FFmpeg
Loading...
Searching...
No Matches
Data Structures | Macros | Functions | Variables
af_speechnorm.c File Reference

Speech Normalizer. More...

#include <float.h>
#include "libavutil/avassert.h"
#include "libavutil/channel_layout.h"
#include "libavutil/mem.h"
#include "libavutil/opt.h"
#include "bufferqueue.h"
#include "audio.h"
#include "avfilter.h"
#include "filters.h"

Go to the source code of this file.

Data Structures

struct  PeriodItem
 
struct  ChannelContext
 
struct  SpeechNormalizerContext
 

Macros

#define FF_BUFQUEUE_SIZE   (1024)
 
#define MAX_ITEMS   882000
 
#define MIN_PEAK   (1. / 32768.)
 
#define OFFSET(x)
 
#define FLAGS   AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_RUNTIME_PARAM
 
#define ANALYZE_CHANNEL(name, ptype, zero, min_peak)
 
#define FILTER_CHANNELS(name, ptype)
 
#define FILTER_LINK_CHANNELS(name, ptype, tlerp)
 

Functions

 AVFILTER_DEFINE_CLASS (speechnorm)
 
static int get_pi_samples (PeriodItem *pi, int start, int end, int remain)
 
static int available_samples (AVFilterContext *ctx)
 
static void consume_pi (ChannelContext *cc, int nb_samples)
 
static double next_gain (AVFilterContext *ctx, double pi_max_peak, int bypass, double state, double pi_rms_sum, int pi_size)
 
static void next_pi (AVFilterContext *ctx, ChannelContext *cc, int bypass)
 
static double min_gain (AVFilterContext *ctx, ChannelContext *cc, int max_size)
 
static double dlerp (double min, double max, double mix)
 
static float flerp (float min, float max, float mix)
 
static int filter_frame (AVFilterContext *ctx)
 
static int activate (AVFilterContext *ctx)
 
static int config_input (AVFilterLink *inlink)
 
static int process_command (AVFilterContext *ctx, const char *cmd, const char *args, char *res, int res_len, int flags)
 
static av_cold void uninit (AVFilterContext *ctx)
 

Variables

static const AVOption speechnorm_options []
 
static const AVFilterPad inputs []
 
const FFFilter ff_af_speechnorm
 

Detailed Description

Speech Normalizer.

Definition in file af_speechnorm.c.

Macro Definition Documentation

◆ FF_BUFQUEUE_SIZE

#define FF_BUFQUEUE_SIZE   (1024)

Definition at line 35 of file af_speechnorm.c.

◆ MAX_ITEMS

#define MAX_ITEMS   882000

Definition at line 42 of file af_speechnorm.c.

Referenced by get_pi_samples(), min_gain(), and next_pi().

◆ MIN_PEAK

#define MIN_PEAK   (1. / 32768.)

Definition at line 43 of file af_speechnorm.c.

◆ OFFSET

#define OFFSET ( x)
Value:
offsetof(SpeechNormalizerContext, x)

Definition at line 94 of file af_speechnorm.c.

◆ FLAGS

#define FLAGS   AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_RUNTIME_PARAM

Definition at line 95 of file af_speechnorm.c.

◆ ANALYZE_CHANNEL

#define ANALYZE_CHANNEL ( name,
ptype,
zero,
min_peak )

Definition at line 235 of file af_speechnorm.c.

