/*
* Copyright (c) 2018 The FFmpeg Project
*
* This file is part of FFmpeg.
*
* FFmpeg is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* FFmpeg is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include <float.h>
#include "libavutil/audio_fifo.h"
#include "libavutil/avstring.h"
#include "libavutil/channel_layout.h"
#include "libavutil/opt.h"
#include "libavcodec/avfft.h"
#include "avfilter.h"
#include "audio.h"
#include "formats.h"
#include "filters.h"
#define C (M_LN10 * 0.1)
#define RATIO 0.98
#define RRATIO (1.0 - RATIO)
enum OutModes {
IN_MODE,
OUT_MODE,
NOISE_MODE,
NB_MODES
};
enum NoiseType {
WHITE_NOISE,
VINYL_NOISE,
SHELLAC_NOISE,
CUSTOM_NOISE,
NB_NOISE
};
typedef struct DeNoiseChannel {
int band_noise[15];
double noise_band_auto_var[15];
double noise_band_sample[15];
double *amt;
double *band_amt;
double *band_excit;
double *gain;
double *prior;
double *prior_band_excit;
double *clean_data;
double *noisy_data;
double *out_samples;
double *spread_function;
double *abs_var;
double *rel_var;
double *min_abs_var;
FFTComplex *fft_data;
FFTContext *fft, *ifft;
double noise_band_norm[15];
double noise_band_avr[15];
double noise_band_avi[15];
double noise_band_var[15];
double sfm_threshold;
double sfm_alpha;
double sfm_results[3];
int sfm_fail_flags[512];
int sfm_fail_total;
} DeNoiseChannel;
typedef struct AudioFFTDeNoiseContext {
const AVClass *class;
float noise_reduction;
float noise_floor;
int noise_type;
char *band_noise_str;
float residual_floor;
int track_noise;
int track_residual;
int output_mode;
float last_residual_floor;
float last_noise_floor;
float last_noise_reduction;
float last_noise_balance;
int64_t block_count;
int64_t pts;
int channels;
int sample_noise;
int sample_noise_start;
int sample_noise_end;
float sample_rate;
int buffer_length;
int fft_length;
int fft_length2;
int bin_count;
int window_length;
int sample_advance;
int number_of_bands;
int band_centre[15];
int *bin2band;
double *window;
double *band_alpha;
double *band_beta;
DeNoiseChannel *dnch;
double max_gain;
double max_var;
double gain_scale;
double window_weight;
double floor;
double sample_floor;
double auto_floor;
int noise_band_edge[17];
int noise_band_count;
double matrix_a[25];
double vector_b[5];
double matrix_b[75];
double matrix_c[75];
AVAudioFifo *fifo;
} AudioFFTDeNoiseContext;
#define OFFSET(x) offsetof(AudioFFTDeNoiseContext, x)
#define AF AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
#define AFR AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_RUNTIME_PARAM
static const AVOption afftdn_options[] = {
{ "nr", "set the noise reduction", OFFSET(noise_reduction), AV_OPT_TYPE_FLOAT, {.dbl = 12}, .01, 97, AFR },
{ "nf", "set the noise floor", OFFSET(noise_floor), AV_OPT_TYPE_FLOAT, {.dbl =-50}, -80,-20, AFR },
{ "nt", "set the noise type", OFFSET(noise_type), AV_OPT_TYPE_INT, {.i64 = WHITE_NOISE}, WHITE_NOISE, NB_NOISE-1, AF, "type" },
{ "w", "white noise", 0, AV_OPT_TYPE_CONST, {.i64 = WHITE_NOISE}, 0, 0, AF, "type" },
{ "v", "vinyl noise", 0, AV_OPT_TYPE_CONST, {.i64 = VINYL_NOISE}, 0, 0, AF, "type" },
{ "s", "shellac noise", 0, AV_OPT_TYPE_CONST, {.i64 = SHELLAC_NOISE}, 0, 0, AF, "type" },
{ "c", "custom noise", 0, AV_OPT_TYPE_CONST, {.i64 = CUSTOM_NOISE}, 0, 0, AF, "type" },
{ "bn", "set the custom bands noise", OFFSET(band_noise_str), AV_OPT_TYPE_STRING, {.str = 0}, 0, 0, AF },
{ "rf", "set the residual floor", OFFSET(residual_floor), AV_OPT_TYPE_FLOAT, {.dbl =-38}, -80,-20, AFR },
{ "tn", "track noise", OFFSET(track_noise), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, AFR },
{ "tr", "track residual", OFFSET(track_residual), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, AFR },
{ "om", "set output mode", OFFSET(output_mode), AV_OPT_TYPE_INT, {.i64 = OUT_MODE}, 0, NB_MODES-1, AFR, "mode" },
{ "i", "input", 0, AV_OPT_TYPE_CONST, {.i64 = IN_MODE}, 0, 0, AFR, "mode" },
{ "o", "output", 0, AV_OPT_TYPE_CONST, {.i64 = OUT_MODE}, 0, 0, AFR, "mode" },
{ "n", "noise", 0, AV_OPT_TYPE_CONST, {.i64 = NOISE_MODE}, 0, 0, AFR, "mode" },
{ NULL }
};
AVFILTER_DEFINE_CLASS(afftdn);
static int get_band_noise(AudioFFTDeNoiseContext *s,
int band, double a,
double b, double c)
{
double d1, d2, d3;
d1 = a / s->band_centre[band];
d1 = 10.0 * log(1.0 + d1 * d1) / M_LN10;
d2 = b / s->band_centre[band];
d2 = 10.0 * log(1.0 + d2 * d2) / M_LN10;
