT_ENH, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
118
119 gst_element_class_add_static_pad_template (gstelement_class,
120 &speex_dec_src_factory);
121 gst_element_class_add_static_pad_template (gstelement_class,
122 &speex_dec_sink_factory);
123 gst_element_class_set_static_metadata (gstelement_class,
124 "Speex audio decoder", "Codec/Decoder/Audio",
125 "decode speex streams to audio", "Wim Taymans <wim@fluendo.com>");
126
127 GST_DEBUG_CATEGORY_INIT (speexdec_debug, "speexdec", 0,
128 "speex decoding element");
129 }
130
131 static void
gst_speex_dec_reset(GstSpeexDec * dec)132 gst_speex_dec_reset (GstSpeexDec * dec)
133 {
134 dec->packetno = 0;
135 dec->frame_size = 0;
136 dec->frame_duration = 0;
137 dec->mode = NULL;
138 speex_header_free (dec->header);
139 dec->header = NULL;
140 speex_bits_destroy (&dec->bits);
141 speex_bits_set_bit_buffer (&dec->bits, NULL, 0);
142
143 gst_buffer_replace (&dec->streamheader, NULL);
144 gst_buffer_replace (&dec->vorbiscomment, NULL);
145
146 if (dec->stereo) {
147 speex_stereo_state_destroy (dec->stereo);
148 dec->stereo = NULL;
149 }
150
151 if (dec->state) {
152 speex_decoder_destroy (dec->state);
153 dec->state = NULL;
154 }
155 }
156
157 static void
gst_speex_dec_init(GstSpeexDec * dec)158 gst_speex_dec_init (GstSpeexDec * dec)
159 {
160 gst_audio_decoder_set_needs_format (GST_AUDIO_DECODER (dec), TRUE);
161 gst_audio_decoder_set_use_default_pad_acceptcaps (GST_AUDIO_DECODER_CAST
162 (dec), TRUE);
163 GST_PAD_SET_ACCEPT_TEMPLATE (GST_AUDIO_DECODER_SINK_PAD (dec));
164
165 dec->enh = DEFAULT_ENH;
166
167 gst_speex_dec_reset (dec);
168 }
169
170 static gboolean
gst_speex_dec_start(GstAudioDecoder * dec)171 gst_speex_dec_start (GstAudioDecoder * dec)
172 {
173 GstSpeexDec *sd = GST_SPEEX_DEC (dec);
174
175 GST_DEBUG_OBJECT (dec, "start");
176 gst_speex_dec_reset (sd);
177
178 /* we know about concealment */
179 gst_audio_decoder_set_plc_aware (dec, TRUE);
180
181 return TRUE;
182 }
183
184 static gboolean
gst_speex_dec_stop(GstAudioDecoder * dec)185 gst_speex_dec_stop (GstAudioDecoder * dec)
186 {
187 GstSpeexDec *sd = GST_SPEEX_DEC (dec);
188
189 GST_DEBUG_OBJECT (dec, "stop");
190 gst_speex_dec_reset (sd);
191
192 return TRUE;
193 }
194
195 static GstFlowReturn
gst_speex_dec_parse_header(GstSpeexDec * dec,GstBuffer * buf)196 gst_speex_dec_parse_header (GstSpeexDec * dec, GstBuffer * buf)
197 {
198 GstMapInfo map;
199 GstAudioInfo info;
200 static const GstAudioChannelPosition chan_pos[2][2] = {
201 {GST_AUDIO_CHANNEL_POSITION_MONO},
202 {GST_AUDIO_CHANNEL_POSITION_FRONT_LEFT,
203 GST_AUDIO_CHANNEL_POSITION_FRONT_RIGHT}
204 };
205
206 /* get the header */
207 gst_buffer_map (buf, &map, GST_MAP_READ);
208 dec->header = speex_packet_to_header ((gchar *) map.data, map.size);
209 gst_buffer_unmap (buf, &map);
210
211 if (!dec->header)
212 goto no_header;
213
214 if (dec->header->mode >= SPEEX_NB_MODES || dec->header->mode < 0)
215 goto mode_too_old;
216
217 dec->mode = speex_lib_get_mode (dec->header->mode);
218
219 /* initialize the decoder */
220 dec->state = speex_decoder_init (dec->mode);
221 if (!dec->state)
222 goto init_failed;
223
224 speex_decoder_ctl (dec->state, SPEEX_SET_ENH, &dec->enh);
225 speex_decoder_ctl (dec->state, SPEEX_GET_FRAME_SIZE, &dec->frame_size);
226
227 if (dec->header->nb_channels != 1) {
228 dec->stereo = speex_stereo_state_init ();
229 dec->callback.callback_id = SPEEX_INBAND_STEREO;
