2012-02-08 19:55:37 +00:00
|
|
|
/***************************************************************************
|
|
|
|
* __________ __ ___.
|
|
|
|
* Open \______ \ ____ ____ | | _\_ |__ _______ ___
|
|
|
|
* Source | _// _ \_/ ___\| |/ /| __ \ / _ \ \/ /
|
|
|
|
* Jukebox | | ( <_> ) \___| < | \_\ ( <_> > < <
|
|
|
|
* Firmware |____|_ /\____/ \___ >__|_ \|___ /\____/__/\_ \
|
|
|
|
* \/ \/ \/ \/ \/
|
|
|
|
* $Id$
|
|
|
|
*
|
|
|
|
* Copyright (C) 2009 Jeffrey Goode
|
|
|
|
*
|
|
|
|
* This program is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU General Public License
|
|
|
|
* as published by the Free Software Foundation; either version 2
|
|
|
|
* of the License, or (at your option) any later version.
|
|
|
|
*
|
|
|
|
* This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
|
|
|
|
* KIND, either express or implied.
|
|
|
|
*
|
|
|
|
****************************************************************************/
|
2013-05-27 07:23:33 +00:00
|
|
|
#include "rbcodecconfig.h"
|
2012-02-08 19:55:37 +00:00
|
|
|
#include "fixedpoint.h"
|
|
|
|
#include "fracmul.h"
|
2012-03-27 23:52:15 +00:00
|
|
|
#include <string.h>
|
2012-02-08 19:55:37 +00:00
|
|
|
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
/* Define LOGF_ENABLE to enable logf output in this file
|
|
|
|
* #define LOGF_ENABLE
|
|
|
|
*/
|
2012-02-08 19:55:37 +00:00
|
|
|
#include "logf.h"
|
2012-03-27 23:52:15 +00:00
|
|
|
#include "dsp_proc_entry.h"
|
2012-04-29 21:31:30 +00:00
|
|
|
#include "compressor.h"
|
2013-05-23 17:58:51 +00:00
|
|
|
#include "dsp_misc.h"
|
2012-03-27 23:52:15 +00:00
|
|
|
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
#define UNITY (1L << 24) /* unity gain in S7.24 format */
|
|
|
|
#define MAX_DLY 960 /* Max number of samples to delay
|
|
|
|
output (960 = 5ms @ 192 kHz)
|
|
|
|
*/
|
|
|
|
#define MAX_CH 4 /* Is there a good malloc() or equal
|
|
|
|
for rockbox?
|
|
|
|
*/
|
|
|
|
#define DLY_TIME 3 /* milliseconds */
|
|
|
|
|
2012-04-26 21:19:16 +00:00
|
|
|
static struct compressor_settings curr_set; /* Cached settings */
|
|
|
|
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
static int32_t comp_makeup_gain IBSS_ATTR; /* S7.24 format */
|
|
|
|
static int32_t comp_curve[66] IBSS_ATTR; /* S7.24 format */
|
|
|
|
static int32_t release_gain IBSS_ATTR; /* S7.24 format */
|
|
|
|
static int32_t release_holdoff IBSS_ATTR; /* S7.24 format */
|
|
|
|
|
|
|
|
/* 1-pole filter coefficients for exponential attack/release times */
|
|
|
|
static int32_t rlsca IBSS_ATTR; /* Release 'alpha' */
|
|
|
|
static int32_t rlscb IBSS_ATTR; /* Release 'beta' */
|
|
|
|
|
|
|
|
static int32_t attca IBSS_ATTR; /* Attack 'alpha' */
|
|
|
|
static int32_t attcb IBSS_ATTR; /* Attack 'beta' */
|
|
|
|
|
|
|
|
static int32_t limitca IBSS_ATTR; /* Limiter Attack 'alpha' */
|
|
|
|
|
|
|
|
/* 1-pole filter coefficients for sidechain pre-emphasis filters */
|
|
|
|
static int32_t hp1ca IBSS_ATTR; /* hpf1 'alpha' */
|
|
|
|
static int32_t hp2ca IBSS_ATTR; /* hpf2 'beta' */
|
|
|
|
|
|
|
|
/* 1-pole hp filter state variables for pre-emphasis filters */
|
|
|
|
static int32_t hpfx1 IBSS_ATTR; /* hpf1 and hpf2 x[n-1] */
|
|
|
|
static int32_t hp1y1 IBSS_ATTR; /* hpf2 y[n-1] */
|
|
|
|
static int32_t hp2y1 IBSS_ATTR; /* hpf2 y[n-1] */
|
|
|
|
|
|
|
|
/* Delay Line for look-ahead compression */
|
|
|
|
static int32_t labuf[MAX_CH][MAX_DLY]; /* look-ahead buffer */
|
|
|
|
static int32_t delay_time;
|
|
|
|
static int32_t delay_write;
|
|
|
|
static int32_t delay_read;
|
|
|
|
|
|
|
|
/** 1-Pole LP Filter first coefficient computation
|
|
|
|
* Returns S7.24 format integer used for "a" coefficient
|
|
|
|
* rc: "RC Time Constant", or time to decay to 1/e
|
|
|
|
* fs: Sampling Rate
|
|
|
|
* Interpret attack and release time as an RC time constant
|
|
|
|
* (time to decay to 1/e)
|
|
|
|
* 1-pole filters use approximation
|
|
|
|
* a0 = 1/(fs*rc + 1)
|
|
|
|
* b1 = 1.0 - a0
|
|
|
|
* fs = Sampling Rate
|
|
|
|
* rc = Time to decay to 1/e
|
|
|
|
* y[n] = a0*x[n] + b1*y[n-1]
|
|
|
|
*
|
|
|
|
* According to simulation on Intel hardware
|
|
|
|
* this algorithm produces < 2% error for rc < ~100ms
|
|
|
|
* For rc 100ms - 1000ms, error approaches 0%
|
|
|
|
* For compressor attack/release times, this is more than adequate.