d3 = s->band_centre[band] / c;
d3 = 10.0 * log(1.0 + d3 * d3) / M_LN10;
return lrint(-d1 + d2 - d3);
}
static void factor(double *array, int size)
{
for (int i = 0; i < size - 1; i++) {
for (int j = i + 1; j < size; j++) {
double d = array[j + i * size] / array[i + i * size];
array[j + i * size] = d;
for (int k = i + 1; k < size; k++) {
array[j + k * size] -= d * array[i + k * size];
}
}
}
}
static void solve(double *matrix, double *vector, int size)
{
for (int i = 0; i < size - 1; i++) {
for (int j = i + 1; j < size; j++) {
double d = matrix[j + i * size];
vector[j] -= d * vector[i];
}
}
vector[size - 1] /= matrix[size * size - 1];
for (int i = size - 2; i >= 0; i--) {
double d = vector[i];
for (int j = i + 1; j < size; j++)
d -= matrix[i + j * size] * vector[j];
vector[i] = d / matrix[i + i * size];
}
}
static int process_get_band_noise(AudioFFTDeNoiseContext *s,
DeNoiseChannel *dnch,
int band)
{
double product, sum, f;
int i = 0;
if (band < 15)
return dnch->band_noise[band];
for (int j = 0; j < 5; j++) {
sum = 0.0;
for (int k = 0; k < 15; k++)
sum += s->matrix_b[i++] * dnch->band_noise[k];
s->vector_b[j] = sum;
}
solve(s->matrix_a, s->vector_b, 5);
f = (0.5 * s->sample_rate) / s->band_centre[14];
f = 15.0 + log(f / 1.5) / log(1.5);
sum = 0.0;
product = 1.0;
for (int j = 0; j < 5; j++) {
sum += product * s->vector_b[j];
product *= f;
}
return lrint(sum);
}
static void calculate_sfm(AudioFFTDeNoiseContext *s,
DeNoiseChannel *dnch,
int start, int end)
{
double d1 = 0.0, d2 = 1.0;
int i = 0, j = 0;
for (int k = start; k < end; k++) {
if (dnch->noisy_data[k] > s->sample_floor) {
j++;
d1 += dnch->noisy_data[k];
d2 *= dnch->noisy_data[k];
if (d2 > 1.0E100) {
d2 *= 1.0E-100;
i++;
} else if (d2 < 1.0E-100) {
d2 *= 1.0E100;
i--;
}
}
}
if (j > 1) {
d1 /= j;
dnch->sfm_results[0] = d1;
d2 = log(d2) + 230.2585 * i;
d2 /= j;
d1 = log(d1);
dnch->sfm_results[1] = d1;
dnch->sfm_results[2] = d1 - d2;
} else {
dnch->sfm_results[0] = s->auto_floor;
dnch->sfm_results[1] = dnch->sfm_threshold;
dnch->sfm_results[2] = dnch->sfm_threshold;
}
}
static double limit_gain(double a, double b)
{
if (a > 1.0)
return (b * a - 1.0) / (b + a - 2.0);
if (a < 1.0)
return (b * a - 2.0 * a + 1.0) / (b - a);
return 1.0;
}
static void process_frame(AudioFFTDeNoiseContext *s, DeNoiseChannel *dnch,
FFTComplex *fft_data,
double *prior, double *prior_band_excit, int track_noise)
{
double d1, d2, d3, gain;
int n, i1;
d1 = fft_data[0].re * fft_data[0].re;
dnch->noisy_data[0] = d1;
d2 = d1 / dnch->abs_var[0];
d3 = RATIO * prior[0] + RRATIO * fmax(d2 - 1.0, 0.0);
gain = d3 / (1.0 + d3);
gain *= (gain + M_PI_4 / fmax(d2, 1.0E-6));
prior[0] = (d2 * gain);
dnch->clean_data[0] = (d1 * gain);
gain = sqrt(gain);
dnch->gain[0] = gain;
n = 0;
for (int i = 1; i < s->fft_length2; i++) {
d1 = fft_data[i].re * fft_data[i].re + fft_data[i].im * fft_data[i].im;
if (d1 > s->sample_floor)
n = i;
dnch->noisy_data[i] = d1;
d2 = d1 / dnch->abs_var[i];
d3 = RATIO * prior[i] + RRATIO * fmax(d2 - 1.0, 0.0);
gain = d3 / (1.0 + d3);
gain *= (gain + M_PI_4 / fmax(d2, 1.0E-6));
prior[i] = d2 * gain;
dnch->clean_data[i] = d1 * gain;
gain = sqrt(gain);
dnch->gain[i] = gain;
}
d1 = fft_data[0].im * fft_data[0].im;
if (d1 > s->sample_floor)
n = s->fft_length2;
dnch->noisy_data[s->fft_length2] = d1;
d2 = d1 / dnch->abs_var[s->fft_length2];
d3 = RATIO * prior[s->fft_length2] + RRATIO * fmax(d2 - 1.0, 0.0);
gain = d3 / (1.0 + d3);
gain *= gain + M_PI_4 / fmax(d2, 1.0E-6);
prior[s->fft_length2] = d2 * gain;
dnch->clean_data[s->fft_length2] = d1 * gain;
gain = sqrt(gain);
dnch->gain[s->fft_length2] = gain;
if (n > s->fft_length2 - 2) {
n = s->bin_count;
i1 = s->noise_band_count;
} else {
i1 = 0;
for (int i = 0; i <= s->noise_band_count; i++) {
if (n > 1.1 * s->noise_band_edge[i]) {
i1 = i;
}
}
}
if (track_noise && (i1 > s->noise_band_count / 2)) {
int j = FFMIN(n, s->noise_band_edge[i1]);
int m = 3, k;
for (k = i1 - 1; k >= 0; k--) {
int i = s->noise_band_edge[k];
calculate_sfm(s, dnch, i, j);
dnch->noise_band_sample[k] = dnch->sfm_results[0];
if (dnch->sfm_results[2] + 0.013 * m * fmax(0.0, dnch->sfm_results[1] - 20.53) >= dnch->sfm_threshold) {
break;
}
j = i;
m++;
}
if (k < i1 - 1) {
double sum = 0.0, min, max;
int i;
for (i = i1 - 1; i > k; i--) {
min = log(dnch->noise_band_sample[i] / dnch->noise_band_auto_var[i]);
sum += min;
}
i = i1 - k - 1;
