2 * Copyright (C) 2004 Wim Taymans <wim@fluendo.com>
3 * Copyright (C) 2006 Tim-Philipp Müller <tim centricular net>
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.
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.
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., 59 Temple Place - Suite 330,
18 * Boston, MA 02111-1307, USA.
22 * SECTION:element-speexdec
23 * @see_also: speexenc, oggdemux
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
31 * <title>Example pipelines</title>
33 * gst-launch -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.
38 * Last reviewed on 2006-04-05 (0.10.2)
45 #include "gstspeexdec.h"
48 #include <gst/tag/tag.h>
49 #include <gst/audio/audio.h>
51 GST_DEBUG_CATEGORY_STATIC (speexdec_debug);
52 #define GST_CAT_DEFAULT speexdec_debug
54 #define DEFAULT_ENH TRUE
62 #define FORMAT_STR GST_AUDIO_NE(S16)
64 static GstStaticPadTemplate speex_dec_src_factory =
65 GST_STATIC_PAD_TEMPLATE ("src",
68 GST_STATIC_CAPS ("audio/x-raw, "
69 "format = (string) " FORMAT_STR ", "
70 "rate = (int) [ 6000, 48000 ], " "channels = (int) [ 1, 2 ]")
73 static GstStaticPadTemplate speex_dec_sink_factory =
74 GST_STATIC_PAD_TEMPLATE ("sink",
77 GST_STATIC_CAPS ("audio/x-speex")
80 #define gst_speex_dec_parent_class parent_class
81 G_DEFINE_TYPE (GstSpeexDec, gst_speex_dec, GST_TYPE_AUDIO_DECODER);
83 static gboolean gst_speex_dec_start (GstAudioDecoder * dec);
84 static gboolean gst_speex_dec_stop (GstAudioDecoder * dec);
85 static gboolean gst_speex_dec_set_format (GstAudioDecoder * bdec,
87 static GstFlowReturn gst_speex_dec_handle_frame (GstAudioDecoder * dec,
90 static void gst_speex_dec_get_property (GObject * object, guint prop_id,
91 GValue * value, GParamSpec * pspec);
92 static void gst_speex_dec_set_property (GObject * object, guint prop_id,
93 const GValue * value, GParamSpec * pspec);
96 gst_speex_dec_class_init (GstSpeexDecClass * klass)
98 GObjectClass *gobject_class;
99 GstElementClass *gstelement_class;
100 GstAudioDecoderClass *base_class;
102 gobject_class = (GObjectClass *) klass;
103 gstelement_class = (GstElementClass *) klass;
104 base_class = (GstAudioDecoderClass *) klass;
106 gobject_class->set_property = gst_speex_dec_set_property;
107 gobject_class->get_property = gst_speex_dec_get_property;
109 base_class->start = GST_DEBUG_FUNCPTR (gst_speex_dec_start);
110 base_class->stop = GST_DEBUG_FUNCPTR (gst_speex_dec_stop);
111 base_class->set_format = GST_DEBUG_FUNCPTR (gst_speex_dec_set_format);
112 base_class->handle_frame = GST_DEBUG_FUNCPTR (gst_speex_dec_handle_frame);
114 g_object_class_install_property (G_OBJECT_CLASS (klass), ARG_ENH,
115 g_param_spec_boolean ("enh", "Enh", "Enable perceptual enhancement",
116 DEFAULT_ENH, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
118 gst_element_class_add_pad_template (gstelement_class,
119 gst_static_pad_template_get (&speex_dec_src_factory));
120 gst_element_class_add_pad_template (gstelement_class,
121 gst_static_pad_template_get (&speex_dec_sink_factory));
122 gst_element_class_set_details_simple (gstelement_class, "Speex audio decoder",
123 "Codec/Decoder/Audio",
124 "decode speex streams to audio", "Wim Taymans <wim@fluendo.com>");
126 GST_DEBUG_CATEGORY_INIT (speexdec_debug, "speexdec", 0,