230 dec->callback.func = speex_std_stereo_request_handler;
231 dec->callback.data = dec->stereo;
232 speex_decoder_ctl (dec->state, SPEEX_SET_HANDLER, &dec->callback);
233 }
234
235 speex_decoder_ctl (dec->state, SPEEX_SET_SAMPLING_RATE, &dec->header->rate);
236
237 dec->frame_duration = gst_util_uint64_scale_int (dec->frame_size,
238 GST_SECOND, dec->header->rate);
239
240 speex_bits_init (&dec->bits);
241
242 /* set caps */
243 gst_audio_info_init (&info);
244 gst_audio_info_set_format (&info,
245 GST_AUDIO_FORMAT_S16,
246 dec->header->rate,
247 dec->header->nb_channels, chan_pos[dec->header->nb_channels - 1]);
248
249 if (!gst_audio_decoder_set_output_format (GST_AUDIO_DECODER (dec), &info))
250 goto nego_failed;
251
252 return GST_FLOW_OK;
253
254 /* ERRORS */
255 no_header:
256 {
257 GST_ELEMENT_ERROR (GST_ELEMENT (dec), STREAM, DECODE,
258 (NULL), ("couldn't read header"));
259 return GST_FLOW_ERROR;
260 }
261 mode_too_old:
262 {
263 GST_ELEMENT_ERROR (GST_ELEMENT (dec), STREAM, DECODE,
264 (NULL),
265 ("Mode number %d does not (yet/any longer) exist in this version",
266 dec->header->mode));
267 return GST_FLOW_ERROR;
268 }
269 init_failed:
270 {
271 GST_ELEMENT_ERROR (GST_ELEMENT (dec), STREAM, DECODE,
272 (NULL), ("couldn't initialize decoder"));
273 return GST_FLOW_ERROR;
274 }
275 nego_failed:
276 {
277 GST_ELEMENT_ERROR (GST_ELEMENT (dec), STREAM, DECODE,
278 (NULL), ("couldn't negotiate format"));
279 return GST_FLOW_NOT_NEGOTIATED;
280 }
281 }
282
283 static GstFlowReturn
gst_speex_dec_parse_comments(GstSpeexDec * dec,GstBuffer * buf)284 gst_speex_dec_parse_comments (GstSpeexDec * dec, GstBuffer * buf)
285 {
286 GstTagList *list;
287 gchar *ver, *encoder = NULL;
288
289 list = gst_tag_list_from_vorbiscomment_buffer (buf, NULL, 0, &encoder);
290
291 if (!list) {
292 GST_WARNING_OBJECT (dec, "couldn't decode comments");
293 list = gst_tag_list_new_empty ();
294 }
295
296 if (encoder) {
297 gst_tag_list_add (list, GST_TAG_MERGE_REPLACE,
298 GST_TAG_ENCODER, encoder, NULL);
299 }
300
301 gst_tag_list_add (list, GST_TAG_MERGE_REPLACE,
302 GST_TAG_AUDIO_CODEC, "Speex", NULL);
303
304 ver = g_strndup (dec->header->speex_version, SPEEX_HEADER_VERSION_LENGTH);
305 g_strstrip (ver);
306
307 if (ver != NULL && *ver != '\0') {
308 gst_tag_list_add (list, GST_TAG_MERGE_REPLACE,
309 GST_TAG_ENCODER_VERSION, ver, NULL);
310 }
311
312 if (dec->header->bitrate > 0) {
313 gst_tag_list_add (list, GST_TAG_MERGE_REPLACE,
314 GST_TAG_BITRATE, (guint) dec->header->bitrate, NULL);
315 }
316
317 GST_INFO_OBJECT (dec, "tags: %" GST_PTR_FORMAT, list);
318
319 gst_audio_decoder_merge_tags (GST_AUDIO_DECODER (dec), list,
320 GST_TAG_MERGE_REPLACE);
321 gst_tag_list_unref (list);
322
323 g_free (encoder);
324 g_free (ver);
325
326 return GST_FLOW_OK;
327 }
328
329 static gboolean
gst_speex_dec_set_format(GstAudioDecoder * bdec,GstCaps * caps)330 gst_speex_dec_set_format (GstAudioDecoder * bdec, GstCaps * caps)
331 {
332 GstSpeexDec *dec = GST_SPEEX_DEC (bdec);
333 gboolean ret = TRUE;
334 GstStructure *s;
335 const GValue *streamheader;
336
337 s = gst_caps_get_structure (caps, 0);
338 if ((streamheader = gst_structure_get_value (s, "streamheader")) &&
339 G_VALUE_HOLDS (streamheader, GST_TYPE_ARRAY) &&
340 gst_value_array_get_size (streamheader) >= 2) {
341 const GValue *header, *vorbiscomment;
342 GstBuffer *buf;
343 GstFlowReturn res = GST_FLOW_OK;
344
345 header = gst_value_array_get_value (streamheader, 0);
