FFmpeg
af_silencedetect.c
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2012 Clément Bœsch <u pkh me>
3  *
4  * This file is part of FFmpeg.
5  *
6  * FFmpeg is free software; you can redistribute it and/or
7  * modify it under the terms of the GNU Lesser General Public
8  * License as published by the Free Software Foundation; either
9  * version 2.1 of the License, or (at your option) any later version.
10  *
11  * FFmpeg 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  * Lesser General Public License for more details.
15  *
16  * You should have received a copy of the GNU Lesser General Public
17  * License along with FFmpeg; if not, write to the Free Software
18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19  */
20 
21 /**
22  * @file
23  * Audio silence detector
24  */
25 
26 #include <float.h> /* DBL_MAX */
27 
28 #include "libavutil/mem.h"
29 #include "libavutil/opt.h"
30 #include "libavutil/timestamp.h"
31 #include "audio.h"
32 #include "avfilter.h"
33 #include "filters.h"
34 
35 typedef struct SilenceDetectContext {
36  const AVClass *class;
37  double noise; ///< noise amplitude ratio
38  int64_t duration; ///< minimum duration of silence until notification
39  int mono; ///< mono mode : check each channel separately (default = check when ALL channels are silent)
40  int channels; ///< number of channels
41  int independent_channels; ///< number of entries in following arrays (always 1 in mono mode)
42  int64_t *nb_null_samples; ///< (array) current number of continuous zero samples
43  int64_t *start; ///< (array) if silence is detected, this value contains the time of the first zero sample (default/unset = INT64_MIN)
44  int64_t frame_end; ///< pts of the end of the current frame (used to compute duration of silence at EOS)
45  int last_sample_rate; ///< last sample rate to check for sample rate changes
46  AVRational time_base; ///< time_base
47 
48  void (*silencedetect)(AVFilterContext *ctx, AVFrame *insamples,
49  int nb_samples, int64_t nb_samples_notify,
50  AVRational time_base);
51 } SilenceDetectContext;
52 
53 #define MAX_DURATION (24*3600*1000000LL)
54 #define OFFSET(x) offsetof(SilenceDetectContext, x)
55 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_AUDIO_PARAM
56 static const AVOption silencedetect_options[] = {
57  { "n", "set noise tolerance", OFFSET(noise), AV_OPT_TYPE_DOUBLE, {.dbl=0.001}, 0, DBL_MAX, FLAGS },
58  { "noise", "set noise tolerance", OFFSET(noise), AV_OPT_TYPE_DOUBLE, {.dbl=0.001}, 0, DBL_MAX, FLAGS },
59  { "d", "set minimum duration in seconds", OFFSET(duration), AV_OPT_TYPE_DURATION, {.i64=2000000}, 0, MAX_DURATION,FLAGS },
60  { "duration", "set minimum duration in seconds", OFFSET(duration), AV_OPT_TYPE_DURATION, {.i64=2000000}, 0, MAX_DURATION,FLAGS },
61  { "mono", "check each channel separately", OFFSET(mono), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, FLAGS },
62  { "m", "check each channel separately", OFFSET(mono), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, FLAGS },
63  { NULL }
64 };
65 
66 AVFILTER_DEFINE_CLASS(silencedetect);
67 
68 static void set_meta(AVFrame *insamples, int channel, const char *key, char *value)
69 {
70  char key2[128];
71 
72  if (channel)
73  snprintf(key2, sizeof(key2), "lavfi.%s.%d", key, channel);
74  else
75  snprintf(key2, sizeof(key2), "lavfi.%s", key);
76  av_dict_set(&insamples->metadata, key2, value, 0);
77 }
78 static av_always_inline void update(AVFilterContext *ctx, AVFrame *insamples,
