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-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.
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 "layout = (string) interleaved, "
71 "rate = (int) [ 6000, 48000 ], " "channels = (int) [ 1, 2 ]")
74 static GstStaticPadTemplate speex_dec_sink_factory =
75 GST_STATIC_PAD_TEMPLATE ("sink",
78 GST_STATIC_CAPS ("audio/x-speex")
81 #define gst_speex_dec_parent_class parent_class
82 G_DEFINE_TYPE (GstSpeexDec, gst_speex_dec, GST_TYPE_AUDIO_DECODER);
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,
88 static GstFlowReturn gst_speex_dec_handle_frame (GstAudioDecoder * dec,
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);
97 gst_speex_dec_class_init (GstSpeexDecClass * klass)
99 GObjectClass *gobject_class;
100 GstElementClass *gstelement_class;
101 GstAudioDecoderClass *base_class;
103 gobject_class = (GObjectClass *) klass;
104 gstelement_class = (GstElementClass *) klass;
105 base_class = (GstAudioDecoderClass *) klass;
107 gobject_class->set_property = gst_speex_dec_set_property;
108 gobject_class->get_property = gst_speex_dec_get_property;
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);
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));
119 gst_element_class_add_pad_template (gstelement_class,
120 gst_static_pad_template_get (&speex_dec_src_factory));
121 gst_element_class_add_pad_template (gstelement_class,
122 gst_static_pad_template_get (&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>");
127 GST_DEBUG_CATEGORY_INIT (speexdec_debug, "speexdec", 0,
128 "speex decoding element");
132 gst_speex_dec_reset (GstSpeexDec * dec)
136 dec->frame_duration = 0;
140 speex_bits_destroy (&dec->bits);
142 gst_buffer_replace (&dec->streamheader, NULL);
143 gst_buffer_replace (&dec->vorbiscomment, NULL);
146 speex_stereo_state_destroy (dec->stereo);
151 speex_decoder_destroy (dec->state);
157 gst_speex_dec_init (GstSpeexDec * dec)
159 dec->enh = DEFAULT_ENH;
161 gst_speex_dec_reset (dec);
165 gst_speex_dec_start (GstAudioDecoder * dec)
167 GstSpeexDec *sd = GST_SPEEX_DEC (dec);
169 GST_DEBUG_OBJECT (dec, "start");
170 gst_speex_dec_reset (sd);
172 /* we know about concealment */
173 gst_audio_decoder_set_plc_aware (dec, TRUE);
179 gst_speex_dec_stop (GstAudioDecoder * dec)
181 GstSpeexDec *sd = GST_SPEEX_DEC (dec);
183 GST_DEBUG_OBJECT (dec, "stop");
184 gst_speex_dec_reset (sd);
190 gst_speex_dec_parse_header (GstSpeexDec * dec, GstBuffer * buf)
194 static const GstAudioChannelPosition chan_pos[2][2] = {
195 {GST_AUDIO_CHANNEL_POSITION_MONO},
196 {GST_AUDIO_CHANNEL_POSITION_FRONT_LEFT,
197 GST_AUDIO_CHANNEL_POSITION_FRONT_RIGHT}
201 gst_buffer_map (buf, &map, GST_MAP_READ);
202 dec->header = speex_packet_to_header ((gchar *) map.data, map.size);
203 gst_buffer_unmap (buf, &map);
208 if (dec->header->mode >= SPEEX_NB_MODES || dec->header->mode < 0)
211 dec->mode = speex_lib_get_mode (dec->header->mode);
213 /* initialize the decoder */
214 dec->state = speex_decoder_init (dec->mode);
218 speex_decoder_ctl (dec->state, SPEEX_SET_ENH, &dec->enh);
219 speex_decoder_ctl (dec->state, SPEEX_GET_FRAME_SIZE, &dec->frame_size);
221 if (dec->header->nb_channels != 1) {
222 dec->stereo = speex_stereo_state_init ();
223 dec->callback.callback_id = SPEEX_INBAND_STEREO;
224 dec->callback.func = speex_std_stereo_request_handler;
225 dec->callback.data = dec->stereo;
226 speex_decoder_ctl (dec->state, SPEEX_SET_HANDLER, &dec->callback);
229 speex_decoder_ctl (dec->state, SPEEX_SET_SAMPLING_RATE, &dec->header->rate);
231 dec->frame_duration = gst_util_uint64_scale_int (dec->frame_size,
232 GST_SECOND, dec->header->rate);
234 speex_bits_init (&dec->bits);
237 gst_audio_info_init (&info);
238 gst_audio_info_set_format (&info,
239 GST_AUDIO_FORMAT_S16,
241 dec->header->nb_channels, chan_pos[dec->header->nb_channels - 1]);
