1
/* $Id: silencedet.c 3664 2011-07-19 03:42:28Z nanang $ */
3
* Copyright (C) 2008-2011 Teluu Inc. (http://www.teluu.com)
4
* Copyright (C) 2003-2008 Benny Prijono <benny@prijono.org>
6
* This program is free software; you can redistribute it and/or modify
7
* it under the terms of the GNU General Public License as published by
8
* the Free Software Foundation; either version 2 of the License, or
9
* (at your option) any later version.
11
* This program 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
14
* GNU General Public License for more details.
16
* You should have received a copy of the GNU General Public License
17
* along with this program; if not, write to the Free Software
18
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20
#include <pjmedia/silencedet.h>
21
#include <pjmedia/alaw_ulaw.h>
22
#include <pjmedia/errno.h>
23
#include <pj/assert.h>
26
#include <pj/string.h>
28
#define THIS_FILE "silencedet.c"
31
# define TRACE_(x) PJ_LOG(5,x)
37
* This enumeration specifies operation mode of silence detector
39
typedef enum pjmedia_silence_det_mode {
43
} pjmedia_silence_det_mode;
48
#define DEF_RECALC_ON_VOICED 4000 /* Time to recalculate threshold
49
in voiced condition, in ms */
50
#define DEF_RECALC_ON_SILENCE 2000 /* Time to recalculate threshold
51
in silence condition, in ms. */
52
#define DEF_BEFORE_SILENCE 400 /* Silence time before really changing
53
state into SILENCE, in ms. */
54
#define DEF_THRESHOLD 1000 /* Default threshold. */
57
* This enumeration specifies the states of the silence detector.
59
enum pjmedia_silence_det_state {
66
* This structure holds the silence detector state.
68
struct pjmedia_silence_det
70
char objname[PJ_MAX_OBJ_NAME]; /**< VAD name. */
72
int mode; /**< VAD mode. */
73
unsigned ptime; /**< Frame time, in msec. */
75
unsigned threshold; /**< Current threshold level. */
76
unsigned sum_level; /**< Total sum of recent level. */
77
unsigned sum_cnt; /**< Number of level summed. */
78
unsigned silence_timer; /**< Silence condition timer. */
79
unsigned voiced_timer; /**< Voiced condition timer. */
81
enum pjmedia_silence_det_state state;/**< Silence detector state. */
82
unsigned recalc_on_voiced; /**< Setting of time to recalc
83
threshold in voiced condition. */
84
unsigned recalc_on_silence; /**< Setting of time to recalc
85
threshold in silence condition.*/
86
unsigned before_silence; /**< Setting of silence time before
87
really changing state into SILENCE,
93
PJ_DEF(pj_status_t) pjmedia_silence_det_create( pj_pool_t *pool,
95
unsigned samples_per_frame,
96
pjmedia_silence_det **p_sd)
98
pjmedia_silence_det *sd;
100
PJ_ASSERT_RETURN(pool && p_sd, PJ_EINVAL);
102
sd = PJ_POOL_ZALLOC_T(pool, pjmedia_silence_det);
104
pj_ansi_snprintf(sd->objname, PJ_MAX_OBJ_NAME, "sd%p", sd);
105
sd->objname[PJ_MAX_OBJ_NAME-1] = '\0';
107
sd->ptime = samples_per_frame * 1000 / clock_rate;
109
/* Default settings */
110
pjmedia_silence_det_set_params(sd, -1, -1, -1);
112
/* Restart in adaptive, silent mode */
113
pjmedia_silence_det_set_adaptive( sd, -1 );
120
PJ_DEF(pj_status_t) pjmedia_silence_det_set_name( pjmedia_silence_det *sd,
123
PJ_ASSERT_RETURN(sd && name, PJ_EINVAL);
125
pj_ansi_snprintf(sd->objname, PJ_MAX_OBJ_NAME, name, sd);
126
sd->objname[PJ_MAX_OBJ_NAME-1] = '\0';
130