|
|
|
|
*
|
|
|
|
* Error was measured against the more rigorous computation:
|
|
|
|
* a0 = 1.0 - e^(-1.0/(fs*rc))
|
|
|
|
*/
|
|
|
|
|
|
|
|
int32_t get_lpf_coeff(int32_t rc, int32_t fs, int32_t rc_units)
|
|
|
|
{
|
|
|
|
int32_t c = fs*rc;
|
|
|
|
c /= rc_units;
|
|
|
|
c += 1;
|
|
|
|
c = UNITY/c;
|
|
|
|
return c;
|
|
|
|
}
|
2012-02-08 19:55:37 +00:00
|
|
|
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
/** Coefficients to get 10dB change per time period "rc"
|
|
|
|
* from 1-pole LP filter topology
|
|
|
|
* This function is better used to match behavior of
|
|
|
|
* linear release which was implemented prior to implementation
|
|
|
|
* of exponential attack/release function
|
|
|
|
*/
|
|
|
|
|
|
|
|
int32_t get_att_rls_coeff(int32_t rc, int32_t fs)
|
|
|
|
{
|
|
|
|
int32_t c = UNITY/fs;
|
|
|
|
c *= 1152; /* 1000 * 10/( 20*log10( 1/e ) ) */
|
|
|
|
c /= rc;
|
|
|
|
return c;
|
|
|
|
}
|
2012-02-08 19:55:37 +00:00
|
|
|
|
|
|
|
/** COMPRESSOR UPDATE
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
* Called via the menu system to configure the compressor process
|
|
|
|
*/
|
2013-05-23 17:58:51 +00:00
|
|
|
static bool compressor_update(struct dsp_config *dsp,
|
|
|
|
const struct compressor_settings *settings)
|
2012-02-08 19:55:37 +00:00
|
|
|
{
|
2012-04-26 21:19:16 +00:00
|
|
|
/* make settings values useful */
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
int threshold = settings->threshold;
|
|
|
|
bool auto_gain = settings->makeup_gain == 1;
|
2012-04-26 21:19:16 +00:00
|
|
|
static const int comp_ratios[] = { 2, 4, 6, 10, 0 };
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
int ratio = comp_ratios[settings->ratio];
|
|
|
|
bool soft_knee = settings->knee == 1;
|
|
|
|
int32_t release = settings->release_time;
|
|
|
|
int32_t attack = settings->attack_time;
|
|
|
|
|
|
|
|
/* Compute Attack and Release Coefficients */
|
|
|
|
int32_t fs = dsp_get_output_frequency(dsp);
|
|
|
|
|
|
|
|
/* Release */
|
|
|
|
rlsca = get_att_rls_coeff(release, fs);
|
|
|
|
rlscb = UNITY - rlsca ;
|
|
|
|
|
|
|
|
/* Attack */
|
|
|
|
if(attack > 0)
|
|
|
|
{
|
|
|
|
attca = get_att_rls_coeff(attack, fs);
|
|
|
|
attcb = UNITY - attca ;
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
attca = UNITY;
|
|
|
|
attcb = 0;
|
|
|
|
}
|
2012-02-08 19:55:37 +00:00
|
|
|
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
|
|
|
|
/* Sidechain pre-emphasis filter coefficients */
|
|
|
|
hp1ca = fs + 0x003C1; /** The "magic" constant is 1/RC. This filter
|
|
|
|
* cut-off is approximately 237 Hz
|
|
|
|
*/
|
|
|
|
hp1ca = UNITY/hp1ca;
|
|
|
|
hp1ca *= fs;
|
|
|
|
|
|
|
|
hp2ca = fs + 0x02065; /* The "magic" constant is 1/RC. This filter
|
|
|
|
* cut-off is approximately 2.18 kHz
|
|
|
|
*/
|
|
|
|
hp2ca = UNITY/hp2ca;
|
|
|
|
hp2ca *= fs;
|
|
|
|
|
|
|
|
bool changed = settings == &curr_set; /* If frequency changes */
|
2012-04-26 21:19:16 +00:00
|
|
|
bool active = threshold < 0;
|
2012-02-08 19:55:37 +00:00
|
|
|
|
2012-04-26 21:19:16 +00:00
|
|
|
if (memcmp(settings, &curr_set, sizeof (curr_set)))
