You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
458 lines
13 KiB
458 lines
13 KiB
/*
|
|
* filter_sox.c -- apply any number of SOX effects using libst
|
|
* Copyright (C) 2003-2004 Ushodaya Enterprises Limited
|
|
* Author: Dan Dennedy <dan@dennedy.org>
|
|
*
|
|
* This library 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.
|
|
*
|
|
* This library 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 this library; if not, write to the Free Software
|
|
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
|
|
*/
|
|
|
|
#include "filter_sox.h"
|
|
|
|
#include <framework/mlt_frame.h>
|
|
#include <framework/mlt_tokeniser.h>
|
|
|
|
#include <stdio.h>
|
|
#include <stdlib.h>
|
|
#include <string.h>
|
|
#include <math.h>
|
|
|
|
#ifdef SOX14
|
|
# include <sox.h>
|
|
# define ST_EOF SOX_EOF
|
|
# define ST_SUCCESS SOX_SUCCESS
|
|
# define st_sample_t sox_sample_t
|
|
# define eff_t sox_effect_t*
|
|
# define st_size_t sox_size_t
|
|
# define ST_LIB_VERSION_CODE SOX_LIB_VERSION_CODE
|
|
# define ST_LIB_VERSION SOX_LIB_VERSION
|
|
# define ST_SIGNED_WORD_TO_SAMPLE(d,clips) SOX_SIGNED_16BIT_TO_SAMPLE(d,clips)
|
|
# define ST_SSIZE_MIN SOX_SSIZE_MIN
|
|
# define ST_SAMPLE_TO_SIGNED_WORD(d,clips) SOX_SAMPLE_TO_SIGNED_16BIT(d,clips)
|
|
#else
|
|
# include <st.h>
|
|
#endif
|
|
|
|
#define BUFFER_LEN 8192
|
|
#define AMPLITUDE_NORM 0.2511886431509580 /* -12dBFS */
|
|
#define AMPLITUDE_MIN 0.00001
|
|
|
|
/** Compute the mean of a set of doubles skipping unset values flagged as -1
|
|
*/
|
|
static inline double mean( double *buf, int count )
|
|
{
|
|
double mean = 0;
|
|
int i;
|
|
int j = 0;
|
|
|
|
for ( i = 0; i < count; i++ )
|
|
{
|
|
if ( buf[ i ] != -1.0 )
|
|
{
|
|
mean += buf[ i ];
|
|
j ++;
|
|
}
|
|
}
|
|
if ( j > 0 )
|
|
mean /= j;
|
|
|
|
return mean;
|
|
}
|
|
|
|
/** Create an effect state instance for a channels
|
|
*/
|
|
static int create_effect( mlt_filter this, char *value, int count, int channel, int frequency )
|
|
{
|
|
mlt_tokeniser tokeniser = mlt_tokeniser_init();
|
|
#ifdef SOX14
|
|
eff_t eff = mlt_pool_alloc( sizeof( sox_effect_t ) );
|
|
#else
|
|
eff_t eff = mlt_pool_alloc( sizeof( struct st_effect ) );
|
|
#endif
|
|
char id[ 256 ];
|
|
int error = 1;
|
|
|
|
// Tokenise the effect specification
|
|
mlt_tokeniser_parse_new( tokeniser, value, " " );
|
|
if ( tokeniser->count < 1 )
|
|
return error;
|
|
|
|
// Locate the effect
|
|
#ifdef SOX14
|
|
//fprintf(stderr, "%s: effect %s count %d\n", __FUNCTION__, tokeniser->tokens[0], tokeniser->count );
|
|
sox_create_effect( eff, sox_find_effect( tokeniser->tokens[0] ) );
|
|
int opt_count = tokeniser->count - 1;
|
|
#else
|
|
int opt_count = st_geteffect_opt( eff, tokeniser->count, tokeniser->tokens );
|
|
#endif
|
|
|
|
// If valid effect
|
|
if ( opt_count != ST_EOF )
|
|
{
|
|
// Supply the effect parameters
|
|
#ifdef SOX14
|
|
if ( ( * eff->handler.getopts )( eff, opt_count, &tokeniser->tokens[ tokeniser->count > 1 ? 1 : 0 ] ) == ST_SUCCESS )
|
|
#else
|
|
if ( ( * eff->h->getopts )( eff, opt_count, &tokeniser->tokens[ tokeniser->count - opt_count ] ) == ST_SUCCESS )
|
|
#endif
|
|
{
|
|
// Set the sox signal parameters
|
|
eff->ininfo.rate = frequency;
|
|
eff->outinfo.rate = frequency;
|
|
eff->ininfo.channels = 1;
|
|
eff->outinfo.channels = 1;
|
|
|
|
// Start the effect
|
|
#ifdef SOX14
|
|
if ( ( * eff->handler.start )( eff ) == ST_SUCCESS )
|
|
#else
|
|
if ( ( * eff->h->start )( eff ) == ST_SUCCESS )
|
|
#endif
|
|
{
|
|
// Construct id
|
|
sprintf( id, "_effect_%d_%d", count, channel );
|
|
|
|
// Save the effect state
|
|
mlt_properties_set_data( MLT_FILTER_PROPERTIES( this ), id, eff, 0, mlt_pool_release, NULL );
|
|
error = 0;
|
|
}
|
|
}
|
|
}
|
|
// Some error occurred so delete the temp effect state
|
|
if ( error == 1 )
|
|
mlt_pool_release( eff );
|
|
|
|
mlt_tokeniser_close( tokeniser );
|
|
|
|
return error;
|
|
}
|
|
|
|
/** Get the audio.
