1 /* GStreamer
2 * Copyright (C) 2011 Tiago Katcipis <tiagokatcipis@gmail.com>
3 * Copyright (C) 2011 Paulo Pizarro <paulo.pizarro@gmail.com>
4 * Copyright (C) 2012-2016 Nicola Murino <nicola.murino@gmail.com>
5 *
6 * This library is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Library General Public
8 * License as published by the Free Software Foundation; either
9 * version 2 of the License, or (at your option) any later version.
10 *
11 * This library is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Library General Public License for more details.
15 *
16 * You should have received a copy of the GNU Library General Public
17 * License along with this library; if not, write to the
18 * Free Software Foundation, Inc., 51 Franklin St, Fifth Floor,
19 * Boston, MA 02110-1301, USA.
20 */
21
22 /**
23 * SECTION:element-removesilence
24 * @title: removesilence
25 *
26 * Removes all silence periods from an audio stream, dropping silence buffers.
27 * If the "silent" property is disabled, removesilence will generate
28 * bus messages named "removesilence".
29 * The message's structure contains one of these fields:
30 *
31 * - #guint64 "silence_detected": the PTS for the first silent buffer after a non silence period.
32 *
33 * - #guint64 "silence_finished": the PTS for the first non silent buffer after a silence period.
34 *
35 * ## Example launch line
36 * |[
37 * gst-launch-1.0 -v -m filesrc location="audiofile" ! decodebin ! removesilence remove=true ! wavenc ! filesink location=without_audio.wav
38 * ]|
39 *
40 */
41
42 #ifdef HAVE_CONFIG_H
43 #include "config.h"
44 #endif
45
46 #include <gst/gst.h>
47 #include <gst/base/gstbasetransform.h>
48 #include <gst/audio/audio.h>
49
50 #include "gstremovesilence.h"
51
52
53 GST_DEBUG_CATEGORY_STATIC (gst_remove_silence_debug);
54 #define GST_CAT_DEFAULT gst_remove_silence_debug
55 #define DEFAULT_VAD_HYSTERESIS 480 /* 60 mseg */
56 #define MINIMUM_SILENCE_BUFFERS_MIN 0
57 #define MINIMUM_SILENCE_BUFFERS_MAX 10000
58 #define MINIMUM_SILENCE_BUFFERS_DEF 0
59 #define MINIMUM_SILENCE_TIME_MIN 0
60 #define MINIMUM_SILENCE_TIME_MAX 10000000000
61 #define MINIMUM_SILENCE_TIME_DEF 0
62 #define DEFAULT_VAD_THRESHOLD -60
63
64 /* Filter signals and args */
65 enum
66 {
67 /* FILL ME */
68 LAST_SIGNAL
69 };
70
71 enum
72 {
73 PROP_0,
74 PROP_REMOVE,
75 PROP_HYSTERESIS,
76 PROP_THRESHOLD,
77 PROP_SQUASH,
78 PROP_SILENT,
79 PROP_MINIMUM_SILENCE_BUFFERS,
80 PROP_MINIMUM_SILENCE_TIME
81 };
82
83
84 static GstStaticPadTemplate sink_template = GST_STATIC_PAD_TEMPLATE ("sink",
85 GST_PAD_SINK,
86 GST_PAD_ALWAYS,
87 GST_STATIC_CAPS ("audio/x-raw, "
88 "format = (string) " GST_AUDIO_NE (S16) ", "
89 "layout = (string) interleaved, "
90 "rate = (int) [ 1, MAX ], " "channels = (int) 1"));
91
92 static GstStaticPadTemplate src_template = GST_STATIC_PAD_TEMPLATE ("src",
93 GST_PAD_SRC,
94 GST_PAD_ALWAYS,
95 GST_STATIC_CAPS ("audio/x-raw, "
96 "format = (string) " GST_AUDIO_NE (S16) ", "
97 "layout = (string) interleaved, "
98 "rate = (int) [ 1, MAX ], " "channels = (int) 1"));
99
100
101 #define DEBUG_INIT(bla) \
102 GST_DEBUG_CATEGORY_INIT (gst_remove_silence_debug, "removesilence", 0, "removesilence element")
103
104 #define gst_remove_silence_parent_class parent_class
105 G_DEFINE_TYPE_WITH_CODE (GstRemoveSilence, gst_remove_silence,
106 GST_TYPE_BASE_TRANSFORM, DEBUG_INIT (0));
107 GST_ELEMENT_REGISTER_DEFINE (removesilence, "removesilence", GST_RANK_NONE,
108 gst_remove_silence_get_type ());
109
110 static void gst_remove_silence_set_property (GObject * object, guint prop_id,
111 const GValue * value, GParamSpec * pspec);
112 static void gst_remove_silence_get_property (GObject * object, guint prop_id,
113 GValue * value, GParamSpec * pspec);
114
115 static gboolean gst_remove_silence_start (GstBaseTransform * trans);