|
2012-02-08 19:55:37 +00:00
|
|
|
{
|
2012-04-26 21:19:16 +00:00
|
|
|
/* Compressor settings have changed since last call */
|
|
|
|
changed = true;
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
|
2012-02-08 19:55:37 +00:00
|
|
|
#if defined(ROCKBOX_HAS_LOGF) && defined(LOGF_ENABLE)
|
2012-04-26 21:19:16 +00:00
|
|
|
if (settings->threshold != curr_set.threshold)
|
|
|
|
{
|
|
|
|
logf(" Compressor Threshold: %d dB\tEnabled: %s",
|
|
|
|
threshold, active ? "Yes" : "No");
|
2012-02-08 19:55:37 +00:00
|
|
|
}
|
|
|
|
|
2012-04-26 21:19:16 +00:00
|
|
|
if (settings->makeup_gain != curr_set.makeup_gain)
|
2012-02-08 19:55:37 +00:00
|
|
|
{
|
2012-04-26 21:19:16 +00:00
|
|
|
logf(" Compressor Makeup Gain: %s",
|
|
|
|
auto_gain ? "Auto" : "Off");
|
|
|
|
}
|
|
|
|
|
|
|
|
if (settings->ratio != cur_set.ratio)
|
2012-02-08 19:55:37 +00:00
|
|
|
{
|
|
|
|
if (ratio)
|
2012-04-26 21:19:16 +00:00
|
|
|
{ logf(" Compressor Ratio: %d:1", ratio); }
|
2012-02-08 19:55:37 +00:00
|
|
|
else
|
2012-04-26 21:19:16 +00:00
|
|
|
{ logf(" Compressor Ratio: Limit"); }
|
2012-02-08 19:55:37 +00:00
|
|
|
}
|
2012-04-26 21:19:16 +00:00
|
|
|
|
|
|
|
if (settings->knee != cur_set.knee)
|
2012-02-08 19:55:37 +00:00
|
|
|
{
|
2012-04-26 21:19:16 +00:00
|
|
|
logf(" Compressor Knee: %s", soft_knee?"Soft":"Hard");
|
2012-02-08 19:55:37 +00:00
|
|
|
}
|
2012-04-26 21:19:16 +00:00
|
|
|
|
|
|
|
if (settings->release_time != cur_set.release_time)
|
2012-02-08 19:55:37 +00:00
|
|
|
{
|
2012-04-26 21:19:16 +00:00
|
|
|
logf(" Compressor Release: %d", release);
|
2012-02-08 19:55:37 +00:00
|
|
|
}
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
if (settings->attack_time != cur_set.attack_time)
|
|
|
|
{
|
|
|
|
logf(" Compressor Attack: %d", attack);
|
|
|
|
}
|
2012-04-26 21:19:16 +00:00
|
|
|
#endif
|
|
|
|
|
|
|
|
curr_set = *settings;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!changed || !active)
|
|
|
|
return active;
|
|
|
|
|
|
|
|
/* configure variables for compressor operation */
|
|
|
|
static const int32_t db[] = {
|
|
|
|
/* positive db equivalents in S15.16 format */
|
|
|
|
0x000000, 0x241FA4, 0x1E1A5E, 0x1A94C8,
|
|
|
|
0x181518, 0x1624EA, 0x148F82, 0x1338BD,
|
|
|
|
0x120FD2, 0x1109EB, 0x101FA4, 0x0F4BB6,
|
|
|
|
0x0E8A3C, 0x0DD840, 0x0D3377, 0x0C9A0E,
|
|
|
|
0x0C0A8C, 0x0B83BE, 0x0B04A5, 0x0A8C6C,
|
|
|
|
0x0A1A5E, 0x09ADE1, 0x094670, 0x08E398,
|
|
|
|
0x0884F6, 0x082A30, 0x07D2FA, 0x077F0F,
|
|
|
|
0x072E31, 0x06E02A, 0x0694C8, 0x064BDF,
|
|
|
|
0x060546, 0x05C0DA, 0x057E78, 0x053E03,
|
|
|
|
0x04FF5F, 0x04C273, 0x048726, 0x044D64,
|
|
|
|
0x041518, 0x03DE30, 0x03A89B, 0x037448,
|
|
|
|
0x03412A, 0x030F32, 0x02DE52, 0x02AE80,
|
|
|
|
0x027FB0, 0x0251D6, 0x0224EA, 0x01F8E2,
|
|
|
|
0x01CDB4, 0x01A359, 0x0179C9, 0x0150FC,
|
|
|
|
0x0128EB, 0x010190, 0x00DAE4, 0x00B4E1,
|
|
|
|
0x008F82, 0x006AC1, 0x004699, 0x002305};
|
|
|
|
|
|
|
|
struct curve_point
|
|
|
|
{
|
|
|
|
int32_t db; /* S15.16 format */
|
|
|
|
int32_t offset; /* S15.16 format */
|
|
|
|
} db_curve[5];
|
|
|
|