127 "speex decoding element");
131 gst_speex_dec_reset (GstSpeexDec * dec)
135 dec->frame_duration = 0;
139 speex_bits_destroy (&dec->bits);
141 gst_buffer_replace (&dec->streamheader, NULL);
142 gst_buffer_replace (&dec->vorbiscomment, NULL);
145 speex_stereo_state_destroy (dec->stereo);
150 speex_decoder_destroy (dec->state);
156 gst_speex_dec_init (GstSpeexDec * dec)
158 dec->enh = DEFAULT_ENH;
160 gst_speex_dec_reset (dec);
164 gst_speex_dec_start (GstAudioDecoder * dec)
166 GstSpeexDec *sd = GST_SPEEX_DEC (dec);
168 GST_DEBUG_OBJECT (dec, "start");
169 gst_speex_dec_reset (sd);
171 /* we know about concealment */
172 gst_audio_decoder_set_plc_aware (dec, TRUE);
178 gst_speex_dec_stop (GstAudioDecoder * dec)
180 GstSpeexDec *sd = GST_SPEEX_DEC (dec);
182 GST_DEBUG_OBJECT (dec, "stop");
183 gst_speex_dec_reset (sd);
189 gst_speex_dec_parse_header (GstSpeexDec * dec, GstBuffer * buf)
196 data = gst_buffer_map (buf, &size, NULL, GST_MAP_READ);
197 dec->header = speex_packet_to_header (data, size);
198 gst_buffer_unmap (buf, data, size);
203 if (dec->header->mode >= SPEEX_NB_MODES || dec->header->mode < 0)
206 dec->mode = speex_lib_get_mode (dec->header->mode);
208 /* initialize the decoder */
209 dec->state = speex_decoder_init (dec->mode);
213 speex_decoder_ctl (dec->state, SPEEX_SET_ENH, &dec->enh);
214 speex_decoder_ctl (dec->state, SPEEX_GET_FRAME_SIZE, &dec->frame_size);
216 if (dec->header->nb_channels != 1) {
217 dec->stereo = speex_stereo_state_init ();
218 dec->callback.callback_id = SPEEX_INBAND_STEREO;
219 dec->callback.func = speex_std_stereo_request_handler;
220 dec->callback.data = dec->stereo;
221 speex_decoder_ctl (dec->state, SPEEX_SET_HANDLER, &dec->callback);
224 speex_decoder_ctl (dec->state, SPEEX_SET_SAMPLING_RATE, &dec->header->rate);
226 dec->frame_duration = gst_util_uint64_scale_int (dec->frame_size,
227 GST_SECOND, dec->header->rate);
229 speex_bits_init (&dec->bits);
232 caps = gst_caps_new_simple ("audio/x-raw",
233 "format", G_TYPE_STRING, FORMAT_STR,
234 "rate", G_TYPE_INT, dec->header->rate,
235 "channels", G_TYPE_INT, dec->header->nb_channels, NULL);
237 if (!gst_pad_set_caps (GST_AUDIO_DECODER_SRC_PAD (dec), caps))
240 gst_caps_unref (caps);
246 GST_ELEMENT_ERROR (GST_ELEMENT (dec), STREAM, DECODE,
247 (NULL), ("couldn't read header"));
248 return GST_FLOW_ERROR;
252 GST_ELEMENT_ERROR (GST_ELEMENT (dec), STREAM, DECODE,
254 ("Mode number %d does not (yet/any longer) exist in this version",
256 return GST_FLOW_ERROR;
260 GST_ELEMENT_ERROR (GST_ELEMENT (dec), STREAM, DECODE,
261 (NULL), ("couldn't initialize decoder"));
262 return GST_FLOW_ERROR;
266 GST_ELEMENT_ERROR (GST_ELEMENT (dec), STREAM, DECODE,
267 (NULL), ("couldn't negotiate format"));
268 gst_caps_unref (caps);
269 return GST_FLOW_NOT_NEGOTIATED;
274 gst_speex_dec_parse_comments (GstSpeexDec * dec, GstBuffer * buf)
277 gchar *ver, *encoder = NULL;
279 list = gst_tag_list_from_vorbiscomment_buffer (buf, NULL, 0, &encoder);
282 GST_WARNING_OBJECT (dec, "couldn't decode comments");
283 list = gst_tag_list_new ();
287 gst_tag_list_add (list, GST_TAG_MERGE_REPLACE,
288 GST_TAG_ENCODER, encoder, NULL);
291 gst_tag_list_add (list, GST_TAG_MERGE_REPLACE,
292 GST_TAG_AUDIO_CODEC, "Speex", NULL);