if (i < 5) {
min = 3.0E-4 * i * i;
} else {
min = 3.0E-4 * (8 * i - 16);
}
if (i < 3) {
max = 2.0E-4 * i * i;
} else {
max = 2.0E-4 * (4 * i - 4);
}
if (s->track_residual) {
if (s->last_noise_floor > s->last_residual_floor + 9) {
min *= 0.5;
max *= 0.75;
} else if (s->last_noise_floor > s->last_residual_floor + 6) {
min *= 0.4;
max *= 1.0;
} else if (s->last_noise_floor > s->last_residual_floor + 4) {
min *= 0.3;
max *= 1.3;
} else if (s->last_noise_floor > s->last_residual_floor + 2) {
min *= 0.2;
max *= 1.6;
} else if (s->last_noise_floor > s->last_residual_floor) {
min *= 0.1;
max *= 2.0;
} else {
min = 0.0;
max *= 2.5;
}
}
sum = av_clipd(sum, -min, max);
sum = exp(sum);
for (int i = 0; i < 15; i++)
dnch->noise_band_auto_var[i] *= sum;
} else if (dnch->sfm_results[2] >= dnch->sfm_threshold) {
dnch->sfm_fail_flags[s->block_count & 0x1FF] = 1;
dnch->sfm_fail_total += 1;
}
}
for (int i = 0; i < s->number_of_bands; i++) {
dnch->band_excit[i] = 0.0;
dnch->band_amt[i] = 0.0;
}
for (int i = 0; i < s->bin_count; i++) {
dnch->band_excit[s->bin2band[i]] += dnch->clean_data[i];
}
for (int i = 0; i < s->number_of_bands; i++) {
dnch->band_excit[i] = fmax(dnch->band_excit[i],
s->band_alpha[i] * dnch->band_excit[i] +
s->band_beta[i] * prior_band_excit[i]);
prior_band_excit[i] = dnch->band_excit[i];
}
for (int j = 0, i = 0; j < s->number_of_bands; j++) {
for (int k = 0; k < s->number_of_bands; k++) {
dnch->band_amt[j] += dnch->spread_function[i++] * dnch->band_excit[k];
}
}
for (int i = 0; i < s->bin_count; i++)
dnch->amt[i] = dnch->band_amt[s->bin2band[i]];
if (dnch->amt[0] > dnch->abs_var[0]) {
dnch->gain[0] = 1.0;
} else if (dnch->amt[0] > dnch->min_abs_var[0]) {
double limit = sqrt(dnch->abs_var[0] / dnch->amt[0]);
dnch->gain[0] = limit_gain(dnch->gain[0], limit);
} else {
dnch->gain[0] = limit_gain(dnch->gain[0], s->max_gain);
}
if (dnch->amt[s->fft_length2] > dnch->abs_var[s->fft_length2]) {
dnch->gain[s->fft_length2] = 1.0;
} else if (dnch->amt[s->fft_length2] > dnch->min_abs_var[s->fft_length2]) {
double limit = sqrt(dnch->abs_var[s->fft_length2] / dnch->amt[s->fft_length2]);
dnch->gain[s->fft_length2] = limit_gain(dnch->gain[s->fft_length2], limit);
} else {
dnch->gain[s->fft_length2] = limit_gain(dnch->gain[s->fft_length2], s->max_gain);
}
for (int i = 1; i < s->fft_length2; i++) {
if (dnch->amt[i] > dnch->abs_var[i]) {
dnch->gain[i] = 1.0;
} else if (dnch->amt[i] > dnch->min_abs_var[i]) {
double limit = sqrt(dnch->abs_var[i] / dnch->amt[i]);
dnch->gain[i] = limit_gain(dnch->gain[i], limit);
} else {
dnch->gain[i] = limit_gain(dnch->gain[i], s->max_gain);
}
}
gain = dnch->gain[0];
dnch->clean_data[0] = (gain * gain * dnch->noisy_data[0]);
fft_data[0].re *= gain;
gain = dnch->gain[s->fft_length2];
dnch->clean_data[s->fft_length2] = (gain * gain * dnch->noisy_data[s->fft_length2]);
fft_data[0].im *= gain;
for (int i = 1; i < s->fft_length2; i++) {
gain = dnch->gain[i];
dnch->clean_data[i] = (gain * gain * dnch->noisy_data[i]);
fft_data[i].re *= gain;
fft_data[i].im *= gain;
}
}
static double freq2bark(double x)
{
double d = x / 7500.0;
return 13.0 * atan(7.6E-4 * x) + 3.5 * atan(d * d);
}
static int get_band_centre(AudioFFTDeNoiseContext *s, int band)
{
if (band == -1)
return lrint(s->band_centre[0] / 1.5);
return s->band_centre[band];
}
static int get_band_edge(AudioFFTDeNoiseContext *s, int band)
{
int i;
if (band == 15) {
i = lrint(s->band_centre[14] * 1.224745);
} else {
i = lrint(s->band_centre[band] / 1.224745);
}
return FFMIN(i, s->sample_rate / 2);
}
static void set_band_parameters(AudioFFTDeNoiseContext *s,
DeNoiseChannel *dnch)
{
double band_noise, d2, d3, d4, d5;
int i = 0, j = 0, k = 0;
d5 = 0.0;
band_noise = process_get_band_noise(s, dnch, 0);
for (int m = j; m <= s->fft_length2; m++) {
if (m == j) {
i = j;
d5 = band_noise;
if (k == 15) {
j = s->bin_count;
} else {
j = s->fft_length * get_band_centre(s, k) / s->sample_rate;
}
d2 = j - i;
band_noise = process_get_band_noise(s, dnch, k);
k++;
}
d3 = (j - m) / d2;
d4 = (m - i) / d2;
dnch->rel_var[m] = exp((d5 * d3 + band_noise * d4) * C);
}
dnch->rel_var[s->fft_length2] = exp(band_noise * C);
for (i = 0; i < 15; i++)
dnch->noise_band_auto_var[i] = s->max_var * exp((process_get_band_noise(s, dnch, i) - 2.0) * C);
for (i = 0; i <= s->fft_length2; i++) {
dnch->abs_var[i] = fmax(s->max_var * dnch->rel_var[i], 1.0);