243 if (!gst_audio_decoder_set_output_format (GST_AUDIO_DECODER (dec), &info))
251 GST_ELEMENT_ERROR (GST_ELEMENT (dec), STREAM, DECODE,
252 (NULL), ("couldn't read header"));
253 return GST_FLOW_ERROR;
257 GST_ELEMENT_ERROR (GST_ELEMENT (dec), STREAM, DECODE,
259 ("Mode number %d does not (yet/any longer) exist in this version",
261 return GST_FLOW_ERROR;
265 GST_ELEMENT_ERROR (GST_ELEMENT (dec), STREAM, DECODE,
266 (NULL), ("couldn't initialize decoder"));
267 return GST_FLOW_ERROR;
271 GST_ELEMENT_ERROR (GST_ELEMENT (dec), STREAM, DECODE,
272 (NULL), ("couldn't negotiate format"));
273 return GST_FLOW_NOT_NEGOTIATED;
278 gst_speex_dec_parse_comments (GstSpeexDec * dec, GstBuffer * buf)
281 gchar *ver, *encoder = NULL;
283 list = gst_tag_list_from_vorbiscomment_buffer (buf, NULL, 0, &encoder);
286 GST_WARNING_OBJECT (dec, "couldn't decode comments");
287 list = gst_tag_list_new_empty ();
291 gst_tag_list_add (list, GST_TAG_MERGE_REPLACE,
292 GST_TAG_ENCODER, encoder, NULL);
295 gst_tag_list_add (list, GST_TAG_MERGE_REPLACE,
296 GST_TAG_AUDIO_CODEC, "Speex", NULL);
298 ver = g_strndup (dec->header->speex_version, SPEEX_HEADER_VERSION_LENGTH);
301 if (ver != NULL && *ver != '\0') {
302 gst_tag_list_add (list, GST_TAG_MERGE_REPLACE,
303 GST_TAG_ENCODER_VERSION, ver, NULL);
306 if (dec->header->bitrate > 0) {
307 gst_tag_list_add (list, GST_TAG_MERGE_REPLACE,
308 GST_TAG_BITRATE, (guint) dec->header->bitrate, NULL);
311 GST_INFO_OBJECT (dec, "tags: %" GST_PTR_FORMAT, list);
313 gst_audio_decoder_merge_tags (GST_AUDIO_DECODER (dec), list,
314 GST_TAG_MERGE_REPLACE);
315 gst_tag_list_unref (list);
324 gst_speex_dec_set_format (GstAudioDecoder * bdec, GstCaps * caps)
326 GstSpeexDec *dec = GST_SPEEX_DEC (bdec);
329 const GValue *streamheader;
331 s = gst_caps_get_structure (caps, 0);
332 if ((streamheader = gst_structure_get_value (s, "streamheader")) &&
333 G_VALUE_HOLDS (streamheader, GST_TYPE_ARRAY) &&
334 gst_value_array_get_size (streamheader) >= 2) {
335 const GValue *header, *vorbiscomment;
337 GstFlowReturn res = GST_FLOW_OK;
339 header = gst_value_array_get_value (streamheader, 0);
340 if (header && G_VALUE_HOLDS (header, GST_TYPE_BUFFER)) {
341 buf = gst_value_get_buffer (header);
342 res = gst_speex_dec_parse_header (dec, buf);
343 if (res != GST_FLOW_OK)
345 gst_buffer_replace (&dec->streamheader, buf);
348 vorbiscomment = gst_value_array_get_value (streamheader, 1);
349 if (vorbiscomment && G_VALUE_HOLDS (vorbiscomment, GST_TYPE_BUFFER)) {
350 buf = gst_value_get_buffer (vorbiscomment);
351 res = gst_speex_dec_parse_comments (dec, buf);
352 if (res != GST_FLOW_OK)
354 gst_buffer_replace (&dec->vorbiscomment, buf);
363 gst_speex_dec_parse_data (GstSpeexDec * dec, GstBuffer * buf)
365 GstFlowReturn res = GST_FLOW_OK;
370 if (!dec->frame_duration)
373 if (G_LIKELY (gst_buffer_get_size (buf))) {
374 /* send data to the bitstream */
375 gst_buffer_map (buf, &map, GST_MAP_READ);
376 speex_bits_read_from (&dec->bits, (gchar *) map.data, map.size);
377 gst_buffer_unmap (buf, &map);
379 fpp = dec->header->frames_per_packet;
382 GST_DEBUG_OBJECT (dec, "received buffer of size %" G_GSIZE_FORMAT
383 ", fpp %d, %d bits", map.size, fpp, speex_bits_remaining (bits));
385 /* FIXME ? actually consider how much concealment is needed */
386 /* concealment data, pass NULL as the bits parameters */
387 GST_DEBUG_OBJECT (dec, "creating concealment data");
388 fpp = dec->header->frames_per_packet;
392 /* now decode each frame, catering for unknown number of them (e.g. rtp) */
393 for (i = 0; i < fpp; i++) {
397 GST_LOG_OBJECT (dec, "decoding frame %d/%d, %d bits remaining", i, fpp,
398 bits ? speex_bits_remaining (bits) : -1);
401 gst_pad_alloc_buffer_and_set_caps (GST_AUDIO_DECODER_SRC_PAD (dec),
402 GST_BUFFER_OFFSET_NONE, dec->frame_size * dec->header->nb_channels * 2,