294 ver = g_strndup (dec->header->speex_version, SPEEX_HEADER_VERSION_LENGTH);
297 if (ver != NULL && *ver != '\0') {
298 gst_tag_list_add (list, GST_TAG_MERGE_REPLACE,
299 GST_TAG_ENCODER_VERSION, ver, NULL);
302 if (dec->header->bitrate > 0) {
303 gst_tag_list_add (list, GST_TAG_MERGE_REPLACE,
304 GST_TAG_BITRATE, (guint) dec->header->bitrate, NULL);
307 GST_INFO_OBJECT (dec, "tags: %" GST_PTR_FORMAT, list);
309 gst_element_found_tags_for_pad (GST_ELEMENT (dec),
310 GST_AUDIO_DECODER_SRC_PAD (dec), list);
319 gst_speex_dec_set_format (GstAudioDecoder * bdec, GstCaps * caps)
321 GstSpeexDec *dec = GST_SPEEX_DEC (bdec);
324 const GValue *streamheader;
326 s = gst_caps_get_structure (caps, 0);
327 if ((streamheader = gst_structure_get_value (s, "streamheader")) &&
328 G_VALUE_HOLDS (streamheader, GST_TYPE_ARRAY) &&
329 gst_value_array_get_size (streamheader) >= 2) {
330 const GValue *header, *vorbiscomment;
332 GstFlowReturn res = GST_FLOW_OK;
334 header = gst_value_array_get_value (streamheader, 0);
335 if (header && G_VALUE_HOLDS (header, GST_TYPE_BUFFER)) {
336 buf = gst_value_get_buffer (header);
337 res = gst_speex_dec_parse_header (dec, buf);
338 if (res != GST_FLOW_OK)
340 gst_buffer_replace (&dec->streamheader, buf);
343 vorbiscomment = gst_value_array_get_value (streamheader, 1);
344 if (vorbiscomment && G_VALUE_HOLDS (vorbiscomment, GST_TYPE_BUFFER)) {
345 buf = gst_value_get_buffer (vorbiscomment);
346 res = gst_speex_dec_parse_comments (dec, buf);
347 if (res != GST_FLOW_OK)
349 gst_buffer_replace (&dec->vorbiscomment, buf);
358 gst_speex_dec_parse_data (GstSpeexDec * dec, GstBuffer * buf)
360 GstFlowReturn res = GST_FLOW_OK;
366 if (!dec->frame_duration)
369 if (G_LIKELY (gst_buffer_get_size (buf))) {
370 /* send data to the bitstream */
371 data = gst_buffer_map (buf, &size, NULL, GST_MAP_READ);
372 speex_bits_read_from (&dec->bits, data, size);
373 gst_buffer_unmap (buf, data, size);
375 fpp = dec->header->frames_per_packet;
378 GST_DEBUG_OBJECT (dec, "received buffer of size %u, fpp %d, %d bits",
379 size, fpp, speex_bits_remaining (bits));
381 /* FIXME ? actually consider how much concealment is needed */
382 /* concealment data, pass NULL as the bits parameters */
383 GST_DEBUG_OBJECT (dec, "creating concealment data");
384 fpp = dec->header->frames_per_packet;
388 /* now decode each frame, catering for unknown number of them (e.g. rtp) */
389 for (i = 0; i < fpp; i++) {
394 GST_LOG_OBJECT (dec, "decoding frame %d/%d, %d bits remaining", i, fpp,
395 bits ? speex_bits_remaining (bits) : -1);
398 gst_pad_alloc_buffer_and_set_caps (GST_AUDIO_DECODER_SRC_PAD (dec),
399 GST_BUFFER_OFFSET_NONE, dec->frame_size * dec->header->nb_channels * 2,
400 GST_PAD_CAPS (GST_AUDIO_DECODER_SRC_PAD (dec)), &outbuf);
402 if (res != GST_FLOW_OK) {
403 GST_DEBUG_OBJECT (dec, "buf alloc flow: %s", gst_flow_get_name (res));
407 /* FIXME, we can use a bufferpool because we have fixed size buffers. We
408 * could also use an allocator */
410 gst_buffer_new_allocate (NULL,
411 dec->frame_size * dec->header->nb_channels * 2, 0);
413 out_data = gst_buffer_map (outbuf, &size, NULL, GST_MAP_WRITE);
414 ret = speex_decode_int (dec->state, bits, out_data);
415 gst_buffer_unmap (outbuf, out_data, size);
418 /* uh? end of stream */