dnch->min_abs_var[i] = s->gain_scale * dnch->abs_var[i];
}
}
static void read_custom_noise(AudioFFTDeNoiseContext *s, int ch)
{
DeNoiseChannel *dnch = &s->dnch[ch];
char *p, *arg, *saveptr = NULL;
int i, ret, band_noise[15] = { 0 };
if (!s->band_noise_str)
return;
p = av_strdup(s->band_noise_str);
if (!p)
return;
for (i = 0; i < 15; i++) {
if (!(arg = av_strtok(p, "| ", &saveptr)))
break;
p = NULL;
ret = av_sscanf(arg, "%d", &band_noise[i]);
if (ret != 1) {
av_log(s, AV_LOG_ERROR, "Custom band noise must be integer.\n");
break;
}
band_noise[i] = av_clip(band_noise[i], -24, 24);
}
av_free(p);
memcpy(dnch->band_noise, band_noise, sizeof(band_noise));
}
static void set_parameters(AudioFFTDeNoiseContext *s)
{
if (s->last_noise_floor != s->noise_floor)
s->last_noise_floor = s->noise_floor;
if (s->track_residual)
s->last_noise_floor = fmaxf(s->last_noise_floor, s->residual_floor);
s->max_var = s->floor * exp((100.0 + s->last_noise_floor) * C);
if (s->track_residual) {
s->last_residual_floor = s->residual_floor;
s->last_noise_reduction = fmax(s->last_noise_floor - s->last_residual_floor, 0);
s->max_gain = exp(s->last_noise_reduction * (0.5 * C));
} else if (s->noise_reduction != s->last_noise_reduction) {
s->last_noise_reduction = s->noise_reduction;
s->last_residual_floor = av_clipf(s->last_noise_floor - s->last_noise_reduction, -80, -20);
s->max_gain = exp(s->last_noise_reduction * (0.5 * C));
}
s->gain_scale = 1.0 / (s->max_gain * s->max_gain);
for (int ch = 0; ch < s->channels; ch++) {
DeNoiseChannel *dnch = &s->dnch[ch];
set_band_parameters(s, dnch);
}
}
static int config_input(AVFilterLink *inlink)
{
AVFilterContext *ctx = inlink->dst;
AudioFFTDeNoiseContext *s = ctx->priv;
double wscale, sar, sum, sdiv;
int i, j, k, m, n;
s->dnch = av_calloc(inlink->channels, sizeof(*s->dnch));
if (!s->dnch)
return AVERROR(ENOMEM);
s->pts = AV_NOPTS_VALUE;
s->channels = inlink->channels;
s->sample_rate = inlink->sample_rate;
s->sample_advance = s->sample_rate / 80;
s->window_length = 3 * s->sample_advance;
s->fft_length2 = 1 << (32 - ff_clz(s->window_length));
s->fft_length = s->fft_length2 * 2;
s->buffer_length = s->fft_length * 2;
s->bin_count = s->fft_length2 + 1;
s->band_centre[0] = 80;
for (i = 1; i < 15; i++) {
s->band_centre[i] = lrint(1.5 * s->band_centre[i - 1] + 5.0);
if (s->band_centre[i] < 1000) {
s->band_centre[i] = 10 * (s->band_centre[i] / 10);
} else if (s->band_centre[i] < 5000) {
s->band_centre[i] = 50 * ((s->band_centre[i] + 20) / 50);
} else if (s->band_centre[i] < 15000) {
s->band_centre[i] = 100 * ((s->band_centre[i] + 45) / 100);
} else {
s->band_centre[i] = 1000 * ((s->band_centre[i] + 495) / 1000);
}
}
for (j = 0; j < 5; j++) {
for (k = 0; k < 5; k++) {
s->matrix_a[j + k * 5] = 0.0;
for (m = 0; m < 15; m++)
s->matrix_a[j + k * 5] += pow(m, j + k);
}
}
factor(s->matrix_a, 5);
i = 0;
for (j = 0; j < 5; j++)
for (k = 0; k < 15; k++)
s->matrix_b[i++] = pow(k, j);
i = 0;
for (j = 0; j < 15; j++)
for (k = 0; k < 5; k++)
s->matrix_c[i++] = pow(j, k);
s->window = av_calloc(s->window_length, sizeof(*s->window));
s->bin2band = av_calloc(s->bin_count, sizeof(*s->bin2band));
if (!s->window || !s->bin2band)
return AVERROR(ENOMEM);
sdiv = s->sample_rate / 17640.0;
for (i = 0; i <= s->fft_length2; i++)
s->bin2band[i] = lrint(sdiv * freq2bark((0.5 * i * s->sample_rate) / s->fft_length2));
s->number_of_bands = s->bin2band[s->fft_length2] + 1;
s->band_alpha = av_calloc(s->number_of_bands, sizeof(*s->band_alpha));
s->band_beta = av_calloc(s->number_of_bands, sizeof(*s->band_beta));
if (!s->band_alpha || !s->band_beta)
return AVERROR(ENOMEM);
for (int ch = 0; ch < inlink->channels; ch++) {
DeNoiseChannel *dnch = &s->dnch[ch];
switch (s->noise_type) {
case WHITE_NOISE:
for (i = 0; i < 15; i++)
dnch->band_noise[i] = 0;
break;
case VINYL_NOISE:
for (i = 0; i < 15; i++)
dnch->band_noise[i] = get_band_noise(s, i, 50.0, 500.5, 2125.0) + FFMAX(i - 7, 0);
break;
case SHELLAC_NOISE:
for (i = 0; i < 15; i++)
dnch->band_noise[i] = get_band_noise(s, i, 1.0, 500.0, 1.0E10) + FFMAX(i - 12, -5);
break;
case CUSTOM_NOISE:
read_custom_noise(s, ch);
break;
default:
return AVERROR_BUG;
}
dnch->sfm_threshold = 0.8;
dnch->sfm_alpha = 0.05;
for (i = 0; i < 512; i++)
dnch->sfm_fail_flags[i] = 0;
dnch->sfm_fail_total = 0;
j = FFMAX((int)(10.0 * (1.3 - dnch->sfm_threshold)), 1);
for (i = 0; i < 512; i += j) {