|
|
*/
|
|
|
|
static int filter_get_audio( mlt_frame frame, int16_t **buffer, mlt_audio_format *format, int *frequency, int *channels, int *samples )
|
|
{
|
|
// Get the properties of the frame
|
|
mlt_properties properties = MLT_FRAME_PROPERTIES( frame );
|
|
|
|
// Get the filter service
|
|
mlt_filter filter = mlt_frame_pop_audio( frame );
|
|
|
|
// Get the filter properties
|
|
mlt_properties filter_properties = MLT_FILTER_PROPERTIES( filter );
|
|
|
|
// Get the properties
|
|
st_sample_t *input_buffer = mlt_properties_get_data( filter_properties, "input_buffer", NULL );
|
|
st_sample_t *output_buffer = mlt_properties_get_data( filter_properties, "output_buffer", NULL );
|
|
int channels_avail = *channels;
|
|
int i; // channel
|
|
int count = mlt_properties_get_int( filter_properties, "_effect_count" );
|
|
|
|
// Get the producer's audio
|
|
mlt_frame_get_audio( frame, buffer, format, frequency, &channels_avail, samples );
|
|
|
|
// Duplicate channels as necessary
|
|
if ( channels_avail < *channels )
|
|
{
|
|
int size = *channels * *samples * sizeof( int16_t );
|
|
int16_t *new_buffer = mlt_pool_alloc( size );
|
|
int j, k = 0;
|
|
|
|
// Duplicate the existing channels
|
|
for ( i = 0; i < *samples; i++ )
|
|
{
|
|
for ( j = 0; j < *channels; j++ )
|
|
{
|
|
new_buffer[ ( i * *channels ) + j ] = (*buffer)[ ( i * channels_avail ) + k ];
|
|
k = ( k + 1 ) % channels_avail;
|
|
}
|
|
}
|
|
|
|
// Update the audio buffer now - destroys the old
|
|
mlt_properties_set_data( properties, "audio", new_buffer, size, ( mlt_destructor )mlt_pool_release, NULL );
|
|
|
|
*buffer = new_buffer;
|
|
}
|
|
else if ( channels_avail == 6 && *channels == 2 )
|
|
{
|
|
// Nasty hack for ac3 5.1 audio - may be a cause of failure?
|
|
int size = *channels * *samples * sizeof( int16_t );
|
|
int16_t *new_buffer = mlt_pool_alloc( size );
|
|
|
|
// Drop all but the first *channels
|
|
for ( i = 0; i < *samples; i++ )
|
|
{
|
|
new_buffer[ ( i * *channels ) + 0 ] = (*buffer)[ ( i * channels_avail ) + 2 ];
|
|
new_buffer[ ( i * *channels ) + 1 ] = (*buffer)[ ( i * channels_avail ) + 3 ];
|
|
}
|
|
|
|
// Update the audio buffer now - destroys the old
|
|
mlt_properties_set_data( properties, "audio", new_buffer, size, ( mlt_destructor )mlt_pool_release, NULL );
|
|
|
|
*buffer = new_buffer;
|
|
}
|
|
|
|
// Even though some effects are multi-channel aware, it is not reliable
|
|
// We must maintain a separate effect state for each channel
|
|
for ( i = 0; i < *channels; i++ )
|
|
{
|
|
char id[ 256 ];
|
|
sprintf( id, "_effect_0_%d", i );
|
|
|
|
// Get an existing effect state
|
|
eff_t e = mlt_properties_get_data( filter_properties, id, NULL );
|
|
|
|
// Validate the existing effect state
|
|
if ( e != NULL && ( e->ininfo.rate != *frequency ||
|
|
e->outinfo.rate != *frequency ) )
|
|
e = NULL;
|
|
|
|
// (Re)Create the effect state
|
|
if ( e == NULL )
|
|
{
|
|
int j = 0;
|
|
|
|
// Reset the count
|
|
count = 0;
|
|
|
|
// Loop over all properties
|
|
for ( j = 0; j < mlt_properties_count( filter_properties ); j ++ )
|
|
{
|
|
// Get the name of this property
|
|
char *name = mlt_properties_get_name( filter_properties, j );
|
|
|
|
// If the name does not contain a . and matches effect
|
|
if ( !strncmp( name, "effect", 6 ) )