346 if (header && G_VALUE_HOLDS (header, GST_TYPE_BUFFER)) {
347 buf = gst_value_get_buffer (header);
348 res = gst_speex_dec_parse_header (dec, buf);
349 if (res != GST_FLOW_OK)
350 goto done;
351 gst_buffer_replace (&dec->streamheader, buf);
352 }
353
354 vorbiscomment = gst_value_array_get_value (streamheader, 1);
355 if (vorbiscomment && G_VALUE_HOLDS (vorbiscomment, GST_TYPE_BUFFER)) {
356 buf = gst_value_get_buffer (vorbiscomment);
357 res = gst_speex_dec_parse_comments (dec, buf);
358 if (res != GST_FLOW_OK)
359 goto done;
360 gst_buffer_replace (&dec->vorbiscomment, buf);
361 }
362 }
363
364 done:
365 return ret;
366 }
367
368 static GstFlowReturn
gst_speex_dec_parse_data(GstSpeexDec * dec,GstBuffer * buf)369 gst_speex_dec_parse_data (GstSpeexDec * dec, GstBuffer * buf)
370 {
371 GstFlowReturn res = GST_FLOW_OK;
372 gint i, fpp;
373 SpeexBits *bits;
374 GstMapInfo map;
375
376 if (!dec->frame_duration)
377 goto not_negotiated;
378
379 if (G_LIKELY (gst_buffer_get_size (buf))) {
380 /* send data to the bitstream */
381 gst_buffer_map (buf, &map, GST_MAP_READ);
382 speex_bits_read_from (&dec->bits, (gchar *) map.data, map.size);
383 gst_buffer_unmap (buf, &map);
384
385 fpp = dec->header->frames_per_packet;
386 bits = &dec->bits;
387
388 GST_DEBUG_OBJECT (dec, "received buffer of size %" G_GSIZE_FORMAT
389 ", fpp %d, %d bits", map.size, fpp, speex_bits_remaining (bits));
390 } else {
391 /* FIXME ? actually consider how much concealment is needed */
392 /* concealment data, pass NULL as the bits parameters */
393 GST_DEBUG_OBJECT (dec, "creating concealment data");
394 fpp = dec->header->frames_per_packet;
395 bits = NULL;
396 }
397
398 /* now decode each frame, catering for unknown number of them (e.g. rtp) */
399 for (i = 0; i < fpp; i++) {
400 GstBuffer *outbuf;
401 gboolean corrupted = FALSE;
402 gint ret;
403
404 GST_LOG_OBJECT (dec, "decoding frame %d/%d, %d bits remaining", i, fpp,
405 bits ? speex_bits_remaining (bits) : -1);
406 #if 0
407 res =
408 gst_pad_alloc_buffer_and_set_caps (GST_AUDIO_DECODER_SRC_PAD (dec),
409 GST_BUFFER_OFFSET_NONE, dec->frame_size * dec->header->nb_channels * 2,
410 GST_PAD_CAPS (GST_AUDIO_DECODER_SRC_PAD (dec)), &outbuf);
411
412 if (res != GST_FLOW_OK) {
413 GST_DEBUG_OBJECT (dec, "buf alloc flow: %s", gst_flow_get_name (res));
414 return res;
415 }
416 #endif
417 /* FIXME, we can use a bufferpool because we have fixed size buffers. We
418 * could also use an allocator */
419 outbuf =
420 gst_buffer_new_allocate (NULL,
421 dec->frame_size * dec->header->nb_channels * 2, NULL);
422
423 gst_buffer_map (outbuf, &map, GST_MAP_WRITE);
424 ret = speex_decode_int (dec->state, bits, (spx_int16_t *) map.data);
425
426 if (ret == -1) {
427 /* uh? end of stream */
428 GST_WARNING_OBJECT (dec, "Unexpected end of stream found");
429 corrupted = TRUE;
430 } else if (ret == -2) {
431 GST_WARNING_OBJECT (dec, "Decoding error: corrupted stream?");
432 corrupted = TRUE;
433 }
434
435 if (bits && speex_bits_remaining (bits) < 0) {
436 GST_WARNING_OBJECT (dec, "Decoding overflow: corrupted stream?");
437 corrupted = TRUE;
438 }
439 if (dec->header->nb_channels == 2)
440 speex_decode_stereo_int ((spx_int16_t *) map.data, dec->frame_size,
441 dec->stereo);
442
443 gst_buffer_unmap (outbuf, &map);
444
445 if (!corrupted) {
446 res = gst_audio_decoder_finish_frame (GST_AUDIO_DECODER (dec), outbuf, 1);
447 } else {
448 res = gst_audio_decoder_finish_frame (GST_AUDIO_DECODER (dec), NULL, 1);