dnch->sfm_fail_flags[i] = 1;
dnch->sfm_fail_total += 1;
}
dnch->amt = av_calloc(s->bin_count, sizeof(*dnch->amt));
dnch->band_amt = av_calloc(s->number_of_bands, sizeof(*dnch->band_amt));
dnch->band_excit = av_calloc(s->number_of_bands, sizeof(*dnch->band_excit));
dnch->gain = av_calloc(s->bin_count, sizeof(*dnch->gain));
dnch->prior = av_calloc(s->bin_count, sizeof(*dnch->prior));
dnch->prior_band_excit = av_calloc(s->number_of_bands, sizeof(*dnch->prior_band_excit));
dnch->clean_data = av_calloc(s->bin_count, sizeof(*dnch->clean_data));
dnch->noisy_data = av_calloc(s->bin_count, sizeof(*dnch->noisy_data));
dnch->out_samples = av_calloc(s->buffer_length, sizeof(*dnch->out_samples));
dnch->abs_var = av_calloc(s->bin_count, sizeof(*dnch->abs_var));
dnch->rel_var = av_calloc(s->bin_count, sizeof(*dnch->rel_var));
dnch->min_abs_var = av_calloc(s->bin_count, sizeof(*dnch->min_abs_var));
dnch->fft_data = av_calloc(s->fft_length2 + 1, sizeof(*dnch->fft_data));
dnch->fft = av_fft_init(av_log2(s->fft_length2), 0);
dnch->ifft = av_fft_init(av_log2(s->fft_length2), 1);
dnch->spread_function = av_calloc(s->number_of_bands * s->number_of_bands,
sizeof(*dnch->spread_function));
if (!dnch->amt ||
!dnch->band_amt ||
!dnch->band_excit ||
!dnch->gain ||
!dnch->prior ||
!dnch->prior_band_excit ||
!dnch->clean_data ||
!dnch->noisy_data ||
!dnch->out_samples ||
!dnch->fft_data ||
!dnch->abs_var ||
!dnch->rel_var ||
!dnch->min_abs_var ||
!dnch->spread_function ||
!dnch->fft ||
!dnch->ifft)
return AVERROR(ENOMEM);
}
for (int ch = 0; ch < inlink->channels; ch++) {
DeNoiseChannel *dnch = &s->dnch[ch];
double *prior_band_excit = dnch->prior_band_excit;
double *prior = dnch->prior;
double min, max;
double p1, p2;
p1 = pow(0.1, 2.5 / sdiv);
p2 = pow(0.1, 1.0 / sdiv);
j = 0;
for (m = 0; m < s->number_of_bands; m++) {
for (n = 0; n < s->number_of_bands; n++) {
if (n < m) {
dnch->spread_function[j++] = pow(p2, m - n);
} else if (n > m) {
dnch->spread_function[j++] = pow(p1, n - m);
} else {
dnch->spread_function[j++] = 1.0;
}
}
}
for (m = 0; m < s->number_of_bands; m++) {
dnch->band_excit[m] = 0.0;
prior_band_excit[m] = 0.0;
}
for (m = 0; m <= s->fft_length2; m++)
dnch->band_excit[s->bin2band[m]] += 1.0;
j = 0;
for (m = 0; m < s->number_of_bands; m++) {
for (n = 0; n < s->number_of_bands; n++)
prior_band_excit[m] += dnch->spread_function[j++] * dnch->band_excit[n];
}
min = pow(0.1, 2.5);
max = pow(0.1, 1.0);
for (int i = 0; i < s->number_of_bands; i++) {
if (i < lrint(12.0 * sdiv)) {
dnch->band_excit[i] = pow(0.1, 1.45 + 0.1 * i / sdiv);
} else {
dnch->band_excit[i] = pow(0.1, 2.5 - 0.2 * (i / sdiv - 14.0));
}
dnch->band_excit[i] = av_clipd(dnch->band_excit[i], min, max);
}
for (int i = 0; i <= s->fft_length2; i++)
prior[i] = RRATIO;
for (int i = 0; i < s->buffer_length; i++)
dnch->out_samples[i] = 0;
j = 0;
for (int i = 0; i < s->number_of_bands; i++)
for (int k = 0; k < s->number_of_bands; k++)
dnch->spread_function[j++] *= dnch->band_excit[i] / prior_band_excit[i];
}
j = 0;
sar = s->sample_advance / s->sample_rate;
for (int i = 0; i <= s->fft_length2; i++) {
if ((i == s->fft_length2) || (s->bin2band[i] > j)) {
double d6 = (i - 1) * s->sample_rate / s->fft_length;
double d7 = fmin(0.008 + 2.2 / d6, 0.03);
s->band_alpha[j] = exp(-sar / d7);
s->band_beta[j] = 1.0 - s->band_alpha[j];
j = s->bin2band[i];
}
}
wscale = sqrt(16.0 / (9.0 * s->fft_length));
sum = 0.0;
for (int i = 0; i < s->window_length; i++) {
double d10 = sin(i * M_PI / s->window_length);
d10 *= wscale * d10;
s->window[i] = d10;
sum += d10 * d10;
}
s->window_weight = 0.5 * sum;
s->floor = (1LL << 48) * exp(-23.025558369790467) * s->window_weight;
s->sample_floor = s->floor * exp(4.144600506562284);
s->auto_floor = s->floor * exp(6.907667510937141);
set_parameters(s);
s->noise_band_edge[0] = FFMIN(s->fft_length2, s->fft_length * get_band_edge(s, 0) / s->sample_rate);
i = 0;
for (int j = 1; j < 16; j++) {
s->noise_band_edge[j] = FFMIN(s->fft_length2, s->fft_length * get_band_edge(s, j) / s->sample_rate);
if (s->noise_band_edge[j] > lrint(1.1 * s->noise_band_edge[j - 1]))
i++;
s->noise_band_edge[16] = i;
}
s->noise_band_count = s->noise_band_edge[16];
s->fifo = av_audio_fifo_alloc(inlink->format, inlink->channels, s->fft_length);
if (!s->fifo)
return AVERROR(ENOMEM);
return 0;
}
static void preprocess(FFTComplex *in, int len)