|
|
{
|
|
// Get the effect specification
|
|
char *value = mlt_properties_get( filter_properties, name );
|
|
|
|
// Create an instance
|
|
if ( create_effect( filter, value, count, i, *frequency ) == 0 )
|
|
count ++;
|
|
}
|
|
}
|
|
|
|
// Save the number of filters
|
|
mlt_properties_set_int( filter_properties, "_effect_count", count );
|
|
|
|
}
|
|
if ( *samples > 0 && count > 0 )
|
|
{
|
|
st_sample_t *p = input_buffer;
|
|
st_sample_t *end = p + *samples;
|
|
int16_t *q = *buffer + i;
|
|
st_size_t isamp = *samples;
|
|
st_size_t osamp = *samples;
|
|
double rms = 0;
|
|
int j;
|
|
char *normalise = mlt_properties_get( filter_properties, "normalise" );
|
|
double normalised_gain = 1.0;
|
|
#if (ST_LIB_VERSION_CODE >= ST_LIB_VERSION(13,0,0))
|
|
st_sample_t dummy_clipped_count = 0;
|
|
#endif
|
|
|
|
// Convert to sox encoding
|
|
while( p != end )
|
|
{
|
|
#if (ST_LIB_VERSION_CODE >= ST_LIB_VERSION(13,0,0))
|
|
*p = ST_SIGNED_WORD_TO_SAMPLE( *q, dummy_clipped_count );
|
|
#else
|
|
*p = ST_SIGNED_WORD_TO_SAMPLE( *q );
|
|
#endif
|
|
// Compute rms amplitude while we are accessing each sample
|
|
rms += ( double )*p * ( double )*p;
|
|
|
|
p ++;
|
|
q += *channels;
|
|
}
|
|
|
|
// Compute final rms amplitude
|
|
rms = sqrt( rms / *samples / ST_SSIZE_MIN / ST_SSIZE_MIN );
|
|
|
|
if ( normalise )
|
|
{
|
|
int window = mlt_properties_get_int( filter_properties, "window" );
|
|
double *smooth_buffer = mlt_properties_get_data( filter_properties, "smooth_buffer", NULL );
|
|
double max_gain = mlt_properties_get_double( filter_properties, "max_gain" );
|
|
|
|
// Default the maximum gain factor to 20dBFS
|
|
if ( max_gain == 0 )
|
|
max_gain = 10.0;
|
|
|
|
// The smoothing buffer prevents radical shifts in the gain level
|
|
if ( window > 0 && smooth_buffer != NULL )
|
|
{
|
|
int smooth_index = mlt_properties_get_int( filter_properties, "_smooth_index" );
|
|
smooth_buffer[ smooth_index ] = rms;
|
|
|
|
// Ignore very small values that adversely affect the mean
|
|
if ( rms > AMPLITUDE_MIN )
|
|
mlt_properties_set_int( filter_properties, "_smooth_index", ( smooth_index + 1 ) % window );
|
|
|
|
// Smoothing is really just a mean over the past N values
|
|
normalised_gain = AMPLITUDE_NORM / mean( smooth_buffer, window );
|
|
}
|
|
else if ( rms > 0 )
|
|
{
|
|
// Determine gain to apply as current amplitude
|
|
normalised_gain = AMPLITUDE_NORM / rms;
|
|
}
|
|
|
|
//printf("filter_sox: rms %.3f gain %.3f\n", rms, normalised_gain );
|
|
|
|
// Govern the maximum gain
|
|
if ( normalised_gain > max_gain )
|
|
normalised_gain = max_gain;
|
|
}
|
|
|
|
// For each effect
|
|
for ( j = 0; j < count; j++ )
|
|
{
|
|
sprintf( id, "_effect_%d_%d", j, i );
|
|
e = mlt_properties_get_data( filter_properties, id, NULL );
|
|
|
|
// We better have this guy
|
|
if ( e != NULL )
|
|
{
|
|
float saved_gain = 1.0;
|
|
|
|
// XXX: hack to apply the normalised gain level to the vol effect
|
|
#ifdef SOX14
|
|
if ( normalise && strcmp( e->handler.name, "vol" ) == 0 )
|
|
#else
|
|
if ( normalise && strcmp( e->name, "vol" ) == 0 )
|
|
#endif
|
|
{
|
|
float *f = ( float * )( e->priv );
|
|
saved_gain = *f;
|
|
*f = saved_gain * normalised_gain;
|
|
}
|
|
|
|
// Apply the effect
|
|
#ifdef SOX14
|
|