116 static gboolean gst_remove_silence_sink_event (GstBaseTransform * trans,
117 GstEvent * event);
118 static GstFlowReturn gst_remove_silence_transform_ip (GstBaseTransform * base,
119 GstBuffer * buf);
120 static void gst_remove_silence_finalize (GObject * obj);
121
122 /* GObject vmethod implementations */
123
124 /* initialize the removesilence's class */
125 static void
gst_remove_silence_class_init(GstRemoveSilenceClass * klass)126 gst_remove_silence_class_init (GstRemoveSilenceClass * klass)
127 {
128 GObjectClass *gobject_class;
129 GstElementClass *gstelement_class;
130 GstBaseTransformClass *base_transform_class;
131
132 gobject_class = (GObjectClass *) klass;
133 gstelement_class = (GstElementClass *) klass;
134 base_transform_class = GST_BASE_TRANSFORM_CLASS (klass);
135
136 gobject_class->finalize = gst_remove_silence_finalize;
137 gobject_class->set_property = gst_remove_silence_set_property;
138 gobject_class->get_property = gst_remove_silence_get_property;
139
140 g_object_class_install_property (gobject_class, PROP_REMOVE,
141 g_param_spec_boolean ("remove", "Remove",
142 "Set to true to remove silence from the stream, false otherwise",
143 FALSE, G_PARAM_READWRITE));
144
145 g_object_class_install_property (gobject_class, PROP_HYSTERESIS,
146 g_param_spec_uint64 ("hysteresis",
147 "Hysteresis",
148 "Set the hysteresis (on samples) used on the internal VAD",
149 1, G_MAXUINT64, DEFAULT_VAD_HYSTERESIS, G_PARAM_READWRITE));
150
151 g_object_class_install_property (gobject_class, PROP_THRESHOLD,
152 g_param_spec_int ("threshold",
153 "Threshold",
154 "Set the silence threshold used on the internal VAD in dB",
155 -70, 70, DEFAULT_VAD_THRESHOLD, G_PARAM_READWRITE));
156
157 g_object_class_install_property (gobject_class, PROP_SQUASH,
158 g_param_spec_boolean ("squash", "Squash",
159 "Set to true to retimestamp buffers when silence is removed and so avoid timestamp gap",
160 FALSE, G_PARAM_READWRITE));
161
162 g_object_class_install_property (gobject_class, PROP_SILENT,
163 g_param_spec_boolean ("silent", "Silent",
164 "Disable/enable bus message notifications for silence detected/finished",
165 TRUE, G_PARAM_READWRITE));
166
167 g_object_class_install_property (gobject_class, PROP_MINIMUM_SILENCE_BUFFERS,
168 g_param_spec_uint ("minimum-silence-buffers", "Minimum silence buffers",
169 "Define the minimum number of consecutive silence buffers before "
170 "removing silence, 0 means disabled. This will not introduce latency",
171 MINIMUM_SILENCE_BUFFERS_MIN, MINIMUM_SILENCE_BUFFERS_MAX,
172 MINIMUM_SILENCE_BUFFERS_DEF, G_PARAM_READWRITE));
173
174 g_object_class_install_property (gobject_class, PROP_MINIMUM_SILENCE_TIME,
175 g_param_spec_uint64 ("minimum_silence_time",
176 "Minimum silence time",
177 "Define the minimum silence time in nanoseconds before removing "
178 " silence, 0 means disabled. This will not introduce latency",
179 MINIMUM_SILENCE_TIME_MIN, MINIMUM_SILENCE_TIME_MAX,
180 MINIMUM_SILENCE_TIME_DEF, G_PARAM_READWRITE));
181
182 gst_element_class_set_static_metadata (gstelement_class,
183 "RemoveSilence",
184 "Filter/Effect/Audio",
185 "Removes all the silence periods from the audio stream.",
186 "Tiago Katcipis <tiagokatcipis@gmail.com>\n \
187 Paulo Pizarro <paulo.pizarro@gmail.com>\n \
188 Nicola Murino <nicola.murino@gmail.com>");
189
190 gst_element_class_add_static_pad_template (gstelement_class, &src_template);
191 gst_element_class_add_static_pad_template (gstelement_class, &sink_template);
192
193 base_transform_class->start = GST_DEBUG_FUNCPTR (gst_remove_silence_start);
194 base_transform_class->sink_event =
195 GST_DEBUG_FUNCPTR (gst_remove_silence_sink_event);
196 base_transform_class->transform_ip =
197 GST_DEBUG_FUNCPTR (gst_remove_silence_transform_ip);
198 }
199
200 static void
gst_remove_silence_reset(GstRemoveSilence * filter)201 gst_remove_silence_reset (GstRemoveSilence * filter)