◆ FILTER_CHANNELS

#define FILTER_CHANNELS ( name,
ptype )
Value:
static void filter_channels_## name (AVFilterContext *ctx, \
AVFrame *in, AVFrame *out, int nb_samples) \
{ \
SpeechNormalizerContext *s = ctx->priv; \
AVFilterLink *inlink = ctx->inputs[0]; \
\
for (int ch = 0; ch < inlink->ch_layout.nb_channels; ch++) { \
ChannelContext *cc = &s->cc[ch]; \
const ptype *src = (const ptype *)in->extended_data[ch]; \
ptype *dst = (ptype *)out->extended_data[ch]; \
enum AVChannel channel = av_channel_layout_channel_from_index(&inlink->ch_layout, ch); \
const int bypass = av_channel_layout_index_from_channel(&s->ch_layout, channel) < 0; \
int n = 0; \
\
while (n < nb_samples) { \
ptype gain; \
int size; \
\
next_pi(ctx, cc, bypass); \
size = FFMIN(nb_samples - n, cc->pi_size); \
av_assert1(size > 0); \
gain = cc->gain_state; \
consume_pi(cc, size); \
for (int i = n; !ctx->is_disabled && i < n + size; i++) \
dst[i] = src[i] * gain; \
n += size; \
} \
} \
}
dst
uint8_t ptrdiff_t const uint8_t ptrdiff_t int intptr_t intptr_t int int16_t * dst
Definition dsp.h:87
i
#define i(width, name, range_min, range_max)
Definition cbs_h264.c:63
s
#define s(width, name)
Definition cbs_vp9.c:198
channel
channel
Use these values when setting the channel map with ebur128_set_channel().
Definition ebur128.h:39
av_channel_layout_index_from_channel
int av_channel_layout_index_from_channel(const AVChannelLayout *channel_layout, enum AVChannel channel)
Get the index of a given channel in a channel layout.
Definition channel_layout.c:715
av_channel_layout_channel_from_index
enum AVChannel av_channel_layout_channel_from_index(const AVChannelLayout *channel_layout, unsigned int idx)
Get the channel with the given index in a channel layout.
Definition channel_layout.c:674
AVChannel
AVChannel
Definition channel_layout.h:47
FFMIN
#define FFMIN(a, b)
Definition macros.h:49
name
const char * name
Definition qsvenc.c:142
AVChannelLayout::nb_channels
int nb_channels
Number of channels in this layout.
Definition channel_layout.h:329
AVFilterContext
An instance of a filter.
Definition avfilter.h:273
AVFrame
This structure describes decoded (raw) audio or video data.
Definition frame.h:472
src
#define src
Definition vp8dsp.c:248
out
static FILE * out
Definition movenc.c:55
ctx
static AVFormatContext * ctx
Definition movenc.c:49
size
int size
Definition twinvq_data.h:10344

Definition at line 314 of file af_speechnorm.c.

◆ FILTER_LINK_CHANNELS

#define FILTER_LINK_CHANNELS ( name,
ptype,
tlerp )

Definition at line 358 of file af_speechnorm.c.

Function Documentation

◆ AVFILTER_DEFINE_CLASS()

AVFILTER_DEFINE_CLASS ( speechnorm )

◆ get_pi_samples()

static int get_pi_samples ( PeriodItem * pi,
int start,
int end,
int remain )
static

Definition at line 123 of file af_speechnorm.c.

Referenced by available_samples().

◆ available_samples()

static int available_samples ( AVFilterContext * ctx)
static

Definition at line 144 of file af_speechnorm.c.

Referenced by activate().

◆ consume_pi()

static void consume_pi ( ChannelContext * cc,
int nb_samples )
static

Definition at line 160 of file af_speechnorm.c.

◆ next_gain()

static double next_gain ( AVFilterContext * ctx,
double pi_max_peak,
int bypass,
double state,
double pi_rms_sum,
int pi_size )
static

Definition at line 169 of file af_speechnorm.c.

Referenced by min_gain(), and next_pi().

◆ next_pi()

static void next_pi ( AVFilterContext * ctx,
ChannelContext * cc,
int bypass )
static

Definition at line 189 of file af_speechnorm.c.

◆ min_gain()

static double min_gain ( AVFilterContext * ctx,
ChannelContext * cc,
int max_size )
static

Definition at line 211 of file af_speechnorm.c.

Referenced by min_gain().

◆ dlerp()

static double dlerp ( double min,
double max,
double mix )
static

Definition at line 348 of file af_speechnorm.c.