449 gst_buffer_unref (outbuf);
450 }
451
452 if (res != GST_FLOW_OK) {
453 GST_DEBUG_OBJECT (dec, "flow: %s", gst_flow_get_name (res));
454 break;
455 }
456 }
457
458 return res;
459
460 /* ERRORS */
461 not_negotiated:
462 {
463 GST_ELEMENT_ERROR (dec, CORE, NEGOTIATION, (NULL),
464 ("decoder not initialized"));
465 return GST_FLOW_NOT_NEGOTIATED;
466 }
467 }
468
469 static gboolean
memcmp_buffers(GstBuffer * buf1,GstBuffer * buf2)470 memcmp_buffers (GstBuffer * buf1, GstBuffer * buf2)
471 {
472 GstMapInfo map;
473 gsize size1, size2;
474 gboolean res;
475
476 size1 = gst_buffer_get_size (buf1);
477 size2 = gst_buffer_get_size (buf2);
478
479 if (size1 != size2)
480 return FALSE;
481
482 gst_buffer_map (buf1, &map, GST_MAP_READ);
483 res = gst_buffer_memcmp (buf2, 0, map.data, map.size) == 0;
484 gst_buffer_unmap (buf1, &map);
485
486 return res;
487 }
488
489 static GstFlowReturn
gst_speex_dec_handle_frame(GstAudioDecoder * bdec,GstBuffer * buf)490 gst_speex_dec_handle_frame (GstAudioDecoder * bdec, GstBuffer * buf)
491 {
492 GstFlowReturn res;
493 GstSpeexDec *dec;
494
495 /* no fancy draining */
496 if (G_UNLIKELY (!buf))
497 return GST_FLOW_OK;
498
499 dec = GST_SPEEX_DEC (bdec);
500
501 /* If we have the streamheader and vorbiscomment from the caps already
502 * ignore them here */
503 if (dec->streamheader && dec->vorbiscomment) {
504 if (memcmp_buffers (dec->streamheader, buf)) {
505 GST_DEBUG_OBJECT (dec, "found streamheader");
506 gst_audio_decoder_finish_frame (bdec, NULL, 1);
507 res = GST_FLOW_OK;
508 } else if (memcmp_buffers (dec->vorbiscomment, buf)) {
509 GST_DEBUG_OBJECT (dec, "found vorbiscomments");
510 gst_audio_decoder_finish_frame (bdec, NULL, 1);
511 res = GST_FLOW_OK;
512 } else {
513 res = gst_speex_dec_parse_data (dec, buf);
514 }
515 } else {
516 /* Otherwise fall back to packet counting and assume that the
517 * first two packets are the headers. */
518 switch (dec->packetno) {
519 case 0:
520 GST_DEBUG_OBJECT (dec, "counted streamheader");
521 res = gst_speex_dec_parse_header (dec, buf);
522 gst_audio_decoder_finish_frame (bdec, NULL, 1);
523 break;
524 case 1:
525 GST_DEBUG_OBJECT (dec, "counted vorbiscomments");
526 res = gst_speex_dec_parse_comments (dec, buf);
527 gst_audio_decoder_finish_frame (bdec, NULL, 1);
528 break;
529 default:
530 {
531 res = gst_speex_dec_parse_data (dec, buf);
532 break;
533 }
534 }
535 }
536
537 dec->packetno++;
538
539 return res;
540 }
541
542 static void
gst_speex_dec_get_property(GObject * object,guint prop_id,GValue * value,GParamSpec * pspec)543 gst_speex_dec_get_property (GObject * object, guint prop_id,
544 GValue * value, GParamSpec * pspec)
545 {
546 GstSpeexDec *speexdec;
547
548 speexdec = GST_SPEEX_DEC (object);
549
550 switch (prop_id) {
551 case ARG_ENH:
552 g_value_set_boolean (value, speexdec->enh);
553 break;
554 default:
555 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
556 break;
557 }
558 }
559
560 static void
gst_speex_dec_set_property(GObject * object,guint prop_id,const GValue * value,GParamSpec * pspec)561 gst_speex_dec_set_property (GObject * object, guint prop_id,
562 const GValue * value, GParamSpec * pspec)
563 {
564 GstSpeexDec *speexdec;
565
566 speexdec = GST_SPEEX_DEC (object);
567
568 switch (prop_id) {
569 case ARG_ENH:
570 speexdec->enh = g_value_get_boolean (value);
571 break;
572 default:
573 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
574 break;
575 }
576 }
577