202 {
203 filter->ts_offset = 0;
204 filter->silence_detected = FALSE;
205 filter->consecutive_silence_buffers = 0;
206 filter->consecutive_silence_time = 0;
207 }
208
209 /* initialize the new element
210 * instantiate pads and add them to element
211 * set pad callback functions
212 * initialize instance structure
213 */
214 static void
gst_remove_silence_init(GstRemoveSilence * filter)215 gst_remove_silence_init (GstRemoveSilence * filter)
216 {
217 filter->vad = vad_new (DEFAULT_VAD_HYSTERESIS, DEFAULT_VAD_THRESHOLD);
218 filter->remove = FALSE;
219 filter->squash = FALSE;
220 filter->silent = TRUE;
221 filter->minimum_silence_buffers = MINIMUM_SILENCE_BUFFERS_DEF;
222 filter->minimum_silence_time = MINIMUM_SILENCE_TIME_DEF;
223
224 gst_remove_silence_reset (filter);
225
226 if (!filter->vad) {
227 GST_DEBUG ("Error initializing VAD !!");
228 return;
229 }
230 }
231
232 static gboolean
gst_remove_silence_start(GstBaseTransform * trans)233 gst_remove_silence_start (GstBaseTransform * trans)
234 {
235 GstRemoveSilence *filter = GST_REMOVE_SILENCE (trans);
236
237 GST_INFO ("reset filter on start");
238 gst_remove_silence_reset (filter);
239
240 return TRUE;
241 }
242
243 static gboolean
gst_remove_silence_sink_event(GstBaseTransform * trans,GstEvent * event)244 gst_remove_silence_sink_event (GstBaseTransform * trans, GstEvent * event)
245 {
246 GstRemoveSilence *filter = GST_REMOVE_SILENCE (trans);
247
248 if (event->type == GST_EVENT_SEGMENT) {
249 GST_INFO ("reset filter on segment event");
250 gst_remove_silence_reset (filter);
251 }
252
253 return
254 GST_BASE_TRANSFORM_CLASS (gst_remove_silence_parent_class)->sink_event
255 (trans, event);
256 }
257
258 static void
gst_remove_silence_finalize(GObject * obj)259 gst_remove_silence_finalize (GObject * obj)
260 {
261 GstRemoveSilence *filter = GST_REMOVE_SILENCE (obj);
262 GST_DEBUG ("Destroying VAD");
263 vad_destroy (filter->vad);
264 filter->vad = NULL;
265 GST_DEBUG ("VAD Destroyed");
266 G_OBJECT_CLASS (parent_class)->finalize (obj);
267 }
268
269 static void
gst_remove_silence_set_property(GObject * object,guint prop_id,const GValue * value,GParamSpec * pspec)270 gst_remove_silence_set_property (GObject * object, guint prop_id,
271 const GValue * value, GParamSpec * pspec)
272 {
273 GstRemoveSilence *filter = GST_REMOVE_SILENCE (object);
274
275 switch (prop_id) {
276 case PROP_REMOVE:
277 filter->remove = g_value_get_boolean (value);
278 break;
279 case PROP_HYSTERESIS:
280 vad_set_hysteresis (filter->vad, g_value_get_uint64 (value));
281 break;
282 case PROP_THRESHOLD:
283 vad_set_threshold (filter->vad, g_value_get_int (value));
284 break;
285 case PROP_SQUASH:
286 filter->squash = g_value_get_boolean (value);
287 break;
288 case PROP_SILENT:
289 filter->silent = g_value_get_boolean (value);
290 break;
291 case PROP_MINIMUM_SILENCE_BUFFERS:
292 filter->minimum_silence_buffers = g_value_get_uint (value);
293 break;
294 case PROP_MINIMUM_SILENCE_TIME:
295 filter->minimum_silence_time = g_value_get_uint64 (value);
296 break;
297 default:
298 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
299 break;
300 }
301 }
302
303 static void
gst_remove_silence_get_property(GObject * object,guint prop_id,GValue * value,GParamSpec * pspec)304 gst_remove_silence_get_property (GObject * object, guint prop_id,
305 GValue * value, GParamSpec * pspec)
306 {
307 GstRemoveSilence *filter = GST_REMOVE_SILENCE (object);
308
309 switch (prop_id) {
310 case PROP_REMOVE:
311 g_value_set_boolean (value, filter->remove);
312 break;
313 case PROP_HYSTERESIS:
314 g_value_set_uint64 (value, vad_get_hysteresis (filter->vad));
315 break;
316 case PROP_THRESHOLD:
317 g_value_set_int (value, vad_get_threshold_as_db (filter->vad));
318 break;
319 case PROP_SQUASH:
320 g_value_set_boolean (value, filter->squash);