if ( ( * e->handler.flow )( e, input_buffer, output_buffer, &isamp, &osamp ) == ST_SUCCESS )
|
|
#else
|
|
if ( ( * e->h->flow )( e, input_buffer, output_buffer, &isamp, &osamp ) == ST_SUCCESS )
|
|
#endif
|
|
{
|
|
// Swap input and output buffer pointers for subsequent effects
|
|
p = input_buffer;
|
|
input_buffer = output_buffer;
|
|
output_buffer = p;
|
|
}
|
|
|
|
// XXX: hack to restore the original vol gain to prevent accumulation
|
|
#ifdef SOX14
|
|
if ( normalise && strcmp( e->handler.name, "vol" ) == 0 )
|
|
#else
|
|
if ( normalise && strcmp( e->name, "vol" ) == 0 )
|
|
#endif
|
|
{
|
|
float *f = ( float * )( e->priv );
|
|
*f = saved_gain;
|
|
}
|
|
}
|
|
}
|
|
|
|
// Convert back to signed 16bit
|
|
p = input_buffer;
|
|
q = *buffer + i;
|
|
end = p + *samples;
|
|
while ( p != end )
|
|
{
|
|
#if (ST_LIB_VERSION_CODE >= ST_LIB_VERSION(13,0,0))
|
|
*q = ST_SAMPLE_TO_SIGNED_WORD( *p ++, dummy_clipped_count );
|
|
#else
|
|
*q = ST_SAMPLE_TO_SIGNED_WORD( *p ++ );
|
|
#endif
|
|
q += *channels;
|
|
}
|
|
}
|
|
}
|
|
|
|
return 0;
|
|
}
|
|
|
|
/** Filter processing.
|
|
*/
|
|
|
|
static mlt_frame filter_process( mlt_filter this, mlt_frame frame )
|
|
{
|
|
if ( mlt_frame_is_test_audio( frame ) == 0 )
|
|
{
|
|
// Add the filter to the frame
|
|
mlt_frame_push_audio( frame, this );
|
|
mlt_frame_push_audio( frame, filter_get_audio );
|
|
|
|
// Parse the window property and allocate smoothing buffer if needed
|
|
mlt_properties properties = MLT_FILTER_PROPERTIES( this );
|
|
int window = mlt_properties_get_int( properties, "window" );
|
|
if ( mlt_properties_get( properties, "smooth_buffer" ) == NULL && window > 1 )
|
|
{
|
|
// Create a smoothing buffer for the calculated "max power" of frame of audio used in normalisation
|
|
double *smooth_buffer = (double*) calloc( window, sizeof( double ) );
|
|
int i;
|
|
for ( i = 0; i < window; i++ )
|
|
smooth_buffer[ i ] = -1.0;
|
|
mlt_properties_set_data( properties, "smooth_buffer", smooth_buffer, 0, free, NULL );
|
|
}
|
|
}
|
|
|
|
return frame;
|
|
}
|
|
|
|
/** Constructor for the filter.
|
|
*/
|
|
|
|
mlt_filter filter_sox_init( char *arg )
|
|
{
|
|
mlt_filter this = mlt_filter_new( );
|
|
if ( this != NULL )
|
|
{
|
|
void *input_buffer = mlt_pool_alloc( BUFFER_LEN );
|
|
void *output_buffer = mlt_pool_alloc( BUFFER_LEN );
|
|
mlt_properties properties = MLT_FILTER_PROPERTIES( this );
|
|
|
|
this->process = filter_process;
|
|
|
|
if ( arg != NULL )
|
|
mlt_properties_set( properties, "effect", arg );
|
|
mlt_properties_set_data( properties, "input_buffer", input_buffer, BUFFER_LEN, mlt_pool_release, NULL );
|
|
mlt_properties_set_data( properties, "output_buffer", output_buffer, BUFFER_LEN, mlt_pool_release, NULL );
|
|
mlt_properties_set_int( properties, "window", 75 );
|
|
}
|
|
return this;
|
|
}
|
|
|
|
// What to do when a libst internal failure occurs
|
|
void cleanup(void){}
|
|
|
|
// Is there a build problem with my sox-devel package?
|
|
#ifndef gsm_create
|
|
void gsm_create(void){}
|
|
#endif
|
|
#ifndef gsm_decode
|
|
void gsm_decode(void){}
|
|
#endif
|
|
#ifndef gdm_encode
|
|
void gsm_encode(void){}
|
|
#endif
|
|
#ifndef gsm_destroy
|
|
void gsm_destroy(void){}
|
|
#endif
|
|
#ifndef gsm_option
|
|
void gsm_option(void){}
|
|
#endif
|