79  int is_silence, int current_sample, int64_t nb_samples_notify,
80  AVRational time_base)
81 {
82  SilenceDetectContext *s = ctx->priv;
83  int channel = current_sample % s->independent_channels;
84  if (is_silence) {
85  if (s->start[channel] == INT64_MIN) {
86  s->nb_null_samples[channel]++;
87  if (s->nb_null_samples[channel] >= nb_samples_notify) {
88  s->start[channel] = insamples->pts + av_rescale_q(current_sample / s->channels + 1 - nb_samples_notify * s->independent_channels / s->channels,
89  (AVRational){ 1, s->last_sample_rate }, time_base);
90  set_meta(insamples, s->mono ? channel + 1 : 0, "silence_start",
91  av_ts2timestr(s->start[channel], &time_base));
92  if (s->mono)
93  av_log(ctx, AV_LOG_INFO, "channel: %d | ", channel);
94  av_log(ctx, AV_LOG_INFO, "silence_start: %s\n",
95  av_ts2timestr(s->start[channel], &time_base));
96  }
97  }
98  } else {
99  if (s->start[channel] > INT64_MIN) {
100  int64_t end_pts = insamples ? insamples->pts + av_rescale_q(current_sample / s->channels,
101  (AVRational){ 1, s->last_sample_rate }, time_base)
102  : s->frame_end;
103  int64_t duration_ts = end_pts - s->start[channel];
104  if (insamples) {
105  set_meta(insamples, s->mono ? channel + 1 : 0, "silence_end",
106  av_ts2timestr(end_pts, &time_base));
107  set_meta(insamples, s->mono ? channel + 1 : 0, "silence_duration",
108  av_ts2timestr(duration_ts, &time_base));
109  }
110  if (s->mono)
111  av_log(ctx, AV_LOG_INFO, "channel: %d | ", channel);
112  av_log(ctx, AV_LOG_INFO, "silence_end: %s | silence_duration: %s\n",
113  av_ts2timestr(end_pts, &time_base),
114  av_ts2timestr(duration_ts, &time_base));
115  }
116  s->nb_null_samples[channel] = 0;
117  s->start[channel] = INT64_MIN;
118  }
119 }
120 
121 #define SILENCE_DETECT(name, type) \
122 static void silencedetect_##name(AVFilterContext *ctx, AVFrame *insamples, \
123  int nb_samples, int64_t nb_samples_notify, \
124  AVRational time_base) \
125 { \
126  SilenceDetectContext *s = ctx->priv; \
127  const type *p = (const type *)insamples->data[0]; \
128  const type noise = s->noise; \
129  int i; \
130  \
131  for (i = 0; i < nb_samples; i++, p++) \
132  update(ctx, insamples, *p < noise && *p > -noise, i, \
133  nb_samples_notify, time_base); \
134 }
135 
136 #define SILENCE_DETECT_PLANAR(name, type) \
137 static void silencedetect_##name(AVFilterContext *ctx, AVFrame *insamples, \
138  int nb_samples, int64_t nb_samples_notify, \
139  AVRational time_base) \
140 { \
141  SilenceDetectContext *s = ctx->priv; \
142  const int channels = insamples->ch_layout.nb_channels; \
143  const type noise = s->noise; \
144  \
145  nb_samples /= channels; \
146  for (int i = 0; i < nb_samples; i++) { \
147  for (int ch = 0; ch < insamples->ch_layout.nb_channels; ch++) { \
148  const type *p = (const type *)insamples->extended_data[ch]; \
149  update(ctx, insamples, p[i] < noise && p[i] > -noise, \
150  channels * i + ch, \
151  nb_samples_notify, time_base); \
152  } \
153  } \
154 }
155 
156 SILENCE_DETECT(dbl, double)
157 SILENCE_DETECT(flt, float)
158 SILENCE_DETECT(s32, int32_t)
159 SILENCE_DETECT(s16, int16_t)
160 
161 SILENCE_DETECT_PLANAR(dblp, double)
162 SILENCE_DETECT_PLANAR(fltp, float)
163 SILENCE_DETECT_PLANAR(s32p, int32_t)
164 SILENCE_DETECT_PLANAR(s16p, int16_t)
165 
166 static int config_input(AVFilterLink *inlink)
167 {
168  AVFilterContext *ctx = inlink->dst;
169  SilenceDetectContext *s = ctx->priv;
170  int c;
171 