321 break;
322 case PROP_SILENT:
323 g_value_set_boolean (value, filter->silent);
324 break;
325 case PROP_MINIMUM_SILENCE_BUFFERS:
326 g_value_set_uint (value, filter->minimum_silence_buffers);
327 break;
328 case PROP_MINIMUM_SILENCE_TIME:
329 g_value_set_uint64 (value, filter->minimum_silence_time);
330 break;
331 default:
332 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
333 break;
334 }
335 }
336
337 static GstFlowReturn
gst_remove_silence_transform_ip(GstBaseTransform * trans,GstBuffer * inbuf)338 gst_remove_silence_transform_ip (GstBaseTransform * trans, GstBuffer * inbuf)
339 {
340 GstRemoveSilence *filter = NULL;
341 int frame_type;
342 GstMapInfo map;
343 gboolean consecutive_silence_reached;
344
345 filter = GST_REMOVE_SILENCE (trans);
346
347 gst_buffer_map (inbuf, &map, GST_MAP_READ);
348 frame_type =
349 vad_update (filter->vad, (gint16 *) map.data, map.size / sizeof (gint16));
350 gst_buffer_unmap (inbuf, &map);
351
352 if (frame_type == VAD_SILENCE) {
353 GST_DEBUG ("Silence detected");
354 filter->consecutive_silence_buffers++;
355 if (GST_BUFFER_DURATION_IS_VALID (inbuf)) {
356 filter->consecutive_silence_time += inbuf->duration;
357 } else {
358 GST_WARNING
359 ("Invalid buffer duration, consecutive_silence_time update not possible");
360 }
361 if (filter->minimum_silence_buffers == 0
362 && filter->minimum_silence_time == 0) {
363 consecutive_silence_reached = TRUE;
364 } else {
365 consecutive_silence_reached =
366 (filter->minimum_silence_buffers > 0
367 && filter->consecutive_silence_buffers >=
368 filter->minimum_silence_buffers)
369 || (filter->minimum_silence_time > 0
370 && filter->consecutive_silence_time >= filter->minimum_silence_time);
371 }
372 if (!filter->silence_detected && consecutive_silence_reached) {
373 if (!filter->silent) {
374 if (GST_BUFFER_PTS_IS_VALID (inbuf)) {
375 GstStructure *s;
376 GstMessage *m;
377 s = gst_structure_new ("removesilence", "silence_detected",
378 G_TYPE_UINT64, GST_BUFFER_PTS (inbuf) - filter->ts_offset, NULL);
379 m = gst_message_new_element (GST_OBJECT (filter), s);
380 gst_element_post_message (GST_ELEMENT (filter), m);
381 }
382 }
383 filter->silence_detected = TRUE;
384 }
385
386 if (filter->remove && consecutive_silence_reached) {
387 GST_DEBUG ("Removing silence");
388 if (filter->squash) {
389 if (GST_BUFFER_DURATION_IS_VALID (inbuf)) {
390 filter->ts_offset += inbuf->duration;
391 } else {
392 GST_WARNING ("Invalid buffer duration: ts_offset not updated");
393 }
394 }
395 return GST_BASE_TRANSFORM_FLOW_DROPPED;
396 }
397
398 } else {
399 filter->consecutive_silence_buffers = 0;
400 filter->consecutive_silence_time = 0;
401 if (filter->silence_detected) {
402 if (!filter->silent) {
403 if (GST_BUFFER_PTS_IS_VALID (inbuf)) {
404 GstStructure *s;
405 GstMessage *m;
406 s = gst_structure_new ("removesilence", "silence_finished",
407 G_TYPE_UINT64, GST_BUFFER_PTS (inbuf) - filter->ts_offset, NULL);
408 m = gst_message_new_element (GST_OBJECT (filter), s);
409 gst_element_post_message (GST_ELEMENT (filter), m);
410 }
411 }
412 filter->silence_detected = FALSE;
413 }
414 }
415
416 if (filter->squash && filter->ts_offset > 0) {
417 if (GST_BUFFER_PTS_IS_VALID (inbuf)) {
418 inbuf = gst_buffer_make_writable (inbuf);
419 GST_BUFFER_PTS (inbuf) -= filter->ts_offset;
420 } else {
421 GST_WARNING ("Invalid buffer pts, update not possible");
422 }
423 }
424
425 return GST_FLOW_OK;
426 }
427
428 /*Plugin init functions*/
429 static gboolean
plugin_init(GstPlugin * plugin)430 plugin_init (GstPlugin * plugin)
431 {
432 return GST_ELEMENT_REGISTER (removesilence, plugin);
433 }
434
435 GST_PLUGIN_DEFINE (GST_VERSION_MAJOR,
436 GST_VERSION_MINOR,
437 removesilence,
438 "Removes silence from an audio stream",
439 plugin_init, VERSION, "LGPL", GST_PACKAGE_NAME, GST_PACKAGE_ORIGIN);
440