Skip to content
体验新版
项目
组织
正在加载...
登录
切换导航
打开侧边栏
小白菜888
Ffmpeg
提交
c07bc1d6
F
Ffmpeg
项目概览
小白菜888
/
Ffmpeg
通知
3
Star
0
Fork
0
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
0
列表
看板
标记
里程碑
合并请求
0
DevOps
流水线
流水线任务
计划
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
F
Ffmpeg
项目概览
项目概览
详情
发布
仓库
仓库
文件
提交
分支
标签
贡献者
分支图
比较
Issue
0
Issue
0
列表
看板
标记
里程碑
合并请求
0
合并请求
0
Pages
DevOps
DevOps
流水线
流水线任务
计划
分析
分析
仓库分析
DevOps
Wiki
0
Wiki
成员
成员
收起侧边栏
关闭侧边栏
动态
分支图
创建新Issue
流水线任务
提交
Issue看板
体验新版 GitCode,发现更多精彩内容 >>
提交
c07bc1d6
编写于
10月 08, 2018
作者:
P
Paul B Mahol
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
avfilter/af_silenceremove: add options to keep min duration of silence
上级
c27c7b49
变更
2
隐藏空白更改
内联
并排
Showing
2 changed file
with
130 addition
and
39 deletion
+130
-39
libavfilter/af_silenceremove.c
libavfilter/af_silenceremove.c
+129
-38
tests/fate/filter-audio.mak
tests/fate/filter-audio.mak
+1
-1
未找到文件。
libavfilter/af_silenceremove.c
浏览文件 @
c07bc1d6
...
...
@@ -45,20 +45,32 @@ typedef struct SilenceRemoveContext {
int
start_periods
;
int64_t
start_duration
;
int64_t
start_duration_opt
;
double
start_threshold
;
int64_t
start_silence
;
int64_t
start_silence_opt
;
int
stop_periods
;
int64_t
stop_duration
;
int64_t
stop_duration_opt
;
double
stop_threshold
;
int64_t
stop_silence
;
int64_t
stop_silence_opt
;
double
*
start_holdoff
;
double
*
start_silence_hold
;
size_t
start_holdoff_offset
;
size_t
start_holdoff_end
;
size_t
start_silence_offset
;
size_t
start_silence_end
;
int
start_found_periods
;
double
*
stop_holdoff
;
double
*
stop_silence_hold
;
size_t
stop_holdoff_offset
;
size_t
stop_holdoff_end
;
size_t
stop_silence_offset
;
size_t
stop_silence_end
;
int
stop_found_periods
;
double
window_ratio
;
...
...
@@ -68,7 +80,6 @@ typedef struct SilenceRemoveContext {
int
window_size
;
double
sum
;
int
leave_silence
;
int
restart
;
int64_t
next_pts
;
...
...
@@ -78,19 +89,21 @@ typedef struct SilenceRemoveContext {
}
SilenceRemoveContext
;
#define OFFSET(x) offsetof(SilenceRemoveContext, x)
#define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_AUDIO_PARAM
#define AF AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_AUDIO_PARAM
static
const
AVOption
silenceremove_options
[]
=
{
{
"start_periods"
,
NULL
,
OFFSET
(
start_periods
),
AV_OPT_TYPE_INT
,
{.
i64
=
0
},
0
,
9000
,
FLAGS
},
{
"start_duration"
,
NULL
,
OFFSET
(
start_duration
),
AV_OPT_TYPE_DURATION
,
{.
i64
=
0
},
0
,
INT32_MAX
,
FLAGS
},
{
"start_threshold"
,
NULL
,
OFFSET
(
start_threshold
),
AV_OPT_TYPE_DOUBLE
,
{.
dbl
=
0
},
0
,
DBL_MAX
,
FLAGS
},
{
"stop_periods"
,
NULL
,
OFFSET
(
stop_periods
),
AV_OPT_TYPE_INT
,
{.
i64
=
0
},
-
9000
,
9000
,
FLAGS
},
{
"stop_duration"
,
NULL
,
OFFSET
(
stop_duration
),
AV_OPT_TYPE_DURATION
,
{.
i64
=
0
},
0
,
INT32_MAX
,
FLAGS
},
{
"stop_threshold"
,
NULL
,
OFFSET
(
stop_threshold
),
AV_OPT_TYPE_DOUBLE
,
{.
dbl
=
0
},
0
,
DBL_MAX
,
FLAGS
},
{
"leave_silence"
,
NULL
,
OFFSET
(
leave_silence
),
AV_OPT_TYPE_BOOL
,
{.