Referenced by dlerp().

◆ flerp()

static float flerp ( float min,
float max,
float mix )
static

Definition at line 353 of file af_speechnorm.c.

◆ filter_frame()

static int filter_frame ( AVFilterContext * ctx)
static

Definition at line 413 of file af_speechnorm.c.

Referenced by activate().

◆ activate()

static int activate ( AVFilterContext * ctx)
static

Definition at line 476 of file af_speechnorm.c.

◆ config_input()

static int config_input ( AVFilterLink * inlink)
static

Definition at line 523 of file af_speechnorm.c.

◆ process_command()

static int process_command ( AVFilterContext * ctx,
const char * cmd,
const char * args,
char * res,
int res_len,
int flags )
static

Definition at line 560 of file af_speechnorm.c.

◆ uninit()

static av_cold void uninit ( AVFilterContext * ctx)
static

Definition at line 576 of file af_speechnorm.c.

Variable Documentation

◆ speechnorm_options

const AVOption speechnorm_options[]
static
Initial value:
= {
{ "peak", "set the peak value", OFFSET(peak_value), AV_OPT_TYPE_DOUBLE, {.dbl=0.95}, 0.0, 1.0, FLAGS },
{ "p", "set the peak value", OFFSET(peak_value), AV_OPT_TYPE_DOUBLE, {.dbl=0.95}, 0.0, 1.0, FLAGS },
{ "expansion", "set the max expansion factor", OFFSET(max_expansion), AV_OPT_TYPE_DOUBLE, {.dbl=2.0}, 1.0, 50.0, FLAGS },
{ "e", "set the max expansion factor", OFFSET(max_expansion), AV_OPT_TYPE_DOUBLE, {.dbl=2.0}, 1.0, 50.0, FLAGS },
{ "compression", "set the max compression factor", OFFSET(max_compression), AV_OPT_TYPE_DOUBLE, {.dbl=2.0}, 1.0, 50.0, FLAGS },
{ "c", "set the max compression factor", OFFSET(max_compression), AV_OPT_TYPE_DOUBLE, {.dbl=2.0}, 1.0, 50.0, FLAGS },
{ "threshold", "set the threshold value", OFFSET(threshold_value), AV_OPT_TYPE_DOUBLE, {.dbl=0}, 0.0, 1.0, FLAGS },
{ "t", "set the threshold value", OFFSET(threshold_value), AV_OPT_TYPE_DOUBLE, {.dbl=0}, 0.0, 1.0, FLAGS },
{ "raise", "set the expansion raising amount", OFFSET(raise_amount), AV_OPT_TYPE_DOUBLE, {.dbl=0.001}, 0.0, 1.0, FLAGS },
{ "r", "set the expansion raising amount", OFFSET(raise_amount), AV_OPT_TYPE_DOUBLE, {.dbl=0.001}, 0.0, 1.0, FLAGS },
{ "fall", "set the compression raising amount", OFFSET(fall_amount), AV_OPT_TYPE_DOUBLE, {.dbl=0.001}, 0.0, 1.0, FLAGS },
{ "f", "set the compression raising amount", OFFSET(fall_amount), AV_OPT_TYPE_DOUBLE, {.dbl=0.001}, 0.0, 1.0, FLAGS },
{ "channels", "set channels to filter", OFFSET(ch_layout_str), AV_OPT_TYPE_STRING, {.str="all"}, 0, 0, FLAGS },
{ "h", "set channels to filter", OFFSET(ch_layout_str), AV_OPT_TYPE_STRING, {.str="all"}, 0, 0, FLAGS },
{ "invert", "set inverted filtering", OFFSET(invert), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, FLAGS },
{ "i", "set inverted filtering", OFFSET(invert), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, FLAGS },
{ "link", "set linked channels filtering", OFFSET(link), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, FLAGS },
{ "l", "set linked channels filtering", OFFSET(link), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, FLAGS },
{ "rms", "set the RMS value", OFFSET(rms_value), AV_OPT_TYPE_DOUBLE, {.dbl=0.0}, 0.0, 1.0, FLAGS },
{ "m", "set the RMS value", OFFSET(rms_value), AV_OPT_TYPE_DOUBLE, {.dbl=0.0}, 0.0, 1.0, FLAGS },
{ NULL }
}
invert
static void invert(float *h, int n)
Definition asrc_sinc.c:186
FLAGS
#define FLAGS
Definition cmdutils.c:598
NULL
#define NULL
Definition coverity.c:32
OFFSET
#define OFFSET(x)
Definition ffmpeg_opt.c:1551
AV_OPT_TYPE_DOUBLE
@ AV_OPT_TYPE_DOUBLE
Underlying C type is double.
Definition opt.h:266
AV_OPT_TYPE_BOOL
@ AV_OPT_TYPE_BOOL
Underlying C type is int.
Definition opt.h:326
AV_OPT_TYPE_STRING
@ AV_OPT_TYPE_STRING
Underlying C type is a uint8_t* that is either NULL or points to a C string allocated with the av_mal...
Definition opt.h:275