172  s->channels = inlink->ch_layout.nb_channels;
173  s->duration = av_rescale(s->duration, inlink->sample_rate, AV_TIME_BASE);
174  s->independent_channels = s->mono ? s->channels : 1;
175  s->nb_null_samples = av_calloc(s->independent_channels,
176  sizeof(*s->nb_null_samples));
177  if (!s->nb_null_samples)
178  return AVERROR(ENOMEM);
179  s->start = av_malloc_array(sizeof(*s->start), s->independent_channels);
180  if (!s->start)
181  return AVERROR(ENOMEM);
182  for (c = 0; c < s->independent_channels; c++)
183  s->start[c] = INT64_MIN;
184 
185  switch (inlink->format) {
186  case AV_SAMPLE_FMT_DBL: s->silencedetect = silencedetect_dbl; break;
187  case AV_SAMPLE_FMT_FLT: s->silencedetect = silencedetect_flt; break;
188  case AV_SAMPLE_FMT_S32:
189  s->noise *= INT32_MAX;
190  s->silencedetect = silencedetect_s32;
191  break;
192  case AV_SAMPLE_FMT_S16:
193  s->noise *= INT16_MAX;
194  s->silencedetect = silencedetect_s16;
195  break;
196  case AV_SAMPLE_FMT_DBLP: s->silencedetect = silencedetect_dblp; break;
197  case AV_SAMPLE_FMT_FLTP: s->silencedetect = silencedetect_fltp; break;
198  case AV_SAMPLE_FMT_S32P:
199  s->noise *= INT32_MAX;
200  s->silencedetect = silencedetect_s32p;
201  break;
202  case AV_SAMPLE_FMT_S16P:
203  s->noise *= INT16_MAX;
204  s->silencedetect = silencedetect_s16p;
205  break;
206  default:
207  return AVERROR_BUG;
208  }
209 
210  return 0;
211 }
212 
213 static int filter_frame(AVFilterLink *inlink, AVFrame *insamples)
214 {
215  AVFilterContext *ctx = inlink->dst;
216  SilenceDetectContext *s = ctx->priv;
217  const int nb_channels = inlink->ch_layout.nb_channels;
218  const int srate = inlink->sample_rate;
219  const int nb_samples = insamples->nb_samples * nb_channels;
220  const int64_t nb_samples_notify = s->duration * (s->mono ? 1 : nb_channels);
221  int c;
222 
223  // scale number of null samples to the new sample rate
224  if (s->last_sample_rate && s->last_sample_rate != srate)
225  for (c = 0; c < s->independent_channels; c++) {
226  s->nb_null_samples[c] = srate * s->nb_null_samples[c] / s->last_sample_rate;
227  }
228  s->last_sample_rate = srate;
229  s->time_base = inlink->time_base;
230  s->frame_end = insamples->pts + av_rescale_q(insamples->nb_samples,
231  (AVRational){ 1, s->last_sample_rate }, inlink->time_base);
232 
233  s->silencedetect(ctx, insamples, nb_samples, nb_samples_notify,
234  inlink->time_base);
235 
236  return ff_filter_frame(inlink->dst->outputs[0], insamples);
237 }
238 
239 static av_cold void uninit(AVFilterContext *ctx)
240 {
241  SilenceDetectContext *s = ctx->priv;
242  int c;
243 
244  for (c = 0; c < s->independent_channels; c++)
245  if (s->start[c] > INT64_MIN)
246  update(ctx, NULL, 0, c, 0, s->time_base);
247  av_freep(&s->nb_null_samples);
248  av_freep(&s->start);
249 }
250 
251 static const AVFilterPad silencedetect_inputs[] = {
252  {
253  .name = "default",
254  .type = AVMEDIA_TYPE_AUDIO,
255  .config_props = config_input,
256  .filter_frame = filter_frame,
257  },
258 };
259 
260 const FFFilter ff_af_silencedetect = {
261  .p.name = "silencedetect",
262  .p.description = NULL_IF_CONFIG_SMALL("Detect silence."),
263  .p.priv_class = &silencedetect_class,
264  .p.flags = AVFILTER_FLAG_METADATA_ONLY,
265  .priv_size = sizeof(SilenceDetectContext),
266  .uninit = uninit,
267  FILTER_INPUTS(silencedetect_inputs),
268  FILTER_OUTPUTS(ff_audio_default_filterpad),
269  FILTER_SAMPLEFMTS(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_DBLP,