i64
=
0
},
0
,
1
,
FLAGS
},
{
"detection"
,
NULL
,
OFFSET
(
detection
),
AV_OPT_TYPE_INT
,
{.
i64
=
1
},
0
,
1
,
FLAGS
,
"detection"
},
{
"peak"
,
0
,
0
,
AV_OPT_TYPE_CONST
,
{.
i64
=
0
},
0
,
0
,
FLAGS
,
"detection"
},
{
"rms"
,
0
,
0
,
AV_OPT_TYPE_CONST
,
{.
i64
=
1
},
0
,
0
,
FLAGS
,
"detection"
},
{
"window"
,
NULL
,
OFFSET
(
window_ratio
),
AV_OPT_TYPE_DOUBLE
,
{.
dbl
=
0
.
02
},
0
,
10
,
FLAGS
},
{
"start_periods"
,
NULL
,
OFFSET
(
start_periods
),
AV_OPT_TYPE_INT
,
{.
i64
=
0
},
0
,
9000
,
AF
},
{
"start_duration"
,
NULL
,
OFFSET
(
start_duration_opt
),
AV_OPT_TYPE_DURATION
,
{.
i64
=
0
},
0
,
INT32_MAX
,
AF
},
{
"start_threshold"
,
NULL
,
OFFSET
(
start_threshold
),
AV_OPT_TYPE_DOUBLE
,
{.
dbl
=
0
},
0
,
DBL_MAX
,
AF
},
{
"start_silence"
,
NULL
,
OFFSET
(
start_silence_opt
),
AV_OPT_TYPE_DURATION
,
{.
i64
=
0
},
0
,
INT32_MAX
,
AF
},
{
"stop_periods"
,
NULL
,
OFFSET
(
stop_periods
),
AV_OPT_TYPE_INT
,
{.
i64
=
0
},
-
9000
,
9000
,
AF
},
{
"stop_duration"
,
NULL
,
OFFSET
(
stop_duration_opt
),
AV_OPT_TYPE_DURATION
,
{.
i64
=
0
},
0
,
INT32_MAX
,
AF
},
{
"stop_threshold"
,
NULL
,
OFFSET
(
stop_threshold
),
AV_OPT_TYPE_DOUBLE
,
{.
dbl
=
0
},
0
,
DBL_MAX
,
AF
},
{
"stop_silence"
,
NULL
,
OFFSET
(
stop_silence_opt
),
AV_OPT_TYPE_DURATION
,
{.
i64
=
0
},
0
,
INT32_MAX
,
AF
},
{
"detection"
,
NULL
,
OFFSET
(
detection
),
AV_OPT_TYPE_INT
,
{.
i64
=
1
},
0
,
1
,
AF
,
"detection"
},
{
"peak"
,
0
,
0
,
AV_OPT_TYPE_CONST
,
{.
i64
=
0
},
0
,
0
,
AF
,
"detection"
},
{
"rms"
,
0
,
0
,
AV_OPT_TYPE_CONST
,
{.
i64
=
1
},
0
,
0
,
AF
,
"detection"
},
{
"window"
,
NULL
,
OFFSET
(
window_ratio
),
AV_OPT_TYPE_DOUBLE
,
{.
dbl
=
0
.
02
},
0
,
10
,
AF
},
{
NULL
}
};
...
...
@@ -158,7 +171,7 @@ static av_cold int init(AVFilterContext *ctx)
s
->
update
=
update_rms
;
s
->
compute
=
compute_rms
;
break
;
}
;
}
return
0
;
}
...
...
@@ -184,19 +197,14 @@ static int config_input(AVFilterLink *inlink)
clear_window
(
s
);
s
->
start_duration
=
av_rescale
(
s
->
start_duration
,
inlink
->
sample_rate
,
s
->
start_duration
=
av_rescale
(
s
->
start_duration
_opt
,
inlink
->
sample_rate
,
AV_TIME_BASE
);
if
(
s
->
start_duration
<
0
)
{
av_log
(
ctx
,
AV_LOG_WARNING
,
"start duration must be non-negative
\n
"
);
s
->
start_duration
=
-
s
->
start_duration
;
}
s
->
stop_duration
=
av_rescale
(
s
->
stop_duration
,
inlink
->
sample_rate
,
s
->
start_silence
=
av_rescale
(
s
->
start_silence_opt
,
inlink
->
sample_rate
,
AV_TIME_BASE
);
s
->
stop_duration
=
av_rescale
(
s
->
stop_duration_opt
,
inlink
->
sample_rate
,
AV_TIME_BASE
);
s
->
stop_silence
=
av_rescale
(
s
->
stop_silence_opt
,
inlink
->
sample_rate
,
AV_TIME_BASE
);
if
(
s
->
stop_duration
<
0
)
{
av_log
(
ctx
,
AV_LOG_WARNING
,
"stop duration must be non-negative
\n
"
);
s
->
stop_duration
=
-
s
->
stop_duration
;
}
s
->
start_holdoff
=
av_malloc_array
(
FFMAX
(
s
->
start_duration
,
1
),
sizeof
(
*
s
->
start_holdoff
)
*
...