PJ_DEF(pj_status_t) pjmedia_silence_det_set_adaptive(pjmedia_silence_det *sd,
133
PJ_ASSERT_RETURN(sd, PJ_EINVAL);
136
threshold = DEF_THRESHOLD;
138
sd->mode = VAD_MODE_ADAPTIVE;
139
sd->threshold = threshold;
144
PJ_DEF(pj_status_t) pjmedia_silence_det_set_fixed( pjmedia_silence_det *sd,
147
PJ_ASSERT_RETURN(sd, PJ_EINVAL);
150
threshold = DEF_THRESHOLD;
152
sd->mode = VAD_MODE_FIXED;
153
sd->threshold = threshold;
158
PJ_DEF(pj_status_t) pjmedia_silence_det_set_params( pjmedia_silence_det *sd,
163
PJ_ASSERT_RETURN(sd, PJ_EINVAL);
165
if (recalc_time1 < 0)
166
recalc_time1 = DEF_RECALC_ON_VOICED;
167
if (recalc_time2 < 0)
168
recalc_time2 = DEF_RECALC_ON_SILENCE;
169
if (before_silence < 0)
170
before_silence = DEF_BEFORE_SILENCE;
172
sd->recalc_on_voiced = recalc_time1;
173
sd->recalc_on_silence = recalc_time2;
174
sd->before_silence = before_silence;
180
PJ_DEF(pj_status_t) pjmedia_silence_det_disable( pjmedia_silence_det *sd )
182
PJ_ASSERT_RETURN(sd, PJ_EINVAL);
184
sd->mode = VAD_MODE_NONE;
190
PJ_DEF(pj_int32_t) pjmedia_calc_avg_signal( const pj_int16_t samples[],
195
const pj_int16_t * pcm = samples;
196
const pj_int16_t * end = samples + count;
208
return (pj_int32_t)(sum / count);
211
PJ_DEF(pj_bool_t) pjmedia_silence_det_apply( pjmedia_silence_det *sd,
214
int avg_recent_level;
216
if (sd->mode == VAD_MODE_NONE)
219
if (sd->mode == VAD_MODE_FIXED)
220
return (level < sd->threshold);
222
/* Calculating recent level */
223
sd->sum_level += level;
225
avg_recent_level = (sd->sum_level / sd->sum_cnt);
227
if (level > sd->threshold ||
228
level >= PJMEDIA_SILENCE_DET_MAX_THRESHOLD)
230
sd->silence_timer = 0;
231
sd->voiced_timer += sd->ptime;
235
if (sd->voiced_timer > sd->recalc_on_voiced) {
236
/* Voiced for long time (>recalc_on_voiced), current
237
* threshold seems to be too low.
239
sd->threshold = (avg_recent_level + sd->threshold) >> 1;
240
TRACE_((THIS_FILE,"Re-adjust threshold (in talk burst)"
241
"to %d", sd->threshold));
243
sd->voiced_timer = 0;
245
/* Reset sig_level */
246
sd->sum_level = avg_recent_level;
252
TRACE_((THIS_FILE,"Starting talk burst (level=%d threshold=%d)",
253
level, sd->threshold));
255
case STATE_START_SILENCE:
256
sd->state = STATE_VOICED;
258
/* Reset sig_level */
259
sd->sum_level = level;
269
sd->voiced_timer = 0;
270
sd->silence_timer += sd->ptime;
274
if (sd->silence_timer >= sd->recalc_on_silence) {
275
sd->threshold = avg_recent_level << 1;
276
TRACE_((THIS_FILE,"Re-adjust threshold (in silence)"
277
"to %d", sd->threshold));
279
sd->silence_timer = 0;
281
/* Reset sig_level */
282
sd->sum_level = avg_recent_level;
288
sd->state = STATE_START_SILENCE;
290
/* Reset sig_level */
291
sd->sum_level = level;
294
case STATE_START_SILENCE:
295
if (sd->silence_timer >= sd->before_silence) {
296
sd->state = STATE_SILENCE;
297
sd->threshold = avg_recent_level << 1;
298
TRACE_((THIS_FILE,"Starting silence (level=%d "
299
"threshold=%d)", level, sd->threshold));
301
/* Reset sig_level */
302
sd->sum_level = avg_recent_level;
313
return (sd->state == STATE_SILENCE);
317
PJ_DEF(pj_bool_t) pjmedia_silence_det_detect( pjmedia_silence_det *sd,
318
const pj_int16_t samples[],
324
/* Calculate average signal level. */
325
level = pjmedia_calc_avg_signal(samples, count);
327
/* Report to caller, if required. */
331
return pjmedia_silence_det_apply(sd, level);