{
double d1, d2, d3, d4, d5, d6, d7, d8, d9, d10;
int n, i, k;
d5 = 2.0 * M_PI / len;
d8 = sin(0.5 * d5);
d8 = -2.0 * d8 * d8;
d7 = sin(d5);
d9 = 1.0 + d8;
d6 = d7;
n = len / 2;
for (i = 1; i < len / 4; i++) {
k = n - i;
d2 = 0.5 * (in[i].re + in[k].re);
d1 = 0.5 * (in[i].im - in[k].im);
d4 = 0.5 * (in[i].im + in[k].im);
d3 = 0.5 * (in[k].re - in[i].re);
in[i].re = d2 + d9 * d4 + d6 * d3;
in[i].im = d1 + d9 * d3 - d6 * d4;
in[k].re = d2 - d9 * d4 - d6 * d3;
in[k].im = -d1 + d9 * d3 - d6 * d4;
d10 = d9;
d9 += d9 * d8 - d6 * d7;
d6 += d6 * d8 + d10 * d7;
}
d2 = in[0].re;
in[0].re = d2 + in[0].im;
in[0].im = d2 - in[0].im;
}
static void postprocess(FFTComplex *in, int len)
{
double d1, d2, d3, d4, d5, d6, d7, d8, d9, d10;
int n, i, k;
d5 = 2.0 * M_PI / len;
d8 = sin(0.5 * d5);
d8 = -2.0 * d8 * d8;
d7 = sin(d5);
d9 = 1.0 + d8;
d6 = d7;
n = len / 2;
for (i = 1; i < len / 4; i++) {
k = n - i;
d2 = 0.5 * (in[i].re + in[k].re);
d1 = 0.5 * (in[i].im - in[k].im);
d4 = 0.5 * (in[i].re - in[k].re);
d3 = 0.5 * (in[i].im + in[k].im);
in[i].re = d2 - d9 * d3 - d6 * d4;
in[i].im = d1 + d9 * d4 - d6 * d3;
in[k].re = d2 + d9 * d3 + d6 * d4;
in[k].im = -d1 + d9 * d4 - d6 * d3;
d10 = d9;
d9 += d9 * d8 - d6 * d7;
d6 += d6 * d8 + d10 * d7;
}
d2 = in[0].re;
in[0].re = 0.5 * (d2 + in[0].im);
in[0].im = 0.5 * (d2 - in[0].im);
}
static void init_sample_noise(DeNoiseChannel *dnch)
{
for (int i = 0; i < 15; i++) {
dnch->noise_band_norm[i] = 0.0;
dnch->noise_band_avr[i] = 0.0;
dnch->noise_band_avi[i] = 0.0;
dnch->noise_band_var[i] = 0.0;
}
}
static void sample_noise_block(AudioFFTDeNoiseContext *s,
DeNoiseChannel *dnch,
AVFrame *in, int ch)
{
float *src = (float *)in->extended_data[ch];
double mag2, var = 0.0, avr = 0.0, avi = 0.0;
int edge, j, k, n, edgemax;
for (int i = 0; i < s->window_length; i++) {
dnch->fft_data[i].re = s->window[i] * src[i] * (1LL << 24);
dnch->fft_data[i].im = 0.0;
}
for (int i = s->window_length; i < s->fft_length2; i++) {
dnch->fft_data[i].re = 0.0;
dnch->fft_data[i].im = 0.0;
}
av_fft_permute(dnch->fft, dnch->fft_data);
av_fft_calc(dnch->fft, dnch->fft_data);
preprocess(dnch->fft_data, s->fft_length);
edge = s->noise_band_edge[0];
j = edge;
k = 0;
n = j;
edgemax = fmin(s->fft_length2, s->noise_band_edge[15]);
dnch->fft_data[s->fft_length2].re = dnch->fft_data[0].im;
dnch->fft_data[0].im = 0.0;
dnch->fft_data[s->fft_length2].im = 0.0;
for (int i = j; i <= edgemax; i++) {
if ((i == j) && (i < edgemax)) {
if (j > edge) {
dnch->noise_band_norm[k - 1] += j - edge;
dnch->noise_band_avr[k - 1] += avr;
dnch->noise_band_avi[k - 1] += avi;
dnch->noise_band_var[k - 1] += var;
}
k++;
edge = j;
j = s->noise_band_edge[k];
if (k == 15) {
j++;
}
var = 0.0;
avr = 0.0;
avi = 0.0;
}
avr += dnch->fft_data[n].re;
avi += dnch->fft_data[n].im;
mag2 = dnch->fft_data[n].re * dnch->fft_data[n].re +
dnch->fft_data[n].im * dnch->fft_data[n].im;
mag2 = fmax(mag2, s->sample_floor);
dnch->noisy_data[i] = mag2;
var += mag2;
n++;
}
dnch->noise_band_norm[k - 1] += j - edge;
dnch->noise_band_avr[k - 1] += avr;
dnch->noise_band_avi[k - 1] += avi;
dnch->noise_band_var[k - 1] += var;
}
static void finish_sample_noise(AudioFFTDeNoiseContext *s,
DeNoiseChannel *dnch,
double *sample_noise)
{
for (int i = 0; i < s->noise_band_count; i++) {
dnch->noise_band_avr[i] /= dnch->noise_band_norm[i];
dnch->noise_band_avi[i] /= dnch->noise_band_norm[i];
dnch->noise_band_var[i] /= dnch->noise_band_norm[i];
dnch->noise_band_var[i] -= dnch->noise_band_avr[i] * dnch->noise_band_avr[i] +
dnch->noise_band_avi[i] * dnch->noise_band_avi[i];
dnch->noise_band_auto_var[i] = dnch->noise_band_var[i];
sample_noise[i] = (1.0 / C) * log(dnch->noise_band_var[i] / s->floor) - 100.0;
}
if (s->noise_band_count < 15) {
for (int i = s->noise_band_count; i < 15; i++)
sample_noise[i] = sample_noise[i - 1];
}
}
static void set_noise_profile(AudioFFTDeNoiseContext *s,
DeNoiseChannel *dnch,
double *sample_noise,
int new_profile)
{
int new_band_noise[15];
double temp[15];
double sum = 0.0, d1;
float new_noise_floor;
int i, n;
for (int m = 0; m < 15; m++)
temp[m] = sample_noise[m];
if (new_profile) {
i = 0;
for (int m = 0; m < 5; m++) {
sum = 0.0;
for (n = 0; n < 15; n++)
sum += s->matrix_b[i++] * temp[n];
s->vector_b[m] = sum;
}
solve(s->matrix_a, s->vector_b, 5);
i = 0;
for (int m = 0; m < 15; m++) {
sum = 0.0;
for (n = 0; n < 5; n++)
