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 * @see_also: speexenc, oggdemux
24 *
25 * This element decodes a Speex stream to raw integer audio.
26 * <ulink url="http://www.speex.org/">Speex</ulink> is a royalty-free
27 * audio codec maintained by the <ulink url="http://www.xiph.org/">Xiph.org
28 * Foundation</ulink>.
29 *
30 * <refsect2>
31 * <title>Example pipelines</title>
32 * |[
33 * gst-launch-1.0 -v filesrc location=speex.ogg ! oggdemux ! speexdec ! audioconvert ! audioresample ! alsasink
34 * ]| Decode an Ogg/Speex file. To create an Ogg/Speex file refer to the
35 * documentation of speexenc.
36 * </refsect2>
37 */
38
39 #ifdef HAVE_CONFIG_H
40 # include "config.h"
41 #endif
42
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
82 static gboolean gst_speex_dec_start (GstAudioDecoder * dec);
83 static gboolean gst_speex_dec_stop (GstAudioDecoder * dec);
84 static gboolean gst_speex_dec_set_format (GstAudioDecoder * bdec,
85 GstCaps * caps);
86 static GstFlowReturn gst_speex_dec_handle_frame (GstAudioDecoder * dec,
87 GstBuffer * buffer);
88
89 static void gst_speex_dec_get_property (GObject * object, guint prop_id,
90 GValue * value, GParamSpec * pspec);
91 static void gst_speex_dec_set_property (GObject * object, guint prop_id,
92 const GValue * value, GParamSpec * pspec);
93
94 static void
gst_speex_dec_class_init(GstSpeexDecClass * klass)95 gst_speex_dec_class_init (GstSpeexDecClass * klass)
96 {
97 GObjectClass *gobject_class;
98 GstElementClass *gstelement_class;
99 GstAudioDecoderClass *base_class;
100
101 gobject_class = (GObjectClass *) klass;
102 gstelement_class = (GstElementClass *) klass;
103 base_class = (GstAudioDecoderClass *) klass;
104
105 gobject_class->set_property = gst_speex_dec_set_property;
106 gobject_class->get_property = gst_speex_dec_get_property;
107
108 base_class->start = GST_DEBUG_FUNCPTR (gst_speex_dec_start);
109 base_class->stop = GST_DEBUG_FUNCPTR (gst_speex_dec_stop);
110 base_class->set_format = GST_DEBUG_FUNCPTR (gst_speex_dec_set_format);
111 base_class->handle_frame = GST_DEBUG_FUNCPTR (gst_speex_dec_handle_frame);
112
113 g_object_class_install_property (G_OBJECT_CLASS (klass), ARG_ENH,
114 g_param_spec_boolean ("enh", "Enh", "Enable perceptual enhancement",
115 DEFAULT_ENH, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
116
117 gst_element_class_add_static_pad_template (gstelement_class,
118 &speex_dec_src_factory);
119 gst_element_class_add_static_pad_template (gstelement_class,
120 &speex_dec_sink_factory);
121 gst_element_class_set_static_metadata (gstelement_class,
122 "Speex audio decoder", "Codec/Decoder/Audio",
123 "decode speex streams to audio", "Wim Taymans <wim@fluendo.com>");
124
125 GST_DEBUG_CATEGORY_INIT (speexdec_debug, "speexdec", 0,
126 "speex decoding element");
127 }
128
129 static void
gst_speex_dec_reset(GstSpeexDec * dec)130 gst_speex_dec_reset (GstSpeexDec * dec)
131 {
132 dec->packetno = 0;
133 dec->frame_size = 0;
134 dec->frame_duration = 0;
135 dec->mode = NULL;
136 free (dec->header);
137 dec->header = NULL;
138 speex_bits_destroy (&dec->bits);
139 speex_bits_set_bit_buffer (&dec->bits, NULL, 0);
140
141 gst_buffer_replace (&dec->streamheader, NULL);