270  AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_FLTP,
271  AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S32P,
272  AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S16P),
273 };
AV_SAMPLE_FMT_FLTP
@ AV_SAMPLE_FMT_FLTP
float, planar
Definition: samplefmt.h:66
AVERROR
Filter the word “frame” indicates either a video frame or a group of audio as stored in an AVFrame structure Format for each input and each output the list of supported formats For video that means pixel format For audio that means channel sample they are references to shared objects When the negotiation mechanism computes the intersection of the formats supported at each end of a all references to both lists are replaced with a reference to the intersection And when a single format is eventually chosen for a link amongst the remaining all references to the list are updated That means that if a filter requires that its input and output have the same format amongst a supported all it has to do is use a reference to the same list of formats query_formats can leave some formats unset and return AVERROR(EAGAIN) to cause the negotiation mechanism toagain later. That can be used by filters with complex requirements to use the format negotiated on one link to set the formats supported on another. Frame references ownership and permissions
opt.h
SILENCE_DETECT
#define SILENCE_DETECT(name, type)
Definition: af_silencedetect.c:121
ff_filter_frame
int ff_filter_frame(AVFilterLink *link, AVFrame *frame)
Send a frame of data to the next filter.
Definition: avfilter.c:1062
SilenceDetectContext::noise
double noise
noise amplitude ratio
Definition: af_silencedetect.c:37
int64_t
long long int64_t
Definition: coverity.c:34
inlink
The exact code depends on how similar the blocks are and how related they are to the and needs to apply these operations to the correct inlink or outlink if there are several Macros are available to factor that when no extra processing is inlink
Definition: filter_design.txt:212
filter_frame
static int filter_frame(AVFilterLink *inlink, AVFrame *insamples)
Definition: af_silencedetect.c:213
FILTER_INPUTS
#define FILTER_INPUTS(array)
Definition: filters.h:263
AVFrame
This structure describes decoded (raw) audio or video data.
Definition: frame.h:427
AVFrame::pts
int64_t pts
Presentation timestamp in time_base units (time when frame should be shown to user).
Definition: frame.h:529
AVOption
AVOption.
Definition: opt.h:429
SilenceDetectContext::frame_end
int64_t frame_end
pts of the end of the current frame (used to compute duration of silence at EOS)
Definition: af_silencedetect.c:44
AV_SAMPLE_FMT_S32P
@ AV_SAMPLE_FMT_S32P
signed 32 bits, planar
Definition: samplefmt.h:65
silencedetect_inputs
static const AVFilterPad silencedetect_inputs[]
Definition: af_silencedetect.c:251
AV_OPT_TYPE_DURATION
@ AV_OPT_TYPE_DURATION
Underlying C type is int64_t.
Definition: opt.h:319
float.h
AVFilter::name
const char * name
Filter name.
Definition: avfilter.h:215
uninit
static av_cold void uninit(AVFilterContext *ctx)
Definition: af_silencedetect.c:239
ff_af_silencedetect
const FFFilter ff_af_silencedetect
Definition: af_silencedetect.c:260
noise
static int noise(AVBSFContext *ctx, AVPacket *pkt)
Definition: noise.c:127
AVFilterPad
A filter pad used for either input or output.
Definition: filters.h:39
av_cold
#define av_cold
Definition: attributes.h:90
FILTER_SAMPLEFMTS
#define FILTER_SAMPLEFMTS(...)