sum += s->matrix_c[i++] * s->vector_b[n];
temp[m] = sum;
}
}
sum = 0.0;
for (int m = 0; m < 15; m++)
sum += temp[m];
d1 = (int)(sum / 15.0 - 0.5);
if (!new_profile)
i = lrint(temp[7] - d1);
for (d1 -= dnch->band_noise[7] - i; d1 > -20.0; d1 -= 1.0)
;
for (int m = 0; m < 15; m++)
temp[m] -= d1;
new_noise_floor = d1 + 2.5;
if (new_profile) {
av_log(s, AV_LOG_INFO, "bn=");
for (int m = 0; m < 15; m++) {
new_band_noise[m] = lrint(temp[m]);
new_band_noise[m] = av_clip(new_band_noise[m], -24, 24);
av_log(s, AV_LOG_INFO, "%d ", new_band_noise[m]);
}
av_log(s, AV_LOG_INFO, "\n");
memcpy(dnch->band_noise, new_band_noise, sizeof(new_band_noise));
}
if (s->track_noise)
s->noise_floor = new_noise_floor;
}
typedef struct ThreadData {
AVFrame *in;
} ThreadData;
static int filter_channel(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
{
AudioFFTDeNoiseContext *s = ctx->priv;
ThreadData *td = arg;
AVFrame *in = td->in;
const int start = (in->channels * jobnr) / nb_jobs;
const int end = (in->channels * (jobnr+1)) / nb_jobs;
for (int ch = start; ch < end; ch++) {
DeNoiseChannel *dnch = &s->dnch[ch];
const float *src = (const float *)in->extended_data[ch];
double *dst = dnch->out_samples;
if (s->track_noise) {
int i = s->block_count & 0x1FF;
if (dnch->sfm_fail_flags[i])
dnch->sfm_fail_total--;
dnch->sfm_fail_flags[i] = 0;
dnch->sfm_threshold *= 1.0 - dnch->sfm_alpha;
dnch->sfm_threshold += dnch->sfm_alpha * (0.5 + (1.0 / 640) * dnch->sfm_fail_total);
}
for (int m = 0; m < s->window_length; m++) {
dnch->fft_data[m].re = s->window[m] * src[m] * (1LL << 24);
dnch->fft_data[m].im = 0;
}
for (int m = s->window_length; m < s->fft_length2; m++) {
dnch->fft_data[m].re = 0;
dnch->fft_data[m].im = 0;
}
av_fft_permute(dnch->fft, dnch->fft_data);
av_fft_calc(dnch->fft, dnch->fft_data);
preprocess(dnch->fft_data, s->fft_length);
process_frame(s, dnch, dnch->fft_data,
dnch->prior,
dnch->prior_band_excit,
s->track_noise);
postprocess(dnch->fft_data, s->fft_length);
av_fft_permute(dnch->ifft, dnch->fft_data);
av_fft_calc(dnch->ifft, dnch->fft_data);
for (int m = 0; m < s->window_length; m++)
dst[m] += s->window[m] * dnch->fft_data[m].re / (1LL << 24);
}
return 0;
}
static void get_auto_noise_levels(AudioFFTDeNoiseContext *s,
DeNoiseChannel *dnch,
double *levels)
{
if (s->noise_band_count > 0) {
for (int i = 0; i < s->noise_band_count; i++) {
levels[i] = (1.0 / C) * log(dnch->noise_band_auto_var[i] / s->floor) - 100.0;
}
if (s->noise_band_count < 15) {
for (int i = s->noise_band_count; i < 15; i++)
levels[i] = levels[i - 1];
}
} else {
for (int i = 0; i < 15; i++) {
levels[i] = -100.0;
}
}
}
static int output_frame(AVFilterLink *inlink)
{
AVFilterContext *ctx = inlink->dst;
AVFilterLink *outlink = ctx->outputs[0];
AudioFFTDeNoiseContext *s = ctx->priv;
AVFrame *out = NULL, *in = NULL;
ThreadData td;
int ret = 0;
in = ff_get_audio_buffer(outlink, s->window_length);
if (!in)
return AVERROR(ENOMEM);
ret = av_audio_fifo_peek(s->fifo, (void **)in->extended_data, s->window_length);
if (ret < 0)
goto end;
if (s->track_noise) {
for (int ch = 0; ch < inlink->channels; ch++) {
DeNoiseChannel *dnch = &s->dnch[ch];
double levels[15];
get_auto_noise_levels(s, dnch, levels);
set_noise_profile(s, dnch, levels, 0);
}
if (s->noise_floor != s->last_noise_floor)
set_parameters(s);
}
if (s->sample_noise_start) {
for (int ch = 0; ch < inlink->channels; ch++) {
DeNoiseChannel *dnch = &s->dnch[ch];
init_sample_noise(dnch);
}
s->sample_noise_start = 0;
s->sample_noise = 1;
}
if (s->sample_noise) {
for (int ch = 0; ch < inlink->channels; ch++) {
DeNoiseChannel *dnch = &s->dnch[ch];
sample_noise_block(s, dnch, in, ch);
}
}
if (s->sample_noise_end) {
for (int ch = 0; ch < inlink->channels; ch++) {
DeNoiseChannel *dnch = &s->dnch[ch];
double sample_noise[15];
finish_sample_noise(s, dnch, sample_noise);
set_noise_profile(s, dnch, sample_noise, 1);
set_band_parameters(s, dnch);
}
s->sample_noise = 0;
s->sample_noise_end = 0;
}
s->block_count++;
td.in = in;
ctx->internal->execute(ctx, filter_channel, &td, NULL,
FFMIN(outlink->channels, ff_filter_get_nb_threads(ctx)));
out = ff_get_audio_buffer(outlink, s->sample_advance);
if (!out) {
ret = AVERROR(ENOMEM);
goto end;
}
for (int ch = 0; ch < inlink->channels; ch++) {
DeNoiseChannel *dnch = &s->dnch[ch];
double *src = dnch->out_samples;
float *orig = (float *)in->extended_data[ch];