...
@@ -204,6 +212,12 @@ static int config_input(AVFilterLink *inlink)
if
(
!
s
->
start_holdoff
)
return
AVERROR
(
ENOMEM
);
s
->
start_silence_hold
=
av_malloc_array
(
FFMAX
(
s
->
start_silence
,
1
),
sizeof
(
*
s
->
start_silence_hold
)
*
inlink
->
channels
);
if
(
!
s
->
start_silence_hold
)
return
AVERROR
(
ENOMEM
);
s
->
start_holdoff_offset
=
0
;
s
->
start_holdoff_end
=
0
;
s
->
start_found_periods
=
0
;
...
...
@@ -214,6 +228,12 @@ static int config_input(AVFilterLink *inlink)
if
(
!
s
->
stop_holdoff
)
return
AVERROR
(
ENOMEM
);
s
->
stop_silence_hold
=
av_malloc_array
(
FFMAX
(
s
->
stop_silence
,
1
),
sizeof
(
*
s
->
stop_silence_hold
)
*
inlink
->
channels
);
if
(
!
s
->
stop_silence_hold
)
return
AVERROR
(
ENOMEM
);
s
->
stop_holdoff_offset
=
0
;
s
->
stop_holdoff_end
=
0
;
s
->
stop_found_periods
=
0
;
...
...
@@ -228,8 +248,10 @@ static int config_input(AVFilterLink *inlink)
static
void
flush
(
SilenceRemoveContext
*
s
,
AVFrame
*
out
,
AVFilterLink
*
outlink
,
int
*
nb_samples_written
,
int
*
ret
)
int
*
nb_samples_written
,
int
*
ret
,
int
flush_silence
)
{
AVFrame
*
silence
;
if
(
*
nb_samples_written
)
{
out
->
nb_samples
=
*
nb_samples_written
/
outlink
->
channels
;
...
...
@@ -239,10 +261,43 @@ static void flush(SilenceRemoveContext *s,
outlink
->
time_base
);
*
ret
=
ff_filter_frame
(
outlink
,
out
);
if
(
*
ret
<
0
)
return
;
*
nb_samples_written
=
0
;
}
else
{
av_frame_free
(
&
out
);
}
if
(
s
->
stop_silence_end
<=
0
||
!
flush_silence
)
return
;
silence
=
ff_get_audio_buffer
(
outlink
,
s
->
stop_silence_end
/
outlink
->
channels
);
if
(
!
silence
)
{
*
ret
=
AVERROR
(
ENOMEM
);
return
;
}
if
(
s
->
stop_silence_offset
<
s
->
stop_silence_end
)
{
memcpy
(
silence
->
data
[
0
],
&
s
->
stop_silence_hold
[
s
->
stop_silence_offset
],
(
s
->
stop_silence_end
-
s
->
stop_silence_offset
)
*
sizeof
(
double
));
}
if
(
s
->
stop_silence_offset
>
0
)
{
memcpy
(
silence
->
data
[
0
]
+
(
s
->
stop_silence_end
-
s
->
stop_silence_offset
)
*
sizeof
(
double
),
&
s
->
stop_silence_hold
[
0
],
s
->
stop_silence_offset
*
sizeof
(
double
));
}
s
->
stop_silence_offset
=
0
;
s
->
stop_silence_end
=
0
;
silence
->
pts
=
s
->
next_pts
;
s
->
next_pts
+=
av_rescale_q
(
silence
->
nb_samples
,
(
AVRational
){
1
,
outlink
->
sample_rate
},
outlink
->
time_base
);
*
ret
=
ff_filter_frame
(
outlink
,
silence
);
}
static
int
filter_frame
(
AVFilterLink
*
inlink
,
AVFrame
*
in
)
...
...