Definition: filters.h:251
FFFilter
Definition: filters.h:266
duration
int64_t duration
Definition: movenc.c:65
s
#define s(width, name)
Definition: cbs_vp9.c:198
AV_OPT_TYPE_DOUBLE
@ AV_OPT_TYPE_DOUBLE
Underlying C type is double.
Definition: opt.h:267
AVMEDIA_TYPE_AUDIO
@ AVMEDIA_TYPE_AUDIO
Definition: avutil.h:201
filters.h
ctx
AVFormatContext * ctx
Definition: movenc.c:49
av_rescale_q
int64_t av_rescale_q(int64_t a, AVRational bq, AVRational cq)
Rescale a 64-bit integer by 2 rational numbers.
Definition: mathematics.c:142
SilenceDetectContext::mono
int mono
mono mode : check each channel separately (default = check when ALL channels are silent)
Definition: af_silencedetect.c:39
MAX_DURATION
#define MAX_DURATION
Definition: af_silencedetect.c:53
key
const char * key
Definition: hwcontext_opencl.c:189
FILTER_OUTPUTS
#define FILTER_OUTPUTS(array)
Definition: filters.h:264
SilenceDetectContext::channels
int channels
number of channels
Definition: af_silencedetect.c:40
AVClass
Describe the class of an AVClass context structure.
Definition: log.h:76
SilenceDetectContext::start
int64_t * start
(array) if silence is detected, this value contains the time of the first zero sample (default/unset ...
Definition: af_silencedetect.c:43
NULL
#define NULL
Definition: coverity.c:32
AVRational
Rational number (pair of numerator and denominator).
Definition: rational.h:58
ff_audio_default_filterpad
const AVFilterPad ff_audio_default_filterpad[1]
An AVFilterPad array whose only entry has name "default" and is of type AVMEDIA_TYPE_AUDIO.
Definition: audio.c:34
SILENCE_DETECT_PLANAR
#define SILENCE_DETECT_PLANAR(name, type)
Definition: af_silencedetect.c:136
c
Undefined Behavior In the C some operations are like signed integer dereferencing freed accessing outside allocated Undefined Behavior must not occur in a C it is not safe even if the output of undefined operations is unused The unsafety may seem nit picking but Optimizing compilers have in fact optimized code on the assumption that no undefined Behavior occurs Optimizing code based on wrong assumptions can and has in some cases lead to effects beyond the output of computations The signed integer overflow problem in speed critical code Code which is highly optimized and works with signed integers sometimes has the problem that often the output of the computation does not c
Definition: undefined.txt:32
SilenceDetectContext::time_base
AVRational time_base
time_base
Definition: af_silencedetect.c:46
av_ts2timestr
#define av_ts2timestr(ts, tb)
Convenience macro, the return value should be used only directly in function arguments but never stan...
Definition: timestamp.h:83
NULL_IF_CONFIG_SMALL
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification.
Definition: internal.h:94
OFFSET
#define OFFSET(x)
Definition: af_silencedetect.c:54
set_meta
static void set_meta(AVFrame *insamples, int channel, const char *key, char *value)
Definition: af_silencedetect.c:68
AV_SAMPLE_FMT_S16P
@ AV_SAMPLE_FMT_S16P
signed 16 bits, planar
Definition: samplefmt.h:64
AV_LOG_INFO
#define AV_LOG_INFO
Standard information.
Definition: log.h:221
AVFrame::nb_samples
int nb_samples
number of audio samples (per channel) described by this frame
Definition: frame.h:507
AV_TIME_BASE
#define AV_TIME_BASE
Internal time base represented as integer.
Definition: avutil.h:253
av_malloc_array
#define av_malloc_array(a, b)
Definition: tableprint_vlc.h:32
FLAGS
#define FLAGS
Definition: af_silencedetect.c:55
av_always_inline
#define av_always_inline
Definition: attributes.h:49
value
it s the only field you need to keep assuming you have a context There is some magic you don t need to care about around this just let it vf default value
Definition: writing_filters.txt:86
SilenceDetectContext
Definition: af_silencedetect.c:35
AV_SAMPLE_FMT_S16
@ AV_SAMPLE_FMT_S16
signed 16 bits
Definition: samplefmt.h:58
AVFilterPad::name
const char * name
Pad name.
Definition: filters.h:45
silencedetect_options
static const AVOption silencedetect_options[]
Definition: af_silencedetect.c:56
av_rescale
int64_t av_rescale(int64_t a, int64_t b, int64_t c)
Rescale a 64-bit integer with rounding to nearest.
Definition: mathematics.c:129
av_calloc
void * av_calloc(size_t nmemb, size_t size)
Definition: mem.c:264
SilenceDetectContext::duration
int64_t duration
minimum duration of silence until notification
Definition: af_silencedetect.c:38
SilenceDetectContext::independent_channels
int independent_channels
number of entries in following arrays (always 1 in mono mode)
Definition: af_silencedetect.c:41
AVFILTER_DEFINE_CLASS
AVFILTER_DEFINE_CLASS(silencedetect)
avfilter.h
AVFrame::metadata
AVDictionary * metadata
metadata.
Definition: frame.h:705
AVFILTER_FLAG_METADATA_ONLY
#define AVFILTER_FLAG_METADATA_ONLY
The filter is a "metadata" filter - it does not modify the frame data in any way.
Definition: avfilter.h:178
update
static av_always_inline void update(AVFilterContext *ctx, AVFrame *insamples, int is_silence, int current_sample, int64_t nb_samples_notify, AVRational time_base)
Definition: af_silencedetect.c:78
AV_SAMPLE_FMT_DBLP
@ AV_SAMPLE_FMT_DBLP
double, planar
Definition: samplefmt.h:67
config_input
static int config_input(AVFilterLink *inlink)
Definition: af_silencedetect.c:166
AVFilterContext
An instance of a filter.
Definition: avfilter.h:269
FFFilter::p
AVFilter p
The public AVFilter.
Definition: filters.h:270
mem.h
audio.h
AV_OPT_TYPE_BOOL
@ AV_OPT_TYPE_BOOL
Underlying C type is int.
Definition: opt.h:327
av_freep
#define av_freep(p)
Definition: tableprint_vlc.h:35
av_dict_set
int av_dict_set(AVDictionary **pm, const char *key, const char *value, int flags)
Set the given entry in *pm, overwriting an existing entry.
Definition: dict.c:86
SilenceDetectContext::nb_null_samples
int64_t * nb_null_samples
(array) current number of continuous zero samples
Definition: af_silencedetect.c:42
int32_t
int32_t
Definition: audioconvert.c:56
timestamp.h
AVERROR_BUG
#define AVERROR_BUG
Internal bug, also see AVERROR_BUG2.
Definition: error.h:52
av_log
#define av_log(a,...)
Definition: tableprint_vlc.h:27
SilenceDetectContext::last_sample_rate
int last_sample_rate
last sample rate to check for sample rate changes
Definition: af_silencedetect.c:45
AV_SAMPLE_FMT_DBL
@ AV_SAMPLE_FMT_DBL
double
Definition: samplefmt.h:61
AV_SAMPLE_FMT_S32
@ AV_SAMPLE_FMT_S32
signed 32 bits
Definition: samplefmt.h:59
snprintf
#define snprintf
Definition: snprintf.h:34
AV_SAMPLE_FMT_FLT
@ AV_SAMPLE_FMT_FLT
float
Definition: samplefmt.h:60
channel
channel
Definition: ebur128.h:39
SilenceDetectContext::silencedetect
void(* silencedetect)(AVFilterContext *ctx, AVFrame *insamples, int nb_samples, int64_t nb_samples_notify, AVRational time_base)
Definition: af_silencedetect.c:48
Generated on Sun Aug 31 2025 19:22:46 for FFmpeg by   doxygen 1.8.17