float *dst = (float *)out->extended_data[ch];
switch (s->output_mode) {
case IN_MODE:
for (int m = 0; m < s->sample_advance; m++)
dst[m] = orig[m];
break;
case OUT_MODE:
for (int m = 0; m < s->sample_advance; m++)
dst[m] = src[m];
break;
case NOISE_MODE:
for (int m = 0; m < s->sample_advance; m++)
dst[m] = orig[m] - src[m];
break;
default:
av_frame_free(&out);
ret = AVERROR_BUG;
goto end;
}
memmove(src, src + s->sample_advance, (s->window_length - s->sample_advance) * sizeof(*src));
memset(src + (s->window_length - s->sample_advance), 0, s->sample_advance * sizeof(*src));
}
av_audio_fifo_drain(s->fifo, s->sample_advance);
out->pts = s->pts;
ret = ff_filter_frame(outlink, out);
if (ret < 0)
goto end;
s->pts += av_rescale_q(s->sample_advance, (AVRational){1, outlink->sample_rate}, outlink->time_base);
end:
av_frame_free(&in);
return ret;
}
static int activate(AVFilterContext *ctx)
{
AVFilterLink *inlink = ctx->inputs[0];
AVFilterLink *outlink = ctx->outputs[0];
AudioFFTDeNoiseContext *s = ctx->priv;
AVFrame *frame = NULL;
int ret;
FF_FILTER_FORWARD_STATUS_BACK(outlink, inlink);
ret = ff_inlink_consume_frame(inlink, &frame);
if (ret < 0)
return ret;
if (ret > 0) {
if (s->pts == AV_NOPTS_VALUE)
s->pts = frame->pts;
ret = av_audio_fifo_write(s->fifo, (void **)frame->extended_data, frame->nb_samples);
av_frame_free(&frame);
if (ret < 0)
return ret;
}
if (av_audio_fifo_size(s->fifo) >= s->window_length)
return output_frame(inlink);
FF_FILTER_FORWARD_STATUS(inlink, outlink);
if (ff_outlink_frame_wanted(outlink) &&
av_audio_fifo_size(s->fifo) < s->window_length) {
ff_inlink_request_frame(inlink);
return 0;
}
return FFERROR_NOT_READY;
}
static av_cold void uninit(AVFilterContext *ctx)
{
AudioFFTDeNoiseContext *s = ctx->priv;
av_freep(&s->window);
av_freep(&s->bin2band);
av_freep(&s->band_alpha);
av_freep(&s->band_beta);
if (s->dnch) {
for (int ch = 0; ch < s->channels; ch++) {
DeNoiseChannel *dnch = &s->dnch[ch];
av_freep(&dnch->amt);
av_freep(&dnch->band_amt);
av_freep(&dnch->band_excit);
av_freep(&dnch->gain);
av_freep(&dnch->prior);
av_freep(&dnch->prior_band_excit);
av_freep(&dnch->clean_data);
av_freep(&dnch->noisy_data);
av_freep(&dnch->out_samples);
av_freep(&dnch->spread_function);
av_freep(&dnch->abs_var);
av_freep(&dnch->rel_var);
av_freep(&dnch->min_abs_var);
av_freep(&dnch->fft_data);
av_fft_end(dnch->fft);
dnch->fft = NULL;
av_fft_end(dnch->ifft);
dnch->ifft = NULL;
}
av_freep(&s->dnch);
}
av_audio_fifo_free(s->fifo);
}
static int query_formats(AVFilterContext *ctx)
{
AVFilterFormats *formats = NULL;
AVFilterChannelLayouts *layouts = NULL;
static const enum AVSampleFormat sample_fmts[] = {
AV_SAMPLE_FMT_FLTP,
AV_SAMPLE_FMT_NONE
};
int ret;
formats = ff_make_format_list(sample_fmts);
if (!formats)
return AVERROR(ENOMEM);
ret = ff_set_common_formats(ctx, formats);
if (ret < 0)
return ret;
layouts = ff_all_channel_counts();
if (!layouts)
return AVERROR(ENOMEM);
ret = ff_set_common_channel_layouts(ctx, layouts);
if (ret < 0)
return ret;
formats = ff_all_samplerates();
return ff_set_common_samplerates(ctx, formats);
}
static int process_command(AVFilterContext *ctx, const char *cmd, const char *args,
char *res, int res_len, int flags)
{
AudioFFTDeNoiseContext *s = ctx->priv;
int need_reset = 0;
int ret = 0;
if (!strcmp(cmd, "sample_noise") ||
!strcmp(cmd, "sn")) {
if (!strcmp(args, "start")) {
s->sample_noise_start = 1;
s->sample_noise_end = 0;
} else if (!strcmp(args, "end") ||
!strcmp(args, "stop")) {
s->sample_noise_start = 0;
s->sample_noise_end = 1;
}
} else {
ret = ff_filter_process_command(ctx, cmd, args, res, res_len, flags);
if (ret < 0)
return ret;
need_reset = 1;
}
if (need_reset)
set_parameters(s);
return 0;
}
static const AVFilterPad inputs[] = {
{
.name = "default",
.type = AVMEDIA_TYPE_AUDIO,
.config_props = config_input,
},
{ NULL }
};
static const AVFilterPad outputs[] = {
{
.name = "default",
.type = AVMEDIA_TYPE_AUDIO,
},
{ NULL }
};
AVFilter ff_af_afftdn = {
.name = "afftdn",
.description = NULL_IF_CONFIG_SMALL("Denoise audio samples using FFT."),
.query_formats = query_formats,
.priv_size = sizeof(AudioFFTDeNoiseContext),
.priv_class = &afftdn_class,
.activate = activate,
.uninit = uninit,
.inputs = inputs,
.outputs = outputs,
.process_command = process_command,
.flags = AVFILTER_FLAG_SUPPORT_TIMELINE_GENERIC |
AVFILTER_FLAG_SLICE_THREADS,
};