Definition at line 97 of file af_speechnorm.c.

◆ inputs

const AVFilterPad inputs[]
static
Initial value:
= {
{
.name = "default",
.type = AVMEDIA_TYPE_AUDIO,
.config_props = config_input,
},
}
config_input
static int config_input(AVFilterLink *inlink)
Definition af_acontrast.c:121
AVMEDIA_TYPE_AUDIO
@ AVMEDIA_TYPE_AUDIO
Definition avutil.h:201

Definition at line 585 of file af_speechnorm.c.

◆ ff_af_speechnorm

const FFFilter ff_af_speechnorm
Initial value:
= {
.p.name = "speechnorm",
.p.description = NULL_IF_CONFIG_SMALL("Speech Normalizer."),
.p.priv_class = &speechnorm_class,
.p.flags = AVFILTER_FLAG_SUPPORT_TIMELINE_INTERNAL,
.priv_size = sizeof(SpeechNormalizerContext),
.activate = activate,
.uninit = uninit,
FILTER_INPUTS(inputs),
FILTER_OUTPUTS(ff_audio_default_filterpad),
FILTER_SAMPLEFMTS(AV_SAMPLE_FMT_FLTP, AV_SAMPLE_FMT_DBLP),
.process_command = process_command,
}
inputs
static const AVFilterPad inputs[]
Definition af_aap.c:299
process_command
static int process_command(AVFilterContext *ctx, const char *cmd, const char *args, char *res, int res_len, int flags)
Definition af_acrusher.c:307
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
AVFILTER_FLAG_SUPPORT_TIMELINE_INTERNAL
#define AVFILTER_FLAG_SUPPORT_TIMELINE_INTERNAL
Same as AVFILTER_FLAG_SUPPORT_TIMELINE_GENERIC, except that the filter will have its filter_frame() c...
Definition avfilter.h:204
AV_SAMPLE_FMT_FLTP
@ AV_SAMPLE_FMT_FLTP
float, planar
Definition samplefmt.h:66
AV_SAMPLE_FMT_DBLP
@ AV_SAMPLE_FMT_DBLP
double, planar
Definition samplefmt.h:67
uninit
static av_cold void uninit(AVBitStreamFilterContext *ctx)
Definition lcevc_merge.c:135
activate
static int activate(AVBitStreamFilterContext *ctx)
Definition lcevc_merge.c:144
FILTER_INPUTS
#define FILTER_INPUTS(array)
Definition filters.h:264
FILTER_OUTPUTS
#define FILTER_OUTPUTS(array)
Definition filters.h:265
FILTER_SAMPLEFMTS
#define FILTER_SAMPLEFMTS(...)
Definition filters.h:252
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:88

Definition at line 593 of file af_speechnorm.c.

Generated on Sun Aug 16 2026 20:21:11 for FFmpeg by doxygen 1.13.2