419 if (fpp == 0 && speex_bits_remaining (bits) < 8) {
420 /* if we did not know how many frames to expect, then we get this
421 at the end if there are leftover bits to pad to the next byte */
422 GST_DEBUG_OBJECT (dec, "Discarding leftover bits");
424 GST_WARNING_OBJECT (dec, "Unexpected end of stream found");
426 gst_audio_decoder_finish_frame (GST_AUDIO_DECODER (dec), NULL, 1);
427 gst_buffer_unref (outbuf);
428 } else if (ret == -2) {
429 GST_WARNING_OBJECT (dec, "Decoding error: corrupted stream?");
430 gst_audio_decoder_finish_frame (GST_AUDIO_DECODER (dec), NULL, 1);
431 gst_buffer_unref (outbuf);
434 if (bits && speex_bits_remaining (bits) < 0) {
435 GST_WARNING_OBJECT (dec, "Decoding overflow: corrupted stream?");
436 gst_audio_decoder_finish_frame (GST_AUDIO_DECODER (dec), NULL, 1);
437 gst_buffer_unref (outbuf);
439 if (dec->header->nb_channels == 2)
440 speex_decode_stereo_int (out_data, dec->frame_size, dec->stereo);
442 res = gst_audio_decoder_finish_frame (GST_AUDIO_DECODER (dec), outbuf, 1);
444 if (res != GST_FLOW_OK) {
445 GST_DEBUG_OBJECT (dec, "flow: %s", gst_flow_get_name (res));
455 GST_ELEMENT_ERROR (dec, CORE, NEGOTIATION, (NULL),
456 ("decoder not initialized"));
457 return GST_FLOW_NOT_NEGOTIATED;
462 memcmp_buffers (GstBuffer * buf1, GstBuffer * buf2)
468 size1 = gst_buffer_get_size (buf1);
469 size2 = gst_buffer_get_size (buf2);
474 data1 = gst_buffer_map (buf1, NULL, NULL, GST_MAP_READ);
475 res = gst_buffer_memcmp (buf2, 0, data1, size1) == 0;
476 gst_buffer_unmap (buf1, data1, size1);
482 gst_speex_dec_handle_frame (GstAudioDecoder * bdec, GstBuffer * buf)
487 /* no fancy draining */
488 if (G_UNLIKELY (!buf))
491 dec = GST_SPEEX_DEC (bdec);
493 /* If we have the streamheader and vorbiscomment from the caps already
494 * ignore them here */
495 if (dec->streamheader && dec->vorbiscomment) {
496 if (memcmp_buffers (dec->streamheader, buf)) {
497 GST_DEBUG_OBJECT (dec, "found streamheader");
498 gst_audio_decoder_finish_frame (bdec, NULL, 1);
500 } else if (memcmp_buffers (dec->vorbiscomment, buf)) {
501 GST_DEBUG_OBJECT (dec, "found vorbiscomments");
502 gst_audio_decoder_finish_frame (bdec, NULL, 1);
505 res = gst_speex_dec_parse_data (dec, buf);
508 /* Otherwise fall back to packet counting and assume that the
509 * first two packets are the headers. */
510 switch (dec->packetno) {
512 GST_DEBUG_OBJECT (dec, "counted streamheader");
513 res = gst_speex_dec_parse_header (dec, buf);
514 gst_audio_decoder_finish_frame (bdec, NULL, 1);
517 GST_DEBUG_OBJECT (dec, "counted vorbiscomments");
518 res = gst_speex_dec_parse_comments (dec, buf);
519 gst_audio_decoder_finish_frame (bdec, NULL, 1);
523 res = gst_speex_dec_parse_data (dec, buf);
535 gst_speex_dec_get_property (GObject * object, guint prop_id,
536 GValue * value, GParamSpec * pspec)
538 GstSpeexDec *speexdec;
540 speexdec = GST_SPEEX_DEC (object);
544 g_value_set_boolean (value, speexdec->enh);
547 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
553 gst_speex_dec_set_property (GObject * object, guint prop_id,
554 const GValue * value, GParamSpec * pspec)
556 GstSpeexDec *speexdec;
558 speexdec = GST_SPEEX_DEC (object);
562 speexdec->enh = g_value_get_boolean (value);
565 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);