@@ -285,12 +340,22 @@ silence_trim:
s
->
start_holdoff_offset
=
0
;
s
->
start_holdoff_end
=
0
;
s
->
start_silence_offset
=
0
;
s
->
start_silence_end
=
0
;
}
}
else
{
s
->
start_holdoff_end
=
0
;
for
(
j
=
0
;
j
<
inlink
->
channels
;
j
++
)
for
(
j
=
0
;
j
<
inlink
->
channels
;
j
++
)
{
s
->
update
(
s
,
ibuf
[
j
]);
if
(
s
->
start_silence
)
{
s
->
start_silence_hold
[
s
->
start_silence_offset
++
]
=
ibuf
[
j
];
s
->
start_silence_end
=
FFMIN
(
s
->
start_silence_end
+
1
,
inlink
->
channels
*
s
->
start_silence
);
if
(
s
->
start_silence_offset
>=
inlink
->
channels
*
s
->
start_silence
)
{
s
->
start_silence_offset
=
0
;
}
}
}
ibuf
+=
inlink
->
channels
;
nb_samples_read
+=
inlink
->
channels
;
...
...
@@ -305,13 +370,28 @@ silence_trim_flush:
if
(
!
nbs
)
break
;
out
=
ff_get_audio_buffer
(
inlink
,
nbs
/
inlink
->
channels
);
out
=
ff_get_audio_buffer
(
inlink
,
nbs
/
inlink
->
channels
+
s
->
start_silence_end
/
inlink
->
channels
);
if
(
!
out
)
{
av_frame_free
(
&
in
);
return
AVERROR
(
ENOMEM
);
}
memcpy
(
out
->
data
[
0
],
&
s
->
start_holdoff
[
s
->
start_holdoff_offset
],
if
(
s
->
start_silence_end
>
0
)
{
if
(
s
->
start_silence_offset
<
s
->
start_silence_end
)
{
memcpy
(
out
->
data
[
0
],
&
s
->
start_silence_hold
[
s
->
start_silence_offset
],
(
s
->
start_silence_end
-
s
->
start_silence_offset
)
*
sizeof
(
double
));
}
if
(
s
->
start_silence_offset
>
0
)
{
memcpy
(
out
->
data
[
0
]
+
(
s
->
start_silence_end
-
s
->
start_silence_offset
)
*
sizeof
(
double
),
&
s
->
start_silence_hold
[
0
],
s
->
start_silence_offset
*
sizeof
(
double
));
}
}
memcpy
(
out
->
data
[
0
]
+
s
->
start_silence_end
*
sizeof
(
double
),
&
s
->
start_holdoff
[
s
->
start_holdoff_offset
],
nbs
*
sizeof
(
double
));
out
->
pts
=
s
->
next_pts
;
...
...
@@ -326,6 +406,8 @@ silence_trim_flush:
if
(
s
->
start_holdoff_offset
==
s
->
start_holdoff_end
)
{
s
->
start_holdoff_offset
=
0
;
s
->
start_holdoff_end
=
0
;
s
->
start_silence_offset
=
0
;
s
->
start_silence_end
=
0
;
s
->
mode
=
SILENCE_COPY
;
goto
silence_copy
;
}
...
...
@@ -350,9 +432,9 @@ silence_copy:
for
(
j
=
0
;
j
<
inlink
->
channels
;
j
++
)
threshold
&=
s
->
compute
(
s
,
ibuf
[
j
])
>
s
->
stop_threshold
;
if
(
threshold
&&
s
->
stop_holdoff_end
&&
!
s
->
leave
_silence
)
{
if
(
threshold
&&
s
->
stop_holdoff_end
&&
!
s
->
stop
_silence
)
{
s
->
mode
=
SILENCE_COPY_FLUSH
;
flush
(
s
,
out
,
outlink
,
&
nb_samples_written
,
&
ret
);
flush
(
s
,
out
,
outlink
,
&
nb_samples_written
,
&
ret
,
0
);
goto
silence_copy_flush
;
}
else
if
(
threshold
)
{
for
(
j
=
0
;
j
<
inlink
->
channels
;
j
++
)
{
...
...
@@ -364,9 +446,12 @@ silence_copy:
}
else
if
(
!
threshold
)
{
for
(
j
=
0
;
j
<
inlink
->
channels
;
j
++
)
{
s
->
update
(
s
,
*
ibuf
);
if
(
s
->
leave_silence
)
{
*
obuf
++
=
*
ibuf
;
nb_samples_written
++
;
if
(
s
->
stop_silence
)
{
s
->
stop_silence_hold
[
s
->
stop_silence_offset
++
]
=
*
ibuf
;
s
->
stop_silence_end
=
FFMIN
(
s
->
stop_silence_end
+
1
,
inlink
->
channels
*
s
->
stop_silence
);
if
(
s
->
stop_silence_offset
>=
inlink
->
channels
*
s
->
stop_silence
)
{
s
->
stop_silence_offset
=
0
;
}
}
s
->
stop_holdoff
[
s
->
stop_holdoff_end
++
]
=
*
ibuf
++
;
...