142 gst_buffer_replace (&dec->vorbiscomment, NULL);
143
144 if (dec->stereo) {
145 speex_stereo_state_destroy (dec->stereo);
146 dec->stereo = NULL;
147 }
148
149 if (dec->state) {
150 speex_decoder_destroy (dec->state);
151 dec->state = NULL;
152 }
153 }
154
155 static void
gst_speex_dec_init(GstSpeexDec * dec)156 gst_speex_dec_init (GstSpeexDec * dec)
157 {
158 gst_audio_decoder_set_needs_format (GST_AUDIO_DECODER (dec), TRUE);
159 gst_audio_decoder_set_use_default_pad_acceptcaps (GST_AUDIO_DECODER_CAST
160 (dec), TRUE);
161 GST_PAD_SET_ACCEPT_TEMPLATE (GST_AUDIO_DECODER_SINK_PAD (dec));
162
163 dec->enh = DEFAULT_ENH;
164
165 gst_speex_dec_reset (dec);
166 }
167
168 static gboolean
gst_speex_dec_start(GstAudioDecoder * dec)169 gst_speex_dec_start (GstAudioDecoder * dec)
170 {
171 GstSpeexDec *sd = GST_SPEEX_DEC (dec);
172
173 GST_DEBUG_OBJECT (dec, "start");
174 gst_speex_dec_reset (sd);
175
176 /* we know about concealment */
177 gst_audio_decoder_set_plc_aware (dec, TRUE);
178
179 return TRUE;
180 }
181
182 static gboolean
gst_speex_dec_stop(GstAudioDecoder * dec)183 gst_speex_dec_stop (GstAudioDecoder * dec)
184 {
185 GstSpeexDec *sd = GST_SPEEX_DEC (dec);
186
187 GST_DEBUG_OBJECT (dec, "stop");
188 gst_speex_dec_reset (sd);
189
190 return TRUE;
191 }
192
193 static GstFlowReturn
gst_speex_dec_parse_header(GstSpeexDec * dec,GstBuffer * buf)194 gst_speex_dec_parse_header (GstSpeexDec * dec, GstBuffer * buf)
195 {
196 GstMapInfo map;
197 GstAudioInfo info;
198 static const GstAudioChannelPosition chan_pos[2][2] = {
199 {GST_AUDIO_CHANNEL_POSITION_MONO},
200 {GST_AUDIO_CHANNEL_POSITION_FRONT_LEFT,
201 GST_AUDIO_CHANNEL_POSITION_FRONT_RIGHT}
202 };
203
204 /* get the header */
205 gst_buffer_map (buf, &map, GST_MAP_READ);
206 dec->header = speex_packet_to_header ((gchar *) map.data, map.size);
207 gst_buffer_unmap (buf, &map);
208
209 if (!dec->header)
210 goto no_header;
211
212 if (dec->header->mode >= SPEEX_NB_MODES || dec->header->mode < 0)
213 goto mode_too_old;
214
215 dec->mode = speex_lib_get_mode (dec->header->mode);
216
217 /* initialize the decoder */
218 dec->state = speex_decoder_init (dec->mode);
219 if (!dec->state)
220 goto init_failed;
221
222 speex_decoder_ctl (dec->state, SPEEX_SET_ENH, &dec->enh);
223 speex_decoder_ctl (dec->state, SPEEX_GET_FRAME_SIZE, &dec->frame_size);
224
225 if (dec->header->nb_channels != 1) {
226 dec->stereo = speex_stereo_state_init ();
227 dec->callback.callback_id = SPEEX_INBAND_STEREO;
228 dec->callback.func = speex_std_stereo_request_handler;
229 dec->callback.data = dec->stereo;
230 speex_decoder_ctl (dec->state, SPEEX_SET_HANDLER, &dec->callback);
231 }
232
233 speex_decoder_ctl (dec->state, SPEEX_SET_SAMPLING_RATE, &dec->header->rate);
234
235 dec->frame_duration = gst_util_uint64_scale_int (dec->frame_size,
236 GST_SECOND, dec->header->rate);
237
238 speex_bits_init (&dec->bits);
239
240 /* set caps */
241 gst_audio_info_init (&info);
242 gst_audio_info_set_format (&info,
243 GST_AUDIO_FORMAT_S16,
244 dec->header->rate,
245 dec->header->nb_channels, chan_pos[dec->header->nb_channels - 1]);
246
247 if (!gst_audio_decoder_set_output_format (GST_AUDIO_DECODER (dec), &info))
248 goto nego_failed;
249
