Skip to content

V
vlc-1.1
Commit 1206810b authored by Gildas Bazin's avatar Gildas Bazin
Browse files
Options 

* modules/audio_filter/resampler/bandlimited.c: fix bug that was affecting
   quality badly + some clean-up.
   Changed the module priority so it is now the default resampler.
parent eb1ba666
Show whitespace changes
Inline Side-by-side
Showing with 37 additions and 33 deletions
modules/audio_filter/resampler/bandlimited.c
View file @ 1206810b
......@@ -2,7 +2,7 @@
* bandlimited.c : bandlimited interpolation resampler
*****************************************************************************
* Copyright (C) 2002 VideoLAN
* $Id: bandlimited.c,v 1.1 2003/03/04 03:27:40 gbazin Exp $
* $Id: bandlimited.c,v 1.2 2003/03/04 19:28:39 gbazin Exp $
*
* Authors: Gildas Bazin <gbazin@netcourrier.com>
*
......@@ -67,7 +67,7 @@ struct aout_filter_sys_t
int i_buf_size;
int i_old_rate;
int d_old_factor;
double d_old_factor;
int i_old_wing;
unsigned int i_remainder; /* remainder of previous sample */
......@@ -80,7 +80,7 @@ struct aout_filter_sys_t
*****************************************************************************/
vlc_module_begin();
set_description( _("audio filter for bandlimited interpolation resampling") );
set_capability( "audio filter", 4 );
set_capability( "audio filter", 20 );
set_callbacks( Create, Close );
vlc_module_end();
......@@ -158,10 +158,11 @@ static void DoWork( aout_instance_t * p_aout, aout_filter_t * p_filter,
int i_nb_channels = aout_FormatNbChannels( &p_filter->input );
int i_in_nb = p_in_buf->i_nb_samples;
int i_in, i_out = 0;
double d_factor = (double)p_aout->mixer.mixer.i_rate
/ p_filter->input.i_rate;
int i_filter_wing, i_left_over;
double d_factor, d_scale_factor, d_old_scale_factor;
int i_filter_wing;
#if 0
int i;
#endif
/* Check if we really need to run the resampler */
if( p_aout->mixer.mixer.i_rate == p_filter->input.i_rate )
......@@ -213,18 +214,11 @@ static void DoWork( aout_instance_t * p_aout, aout_filter_t * p_filter,
}
#if 0
msg_Err( p_filter, "old rate: %i, old factor: %i, old wing: %i, i_in: %i",
msg_Err( p_filter, "old rate: %i, old factor: %f, old wing: %i, i_in: %i",
p_filter->p_sys->i_old_rate, p_filter->p_sys->d_old_factor,
p_filter->p_sys->i_old_wing, i_in_nb );
#endif
/* Calculate the length of the filter wing */
d_factor = (double)p_aout->mixer.mixer.i_rate / p_filter->input.i_rate;
i_filter_wing = ((SMALL_FILTER_NMULT+1)/2.0) * __MAX(1.0,1.0/d_factor) + 1;
/* Check if we have enough buffered data to start with the new rate. */
i_left_over = i_filter_wing - p_filter->p_sys->i_old_wing;
/* Prepare the source buffer */
i_in_nb += (p_filter->p_sys->i_old_wing * 2);
#ifdef HAVE_ALLOCA
......@@ -254,16 +248,20 @@ static void DoWork( aout_instance_t * p_aout, aout_filter_t * p_filter,
/* Make sure the output buffer is reset */
memset( p_out, 0, p_out_buf->i_size );
#if 0
/* Calculate the new length of the filter wing */
d_factor = (double)p_aout->mixer.mixer.i_rate / p_filter->input.i_rate;
i_filter_wing = ((SMALL_FILTER_NMULT+1)/2.0) * __MAX(1.0,1.0/d_factor) + 1;
/* Account for increased filter gain when using factors less than 1 */
if( d_factor < 1 )
{
LpScl = SMALL_FILTER_SCALE * d_factor + 0.5;
}
#endif
d_old_scale_factor = SMALL_FILTER_SCALE *
p_filter->p_sys->d_old_factor + 0.5;
d_scale_factor = SMALL_FILTER_SCALE * d_factor + 0.5;
/* Apply the old rate until we have enough samples for the new one */
for( i_in = p_filter->p_sys->i_old_wing; i_in < i_left_over; i_in++ )
/* TODO: Check we have enough samples */
i_in = p_filter->p_sys->i_old_wing;
p_in += p_filter->p_sys->i_old_wing * i_nb_channels;
for( ; i_in < i_filter_wing; i_in++ )
{
if( p_filter->p_sys->d_old_factor == 1 )
{
......@@ -297,6 +295,13 @@ static void DoWork( aout_instance_t * p_aout, aout_filter_t * p_filter,
p_filter->output.i_rate,
1, i_nb_channels );
#if 0
/* Normalize for unity filter gain */
for( i = 0; i < i_nb_channels; i++ )
{
*(p_out+i) *= d_old_scale_factor;
}
#endif
}
else
{
......@@ -315,10 +320,6 @@ static void DoWork( aout_instance_t * p_aout, aout_filter_t * p_filter,
1, i_nb_channels );
}
#if 0
v *= LpScl; /* Normalize for unity filter gain */
#endif
p_out += i_nb_channels;
i_out++;
......@@ -330,6 +331,7 @@ static void DoWork( aout_instance_t * p_aout, aout_filter_t * p_filter,
}
/* Apply the new rate for the rest of the samples */
/* TODO: Check we have enough future samples for the new rate */
if( i_in < i_in_nb - i_filter_wing )
{
p_filter->p_sys->i_old_rate = p_filter->input.i_rate;
......@@ -359,6 +361,14 @@ static void DoWork( aout_instance_t * p_aout, aout_filter_t * p_filter,
p_filter->p_sys->i_remainder,
p_filter->output.i_rate,
1, i_nb_channels );
#if 0
/* Normalize for unity filter gain */
for( i = 0; i < i_nb_channels; i++ )
{
*(p_out+i) *= d_old_scale_factor;
}
#endif
}
else
{
......@@ -377,10 +387,6 @@ static void DoWork( aout_instance_t * p_aout, aout_filter_t * p_filter,
1, i_nb_channels );
}
#if 0
v *= LpScl; /* Normalize for unity filter gain */
#endif
p_out += i_nb_channels;
i_out++;
......@@ -401,7 +407,7 @@ static void DoWork( aout_instance_t * p_aout, aout_filter_t * p_filter,
}
#if 0
msg_Err( p_filter, "pout size: %i, nb_samples out: %i", p_out_buf->i_size,
msg_Err( p_filter, "p_out size: %i, nb bytes out: %i", p_out_buf->i_size,
i_out * p_filter->input.i_bytes_per_frame );
#endif
......@@ -506,8 +512,6 @@ void FilterFloatUD( float Imp[], float ImpD[], uint16_t Nwing, float *p_in,
((ui_output_rate * ui_counter + ui_remainder)<< Nhc) -
((ui_output_rate * ui_counter + ui_remainder)<< Nhc) /
ui_input_rate * ui_input_rate;
//(ui_remainder<<Nhc)* ui_output_rate/ui_input_rate -
//(ui_remainder<<Nhc) / ui_input_rate * ui_output_rate;
t += *Hdp * ui_linear_remainder / ui_input_rate / Npc;
for( i = 0; i < i_nb_channels; i++ )
{
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment