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/opt.h"
29 #include "libavutil/timestamp.h"
30 #include "audio.h"
31 #include "formats.h"
32 #include "avfilter.h"
33 #include "internal.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)(struct SilenceDetectContext *s, AVFrame *insamples,
49  int nb_samples, int64_t nb_samples_notify,
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 
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 }
79  int is_silence, int current_sample, int64_t nb_samples_notify,
81 {
82  int channel = current_sample % s->independent_channels;
83  if (is_silence) {
84  if (s->start[channel] == INT64_MIN) {
85  s->nb_null_samples[channel]++;
86  if (s->nb_null_samples[channel] >= nb_samples_notify) {
87  s->start[channel] = insamples->pts + av_rescale_q(current_sample / s->channels + 1 - nb_samples_notify * s->independent_channels / s->channels,
88  (AVRational){ 1, s->last_sample_rate }, time_base);
89  set_meta(insamples, s->mono ? channel + 1 : 0, "silence_start",
90  av_ts2timestr(s->start[channel], &time_base));
91  if (s->mono)
92  av_log(s, AV_LOG_INFO, "channel: %d | ", channel);
93  av_log(s, AV_LOG_INFO, "silence_start: %s\n",
94  av_ts2timestr(s->start[channel], &time_base));
95  }
96  }
97  } else {
98  if (s->start[channel] > INT64_MIN) {
99  int64_t end_pts = insamples ? insamples->pts + av_rescale_q(current_sample / s->channels,
100  (AVRational){ 1, s->last_sample_rate }, time_base)
101  : s->frame_end;
102  int64_t duration_ts = end_pts - s->start[channel];
103  if (insamples) {
104  set_meta(insamples, s->mono ? channel + 1 : 0, "silence_end",
105  av_ts2timestr(end_pts, &time_base));
106  set_meta(insamples, s->mono ? channel + 1 : 0, "silence_duration",
107  av_ts2timestr(duration_ts, &time_base));
108  }
109  if (s->mono)
110  av_log(s, AV_LOG_INFO, "channel: %d | ", channel);
111  av_log(s, AV_LOG_INFO, "silence_end: %s | silence_duration: %s\n",
112  av_ts2timestr(end_pts, &time_base),
113  av_ts2timestr(duration_ts, &time_base));
114  }
115  s->nb_null_samples[channel] = 0;
116  s->start[channel] = INT64_MIN;
117  }
118 }
119 
120 #define SILENCE_DETECT(name, type) \
121 static void silencedetect_##name(SilenceDetectContext *s, AVFrame *insamples, \
122  int nb_samples, int64_t nb_samples_notify, \
123  AVRational time_base) \
124 { \
125  const type *p = (const type *)insamples->data[0]; \
126  const type noise = s->noise; \
127  int i; \
128  \
129  for (i = 0; i < nb_samples; i++, p++) \
130  update(s, insamples, *p < noise && *p > -noise, i, \
131  nb_samples_notify, time_base); \
132 }
133 
134 #define SILENCE_DETECT_PLANAR(name, type) \
135 static void silencedetect_##name(SilenceDetectContext *s, AVFrame *insamples, \
136  int nb_samples, int64_t nb_samples_notify, \
137  AVRational time_base) \
138 { \
139  const int channels = insamples->channels; \
140  const type noise = s->noise; \
141  \
142  nb_samples /= channels; \
143  for (int i = 0; i < nb_samples; i++) { \
144  for (int ch = 0; ch < insamples->channels; ch++) { \
145  const type *p = (const type *)insamples->extended_data[ch]; \
146  update(s, insamples, p[i] < noise && p[i] > -noise, \
147  channels * i + ch, \
148  nb_samples_notify, time_base); \
149  } \
150  } \
151 }
152 
153 SILENCE_DETECT(dbl, double)
154 SILENCE_DETECT(flt, float)
156 SILENCE_DETECT(s16, int16_t)
157 
158 SILENCE_DETECT_PLANAR(dblp, double)
159 SILENCE_DETECT_PLANAR(fltp, float)
161 SILENCE_DETECT_PLANAR(s16p, int16_t)
162 
164 {
165  AVFilterContext *ctx = inlink->dst;
166  SilenceDetectContext *s = ctx->priv;
167  int c;
168 
169  s->channels = inlink->channels;
170  s->duration = av_rescale(s->duration, inlink->sample_rate, AV_TIME_BASE);
171  s->independent_channels = s->mono ? s->channels : 1;
172  s->nb_null_samples = av_calloc(s->independent_channels,
173  sizeof(*s->nb_null_samples));
174  if (!s->nb_null_samples)
175  return AVERROR(ENOMEM);
176  s->start = av_malloc_array(sizeof(*s->start), s->independent_channels);
177  if (!s->start)
178  return AVERROR(ENOMEM);
179  for (c = 0; c < s->independent_channels; c++)
180  s->start[c] = INT64_MIN;
181 
182  switch (inlink->format) {
183  case AV_SAMPLE_FMT_DBL: s->silencedetect = silencedetect_dbl; break;
184  case AV_SAMPLE_FMT_FLT: s->silencedetect = silencedetect_flt; break;
185  case AV_SAMPLE_FMT_S32:
186  s->noise *= INT32_MAX;
187  s->silencedetect = silencedetect_s32;
188  break;
189  case AV_SAMPLE_FMT_S16:
190  s->noise *= INT16_MAX;
191  s->silencedetect = silencedetect_s16;
192  break;
193  case AV_SAMPLE_FMT_DBLP: s->silencedetect = silencedetect_dblp; break;
194  case AV_SAMPLE_FMT_FLTP: s->silencedetect = silencedetect_fltp; break;
195  case AV_SAMPLE_FMT_S32P:
196  s->noise *= INT32_MAX;
197  s->silencedetect = silencedetect_s32p;
198  break;
199  case AV_SAMPLE_FMT_S16P:
200  s->noise *= INT16_MAX;
201  s->silencedetect = silencedetect_s16p;
202  break;
203  default:
204  return AVERROR_BUG;
205  }
206 
207  return 0;
208 }
209 
210 static int filter_frame(AVFilterLink *inlink, AVFrame *insamples)
211 {
212  SilenceDetectContext *s = inlink->dst->priv;
213  const int nb_channels = inlink->channels;
214  const int srate = inlink->sample_rate;
215  const int nb_samples = insamples->nb_samples * nb_channels;
216  const int64_t nb_samples_notify = s->duration * (s->mono ? 1 : nb_channels);
217  int c;
218 
219  // scale number of null samples to the new sample rate
220  if (s->last_sample_rate && s->last_sample_rate != srate)
221  for (c = 0; c < s->independent_channels; c++) {
222  s->nb_null_samples[c] = srate * s->nb_null_samples[c] / s->last_sample_rate;
223  }
224  s->last_sample_rate = srate;
225  s->time_base = inlink->time_base;
226  s->frame_end = insamples->pts + av_rescale_q(insamples->nb_samples,
227  (AVRational){ 1, s->last_sample_rate }, inlink->time_base);
228 
229  s->silencedetect(s, insamples, nb_samples, nb_samples_notify,
230  inlink->time_base);
231 
232  return ff_filter_frame(inlink->dst->outputs[0], insamples);
233 }
234 
236 {
237  SilenceDetectContext *s = ctx->priv;
238  int c;
239 
240  for (c = 0; c < s->independent_channels; c++)
241  if (s->start[c] > INT64_MIN)
242  update(s, NULL, 0, c, 0, s->time_base);
243  av_freep(&s->nb_null_samples);
244  av_freep(&s->start);
245 }
246 
248  {
249  .name = "default",
250  .type = AVMEDIA_TYPE_AUDIO,
251  .config_props = config_input,
252  .filter_frame = filter_frame,
253  },
254 };
255 
257  {
258  .name = "default",
259  .type = AVMEDIA_TYPE_AUDIO,
260  },
261 };
262 
264  .name = "silencedetect",
265  .description = NULL_IF_CONFIG_SMALL("Detect silence."),
266  .priv_size = sizeof(SilenceDetectContext),
267  .uninit = uninit,
274  .priv_class = &silencedetect_class,
276 };
AV_SAMPLE_FMT_FLTP
@ AV_SAMPLE_FMT_FLTP
float, planar
Definition: samplefmt.h:69
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:120
ff_filter_frame
int ff_filter_frame(AVFilterLink *link, AVFrame *frame)
Send a frame of data to the next filter.
Definition: avfilter.c:1018
SilenceDetectContext::noise
double noise
noise amplitude ratio
Definition: af_silencedetect.c:37
silencedetect_outputs
static const AVFilterPad silencedetect_outputs[]
Definition: af_silencedetect.c:256
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:210
AVFrame
This structure describes decoded (raw) audio or video data.
Definition: frame.h:317
AVFrame::pts
int64_t pts
Presentation timestamp in time_base units (time when frame should be shown to user).
Definition: frame.h:424
AVOption
AVOption.
Definition: opt.h:247
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:68
silencedetect_inputs
static const AVFilterPad silencedetect_inputs[]
Definition: af_silencedetect.c:247
AV_OPT_TYPE_DURATION
@ AV_OPT_TYPE_DURATION
Definition: opt.h:238
float.h
AVFilter::name
const char * name
Filter name.
Definition: avfilter.h:169
uninit
static av_cold void uninit(AVFilterContext *ctx)
Definition: af_silencedetect.c:235
formats.h
update
static av_always_inline void update(SilenceDetectContext *s, AVFrame *insamples, int is_silence, int current_sample, int64_t nb_samples_notify, AVRational time_base)
Definition: af_silencedetect.c:78
AVFilterPad
A filter pad used for either input or output.
Definition: internal.h:50
av_cold
#define av_cold
Definition: attributes.h:90
duration
int64_t duration
Definition: movenc.c:64
s
#define s(width, name)
Definition: cbs_vp9.c:257
AV_OPT_TYPE_DOUBLE
@ AV_OPT_TYPE_DOUBLE
Definition: opt.h:226
AVMEDIA_TYPE_AUDIO
@ AVMEDIA_TYPE_AUDIO
Definition: avutil.h:202
ctx
AVFormatContext * ctx
Definition: movenc.c:48
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:141
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:168
FILTER_INPUTS
#define FILTER_INPUTS(array)
Definition: internal.h:191
SilenceDetectContext::channels
int channels
number of channels
Definition: af_silencedetect.c:40
AVClass
Describe the class of an AVClass context structure.
Definition: log.h:66
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_af_silencedetect
const AVFilter ff_af_silencedetect
Definition: af_silencedetect.c:263
SILENCE_DETECT_PLANAR
#define SILENCE_DETECT_PLANAR(name, type)
Definition: af_silencedetect.c:134
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:76
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:117
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:67
AV_LOG_INFO
#define AV_LOG_INFO
Standard information.
Definition: log.h:191
internal.h
AVFrame::nb_samples
int nb_samples
number of audio samples (per channel) described by this frame
Definition: frame.h:397
AV_TIME_BASE
#define AV_TIME_BASE
Internal time base represented as integer.
Definition: avutil.h:254
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:61
AVFilterPad::name
const char * name
Pad name.
Definition: internal.h:56
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:128
av_calloc
void * av_calloc(size_t nmemb, size_t size)
Definition: mem.c:271
AVFilter
Filter definition.
Definition: avfilter.h:165
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)
noise
static int noise(AVBSFContext *ctx, AVPacket *pkt)
Definition: noise_bsf.c:121
avfilter.h
AVFrame::metadata
AVDictionary * metadata
metadata.
Definition: frame.h:608
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:137
AV_SAMPLE_FMT_DBLP
@ AV_SAMPLE_FMT_DBLP
double, planar
Definition: samplefmt.h:70
config_input
static int config_input(AVFilterLink *inlink)
Definition: af_silencedetect.c:163
AVFilterContext
An instance of a filter.
Definition: avfilter.h:402
audio.h
SilenceDetectContext::silencedetect
void(* silencedetect)(struct SilenceDetectContext *s, AVFrame *insamples, int nb_samples, int64_t nb_samples_notify, AVRational time_base)
Definition: af_silencedetect.c:48
AV_OPT_TYPE_BOOL
@ AV_OPT_TYPE_BOOL
Definition: opt.h:241
FILTER_OUTPUTS
#define FILTER_OUTPUTS(array)
Definition: internal.h:192
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:70
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:28
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:64
AV_SAMPLE_FMT_S32
@ AV_SAMPLE_FMT_S32
signed 32 bits
Definition: samplefmt.h:62
snprintf
#define snprintf
Definition: snprintf.h:34
AV_SAMPLE_FMT_FLT
@ AV_SAMPLE_FMT_FLT
float
Definition: samplefmt.h:63
channel
channel
Definition: ebur128.h:39
FILTER_SAMPLEFMTS
#define FILTER_SAMPLEFMTS(...)
Definition: internal.h:179
nb_channels
int nb_channels
Definition: channel_layout.c:81