250 return GST_FLOW_OK;
251
252 /* ERRORS */
253 no_header:
254 {
255 GST_ELEMENT_ERROR (GST_ELEMENT (dec), STREAM, DECODE,
256 (NULL), ("couldn't read header"));
257 return GST_FLOW_ERROR;
258 }
259 mode_too_old:
260 {
261 GST_ELEMENT_ERROR (GST_ELEMENT (dec), STREAM, DECODE,
262 (NULL),
263 ("Mode number %d does not (yet/any longer) exist in this version",
264 dec->header->mode));
265 return GST_FLOW_ERROR;
266 }
267 init_failed:
268 {
269 GST_ELEMENT_ERROR (GST_ELEMENT (dec), STREAM, DECODE,
270 (NULL), ("couldn't initialize decoder"));
271 return GST_FLOW_ERROR;
272 }
273 nego_failed:
274 {
275 GST_ELEMENT_ERROR (GST_ELEMENT (dec), STREAM, DECODE,
276 (NULL), ("couldn't negotiate format"));
277 return GST_FLOW_NOT_NEGOTIATED;
278 }
279 }
280
281 static GstFlowReturn
gst_speex_dec_parse_comments(GstSpeexDec * dec,GstBuffer * buf)282 gst_speex_dec_parse_comments (GstSpeexDec * dec, GstBuffer * buf)
283 {
284 GstTagList *list;
285 gchar *ver, *encoder = NULL;
286
287 list = gst_tag_list_from_vorbiscomment_buffer (buf, NULL, 0, &encoder);
288
289 if (!list) {
290 GST_WARNING_OBJECT (dec, "couldn't decode comments");
291 list = gst_tag_list_new_empty ();
292 }
293
294 if (encoder) {
295 gst_tag_list_add (list, GST_TAG_MERGE_REPLACE,
296 GST_TAG_ENCODER, encoder, NULL);
297 }
298
299 gst_tag_list_add (list, GST_TAG_MERGE_REPLACE,
300 GST_TAG_AUDIO_CODEC, "Speex", NULL);
301
302 ver = g_strndup (dec->header->speex_version, SPEEX_HEADER_VERSION_LENGTH);
303 g_strstrip (ver);
304
305 if (ver != NULL && *ver != '\0') {
306 gst_tag_list_add (list, GST_TAG_MERGE_REPLACE,
307 GST_TAG_ENCODER_VERSION, ver, NULL);
308 }
309
310 if (dec->header->bitrate > 0) {
311 gst_tag_list_add (list, GST_TAG_MERGE_REPLACE,
312 GST_TAG_BITRATE, (guint) dec->header->bitrate, NULL);
313 }
314
315 GST_INFO_OBJECT (dec, "tags: %" GST_PTR_FORMAT, list);
316
317 gst_audio_decoder_merge_tags (GST_AUDIO_DECODER (dec), list,
318 GST_TAG_MERGE_REPLACE);
319 gst_tag_list_unref (list);
320
321 g_free (encoder);
322 g_free (ver);
323
324 return GST_FLOW_OK;
325 }
326
327 static gboolean
gst_speex_dec_set_format(GstAudioDecoder * bdec,GstCaps * caps)328 gst_speex_dec_set_format (GstAudioDecoder * bdec, GstCaps * caps)
329 {
330 GstSpeexDec *dec = GST_SPEEX_DEC (bdec);
331 gboolean ret = TRUE;
332 GstStructure *s;
333 const GValue *streamheader;
334
335 s = gst_caps_get_structure (caps, 0);
336 if ((streamheader = gst_structure_get_value (s, "streamheader")) &&
337 G_VALUE_HOLDS (streamheader, GST_TYPE_ARRAY) &&
338 gst_value_array_get_size (streamheader) >= 2) {
339 const GValue *header, *vorbiscomment;
340 GstBuffer *buf;
341 GstFlowReturn res = GST_FLOW_OK;
342
343 header = gst_value_array_get_value (streamheader, 0);
344 if (header && G_VALUE_HOLDS (header, GST_TYPE_BUFFER)) {
345 buf = gst_value_get_buffer (header);
346 res = gst_speex_dec_parse_header (dec, buf);
347 if (res != GST_FLOW_OK)
348 goto done;
349 gst_buffer_replace (&dec->streamheader, buf);
350 }
351
352 vorbiscomment = gst_value_array_get_value (streamheader, 1);
353 if (vorbiscomment && G_VALUE_HOLDS (vorbiscomment, GST_TYPE_BUFFER)) {
354 buf = gst_value_get_buffer (vorbiscomment);
355 res = gst_speex_dec_parse_comments (dec, buf);