403 GST_PAD_CAPS (GST_AUDIO_DECODER_SRC_PAD (dec)), &outbuf);
405 if (res != GST_FLOW_OK) {
406 GST_DEBUG_OBJECT (dec, "buf alloc flow: %s", gst_flow_get_name (res));
410 /* FIXME, we can use a bufferpool because we have fixed size buffers. We
411 * could also use an allocator */
413 gst_buffer_new_allocate (NULL,
414 dec->frame_size * dec->header->nb_channels * 2, NULL);
416 gst_buffer_map (outbuf, &map, GST_MAP_WRITE);
417 ret = speex_decode_int (dec->state, bits, (spx_int16_t *) map.data);
420 /* uh? end of stream */
421 if (fpp == 0 && speex_bits_remaining (bits) < 8) {
422 /* if we did not know how many frames to expect, then we get this
423 at the end if there are leftover bits to pad to the next byte */
424 GST_DEBUG_OBJECT (dec, "Discarding leftover bits");
426 GST_WARNING_OBJECT (dec, "Unexpected end of stream found");
428 gst_audio_decoder_finish_frame (GST_AUDIO_DECODER (dec), NULL, 1);
429 gst_buffer_unref (outbuf);
430 } else if (ret == -2) {
431 GST_WARNING_OBJECT (dec, "Decoding error: corrupted stream?");
432 gst_audio_decoder_finish_frame (GST_AUDIO_DECODER (dec), NULL, 1);
433 gst_buffer_unref (outbuf);
436 if (bits && speex_bits_remaining (bits) < 0) {
437 GST_WARNING_OBJECT (dec, "Decoding overflow: corrupted stream?");
438 gst_audio_decoder_finish_frame (GST_AUDIO_DECODER (dec), NULL, 1);
439 gst_buffer_unref (outbuf);
441 if (dec->header->nb_channels == 2)
442 speex_decode_stereo_int ((spx_int16_t *) map.data, dec->frame_size,
445 gst_buffer_unmap (outbuf, &map);
447 res = gst_audio_decoder_finish_frame (GST_AUDIO_DECODER (dec), outbuf, 1);
449 if (res != GST_FLOW_OK) {
450 GST_DEBUG_OBJECT (dec, "flow: %s", gst_flow_get_name (res));
460 GST_ELEMENT_ERROR (dec, CORE, NEGOTIATION, (NULL),
461 ("decoder not initialized"));
462 return GST_FLOW_NOT_NEGOTIATED;
467 memcmp_buffers (GstBuffer * buf1, GstBuffer * buf2)
473 size1 = gst_buffer_get_size (buf1);
474 size2 = gst_buffer_get_size (buf2);
479 gst_buffer_map (buf1, &map, GST_MAP_READ);
480 res = gst_buffer_memcmp (buf2, 0, map.data, map.size) == 0;
481 gst_buffer_unmap (buf1, &map);
487 gst_speex_dec_handle_frame (GstAudioDecoder * bdec, GstBuffer * buf)
492 /* no fancy draining */
493 if (G_UNLIKELY (!buf))
496 dec = GST_SPEEX_DEC (bdec);
498 /* If we have the streamheader and vorbiscomment from the caps already
499 * ignore them here */
500 if (dec->streamheader && dec->vorbiscomment) {
501 if (memcmp_buffers (dec->streamheader, buf)) {
502 GST_DEBUG_OBJECT (dec, "found streamheader");
503 gst_audio_decoder_finish_frame (bdec, NULL, 1);
505 } else if (memcmp_buffers (dec->vorbiscomment, buf)) {
506 GST_DEBUG_OBJECT (dec, "found vorbiscomments");
507 gst_audio_decoder_finish_frame (bdec, NULL, 1);
510 res = gst_speex_dec_parse_data (dec, buf);
513 /* Otherwise fall back to packet counting and assume that the
514 * first two packets are the headers. */
515 switch (dec->packetno) {
517 GST_DEBUG_OBJECT (dec, "counted streamheader");
518 res = gst_speex_dec_parse_header (dec, buf);
519 gst_audio_decoder_finish_frame (bdec, NULL, 1);
522 GST_DEBUG_OBJECT (dec, "counted vorbiscomments");
523 res = gst_speex_dec_parse_comments (dec, buf);
524 gst_audio_decoder_finish_frame (bdec, NULL, 1);
528 res = gst_speex_dec_parse_data (dec, buf);
540 gst_speex_dec_get_property (GObject * object, guint prop_id,
541 GValue * value, GParamSpec * pspec)
543 GstSpeexDec *speexdec;
545 speexdec = GST_SPEEX_DEC (object);
549 g_value_set_boolean (value, speexdec->enh);
552 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
558 gst_speex_dec_set_property (GObject * object, guint prop_id,
559 const GValue * value, GParamSpec * pspec)
561 GstSpeexDec *speexdec;
563 speexdec = GST_SPEEX_DEC (object);
567 speexdec->enh = g_value_get_boolean (value);
570 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);