...
@@ -380,26 +465,28 @@ silence_copy:
if
(
!
s
->
restart
)
{
s
->
mode
=
SILENCE_STOP
;
flush
(
s
,
out
,
outlink
,
&
nb_samples_written
,
&
ret
);
flush
(
s
,
out
,
outlink
,
&
nb_samples_written
,
&
ret
,
1
);
goto
silence_stop
;
}
else
{
s
->
stop_found_periods
=
0
;
s
->
start_found_periods
=
0
;
s
->
start_holdoff_offset
=
0
;
s
->
start_holdoff_end
=
0
;
s
->
start_silence_offset
=
0
;
s
->
start_silence_end
=
0
;
clear_window
(
s
);
s
->
mode
=
SILENCE_TRIM
;
flush
(
s
,
out
,
outlink
,
&
nb_samples_written
,
&
ret
);
flush
(
s
,
out
,
outlink
,
&
nb_samples_written
,
&
ret
,
1
);
goto
silence_trim
;
}
}
s
->
mode
=
SILENCE_COPY_FLUSH
;
flush
(
s
,
out
,
outlink
,
&
nb_samples_written
,
&
ret
);
flush
(
s
,
out
,
outlink
,
&
nb_samples_written
,
&
ret
,
0
);
goto
silence_copy_flush
;
}
}
}
flush
(
s
,
out
,
outlink
,
&
nb_samples_written
,
&
ret
);
flush
(
s
,
out
,
outlink
,
&
nb_samples_written
,
&
ret
,
0
);
}
else
{
memcpy
(
obuf
,
ibuf
,
sizeof
(
double
)
*
nbs
*
inlink
->
channels
);
...
...
@@ -439,6 +526,8 @@ silence_copy_flush:
if
(
s
->
stop_holdoff_offset
==
s
->
stop_holdoff_end
)
{
s
->
stop_holdoff_offset
=
0
;
s
->
stop_holdoff_end
=
0
;
s
->
stop_silence_offset
=
0
;
s
->
stop_silence_end
=
0
;
s
->
mode
=
SILENCE_COPY
;
goto
silence_copy
;
}
...
...
@@ -519,7 +608,9 @@ static av_cold void uninit(AVFilterContext *ctx)
SilenceRemoveContext
*
s
=
ctx
->
priv
;
av_freep
(
&
s
->
start_holdoff
);
av_freep
(
&
s
->
start_silence_hold
);
av_freep
(
&
s
->
stop_holdoff
);
av_freep
(
&
s
->
stop_silence_hold
);
av_freep
(
&
s
->
window
);
}
...
...
tests/fate/filter-audio.mak
浏览文件 @
c07bc1d6
...
...
@@ -180,7 +180,7 @@ fate-filter-pan-downmix2: CMD = framecrc -ss 3.14 -i $(SRC) -frames:a 20 -filter
FATE_AFILTER_SAMPLES-$(call
FILTERDEMDECENCMUX,
SILENCEREMOVE,
WAV,
PCM_S16LE,
PCM_S16LE,
WAV)
+=
fate-filter-silenceremove
fate-filter-silenceremove
:
SRC = $(TARGET_SAMPLES)/audio-reference/divertimenti_2ch_96kHz_s24.wav
fate-filter-silenceremove
:
CMD = framecrc -i $(SRC) -frames:a 30 -af silenceremove=
0:0:0:-1:0:
-90dB
fate-filter-silenceremove
:
CMD = framecrc -i $(SRC) -frames:a 30 -af silenceremove=
start_periods=0:start_duration=0:start_threshold=0:stop_periods=-1:stop_duration=0:stop_threshold=
-90dB
FATE_AFILTER_SAMPLES-$(call
FILTERDEMDECENCMUX,
STEREOTOOLS,
WAV,
PCM_S16LE,
PCM_S16LE,
WAV)
+=
fate-filter-stereotools
fate-filter-stereotools
:
SRC = $(TARGET_SAMPLES)/audio-reference/luckynight_2ch_44kHz_s16.wav
...
...
编辑
预览
Markdown
is supported
0%
请重试
或
添加新附件
.
添加附件
取消
You are about to add
0
people
to the discussion. Proceed with caution.
先完成此消息的编辑!
取消
想要评论请
注册
或
登录