356 if (res != GST_FLOW_OK)
357 goto done;
358 gst_buffer_replace (&dec->vorbiscomment, buf);
359 }
360 }
361
362 done:
363 return ret;
364 }
365
366 static GstFlowReturn
gst_speex_dec_parse_data(GstSpeexDec * dec,GstBuffer * buf)367 gst_speex_dec_parse_data (GstSpeexDec * dec, GstBuffer * buf)
368 {
369 GstFlowReturn res = GST_FLOW_OK;
370 gint i, fpp;
371 SpeexBits *bits;
372 GstMapInfo map;
373
374 if (!dec->frame_duration)
375 goto not_negotiated;
376
377 if (G_LIKELY (gst_buffer_get_size (buf))) {
378 /* send data to the bitstream */
379 gst_buffer_map (buf, &map, GST_MAP_READ);
380 speex_bits_read_from (&dec->bits, (gchar *) map.data, map.size);
381 gst_buffer_unmap (buf, &map);
382
383 fpp = dec->header->frames_per_packet;
384 bits = &dec->bits;
385
386 GST_DEBUG_OBJECT (dec, "received buffer of size %" G_GSIZE_FORMAT
387 ", fpp %d, %d bits", map.size, fpp, speex_bits_remaining (bits));
388 } else {
389 /* FIXME ? actually consider how much concealment is needed */
390 /* concealment data, pass NULL as the bits parameters */
391 GST_DEBUG_OBJECT (dec, "creating concealment data");
392 fpp = dec->header->frames_per_packet;
393 bits = NULL;
394 }
395
396 /* now decode each frame, catering for unknown number of them (e.g. rtp) */
397 for (i = 0; i < fpp; i++) {
398 GstBuffer *outbuf;
399 gboolean corrupted = FALSE;
400 gint ret;
401
402 GST_LOG_OBJECT (dec, "decoding frame %d/%d, %d bits remaining", i, fpp,
403 bits ? speex_bits_remaining (bits) : -1);
404 #if 0
405 res =
406 gst_pad_alloc_buffer_and_set_caps (GST_AUDIO_DECODER_SRC_PAD (dec),
407 GST_BUFFER_OFFSET_NONE, dec->frame_size * dec->header->nb_channels * 2,
408 GST_PAD_CAPS (GST_AUDIO_DECODER_SRC_PAD (dec)), &outbuf);
409
410 if (res != GST_FLOW_OK) {
411 GST_DEBUG_OBJECT (dec, "buf alloc flow: %s", gst_flow_get_name (res));
412 return res;
413 }
414 #endif
415 /* FIXME, we can use a bufferpool because we have fixed size buffers. We
416 * could also use an allocator */
417 outbuf =
418 gst_buffer_new_allocate (NULL,
419 dec->frame_size * dec->header->nb_channels * 2, NULL);
420
421 gst_buffer_map (outbuf, &map, GST_MAP_WRITE);
422 ret = speex_decode_int (dec->state, bits, (spx_int16_t *) map.data);
423
424 if (ret == -1) {
425 /* uh? end of stream */
426 GST_WARNING_OBJECT (dec, "Unexpected end of stream found");
427 corrupted = TRUE;
428 } else if (ret == -2) {
429 GST_WARNING_OBJECT (dec, "Decoding error: corrupted stream?");
430 corrupted = TRUE;
431 }
432
433 if (bits && speex_bits_remaining (bits) < 0) {
434 GST_WARNING_OBJECT (dec, "Decoding overflow: corrupted stream?");
435 corrupted = TRUE;
436 }
437 if (dec->header->nb_channels == 2)
438 speex_decode_stereo_int ((spx_int16_t *) map.data, dec->frame_size,
439 dec->stereo);
440
441 gst_buffer_unmap (outbuf, &map);
442
443 if (!corrupted) {
444 res = gst_audio_decoder_finish_frame (GST_AUDIO_DECODER (dec), outbuf, 1);
445 } else {
446 res = gst_audio_decoder_finish_frame (GST_AUDIO_DECODER (dec), NULL, 1);
447 gst_buffer_unref (outbuf);
448 }
449
450 if (res != GST_FLOW_OK) {
451 GST_DEBUG_OBJECT (dec, "flow: %s", gst_flow_get_name (res));
452 break;
453 }
454 }
455
456 return res;
457
458 /* ERRORS */
459 not_negotiated:
460 {
461 GST_ELEMENT_ERROR (dec, CORE, NEGOTIATION, (NULL),
462 ("decoder not initialized"));
463 return GST_FLOW_NOT_NEGOTIATED;
464 }
465 }
466
467 static gboolean
memcmp_buffers(GstBuffer * buf1,GstBuffer * buf2)468 memcmp_buffers (GstBuffer * buf1, GstBuffer * buf2)
469 {
470 GstMapInfo map;
471 gsize size1, size2;
472 gboolean res;
473
474 size1 = gst_buffer_get_size (buf1);
475 size2 = gst_buffer_get_size (buf2);
476
477 if (size1 != size2)
478 return FALSE;
479
480 gst_buffer_map (buf1, &map, GST_MAP_READ);
481 res = gst_buffer_memcmp (buf2, 0, map.data, map.size) == 0;
482 gst_buffer_unmap (buf1, &map);
483
484 return res;
485 }
486
487 static GstFlowReturn
gst_speex_dec_handle_frame(GstAudioDecoder * bdec,GstBuffer * buf)488 gst_speex_dec_handle_frame (GstAudioDecoder * bdec, GstBuffer * buf)
489 {
490 GstFlowReturn res;
491 GstSpeexDec *dec;
492
493 /* no fancy draining */
494 if (G_UNLIKELY (!buf))
495 return GST_FLOW_OK;
496
497 dec = GST_SPEEX_DEC (bdec);
498
499 /* If we have the streamheader and vorbiscomment from the caps already
500 * ignore them here */
501 if (dec->streamheader && dec->vorbiscomment) {
502 if (memcmp_buffers (dec->streamheader, buf)) {
503 GST_DEBUG_OBJECT (dec, "found streamheader");
504 gst_audio_decoder_finish_frame (bdec, NULL, 1);
505 res = GST_FLOW_OK;
506 } else if (memcmp_buffers (dec->vorbiscomment, buf)) {
507 GST_DEBUG_OBJECT (dec, "found vorbiscomments");
508 gst_audio_decoder_finish_frame (bdec, NULL, 1);
509 res = GST_FLOW_OK;
510 } else {
511 res = gst_speex_dec_parse_data (dec, buf);
512 }
513 } else {
514 /* Otherwise fall back to packet counting and assume that the
515 * first two packets are the headers. */
516 switch (dec->packetno) {
517 case 0:
518 GST_DEBUG_OBJECT (dec, "counted streamheader");
519 res = gst_speex_dec_parse_header (dec, buf);
520 gst_audio_decoder_finish_frame (bdec, NULL, 1);
521 break;
522 case 1:
523 GST_DEBUG_OBJECT (dec, "counted vorbiscomments");
524 res = gst_speex_dec_parse_comments (dec, buf);
525 gst_audio_decoder_finish_frame (bdec, NULL, 1);
526 break;
527 default:
528 {
529 res = gst_speex_dec_parse_data (dec, buf);
530 break;
531 }
532 }
533 }
534
535 dec->packetno++;
536
537 return res;
538 }
539
540 static void
gst_speex_dec_get_property(GObject * object,guint prop_id,GValue * value,GParamSpec * pspec)541 gst_speex_dec_get_property (GObject * object, guint prop_id,
542 GValue * value, GParamSpec * pspec)
543 {
544 GstSpeexDec *speexdec;
545
546 speexdec = GST_SPEEX_DEC (object);
547
548 switch (prop_id) {
549 case ARG_ENH:
550 g_value_set_boolean (value, speexdec->enh);
551 break;
552 default:
553 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
554 break;
555 }
556 }
557
558 static void
gst_speex_dec_set_property(GObject * object,guint prop_id,const GValue * value,GParamSpec * pspec)559 gst_speex_dec_set_property (GObject * object, guint prop_id,
560 const GValue * value, GParamSpec * pspec)
561 {
562 GstSpeexDec *speexdec;
563
564 speexdec = GST_SPEEX_DEC (object);
565
566 switch (prop_id) {
567 case ARG_ENH:
568 speexdec->enh = g_value_get_boolean (value);
569 break;
570 default:
571 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
572 break;
573 }
574 }
575