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
/** Set up the shape of the compression curve first as decibel values
|
|
|
|
* db_curve[0] = bottom of knee
|
|
|
|
* [1] = threshold
|
|
|
|
* [2] = top of knee
|
|
|
|
* [3] = 0 db input
|
|
|
|
* [4] = ~+12db input (2 bits clipping overhead)
|
|
|
|
*/
|
2012-04-26 21:19:16 +00:00
|
|
|
|
|
|
|
db_curve[1].db = threshold << 16;
|
|
|
|
if (soft_knee)
|
|
|
|
{
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
/* bottom of knee is 3dB below the threshold for soft knee */
|
2012-04-26 21:19:16 +00:00
|
|
|
db_curve[0].db = db_curve[1].db - (3 << 16);
|
|
|
|
/* top of knee is 3dB above the threshold for soft knee */
|
|
|
|
db_curve[2].db = db_curve[1].db + (3 << 16);
|
|
|
|
if (ratio)
|
|
|
|
/* offset = -3db * (ratio - 1) / ratio */
|
|
|
|
db_curve[2].offset = (int32_t)((long long)(-3 << 16)
|
|
|
|
* (ratio - 1) / ratio);
|
2012-02-08 19:55:37 +00:00
|
|
|
else
|
2012-04-26 21:19:16 +00:00
|
|
|
/* offset = -3db for hard limit */
|
|
|
|
db_curve[2].offset = (-3 << 16);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
/* bottom of knee is at the threshold for hard knee */
|
|
|
|
db_curve[0].db = threshold << 16;
|
|
|
|
/* top of knee is at the threshold for hard knee */
|
|
|
|
db_curve[2].db = threshold << 16;
|
|
|
|
db_curve[2].offset = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Calculate 0db and ~+12db offsets */
|
|
|
|
db_curve[4].db = 0xC0A8C; /* db of 2 bits clipping */
|
|
|
|
if (ratio)
|
|
|
|
{
|
|
|
|
/* offset = threshold * (ratio - 1) / ratio */
|
|
|
|
db_curve[3].offset = (int32_t)((long long)(threshold << 16)
|
|
|
|
* (ratio - 1) / ratio);
|
|
|
|
db_curve[4].offset = (int32_t)((long long)-db_curve[4].db
|
|
|
|
* (ratio - 1) / ratio) + db_curve[3].offset;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
/* offset = threshold for hard limit */
|
|
|
|
db_curve[3].offset = (threshold << 16);
|
|
|
|
db_curve[4].offset = -db_curve[4].db + db_curve[3].offset;
|
|
|
|
}
|
|
|
|
|
|
|
|
/** Now set up the comp_curve table with compression offsets in the
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
* form of gain factors in S7.24 format
|
|
|
|
* comp_curve[0] is 0 (-infinity db) input
|
|
|
|
*/
|
2012-04-26 21:19:16 +00:00
|
|
|
comp_curve[0] = UNITY;
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
/** comp_curve[1 to 63] are intermediate compression values
|
|
|
|
* corresponding to the 6 MSB of the input values of a non-clipped
|
|
|
|
* signal
|
|
|
|
*/
|
2012-04-26 21:19:16 +00:00
|
|
|
for (int i = 1; i < 64; i++)
|
|
|
|
{
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
/** db constants are stored as positive numbers;
|
|
|
|
* make them negative here
|
|
|
|
*/
|
2012-04-26 21:19:16 +00:00
|
|
|
int32_t this_db = -db[i];
|
|
|
|
|
|
|
|
/* no compression below the knee */
|
|
|
|
if (this_db <= db_curve[0].db)
|
|
|
|
comp_curve[i] = UNITY;
|
|
|
|
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
/** if soft knee and below top of knee,
|
|
|
|
* interpolate along soft knee slope
|
|
|
|
*/
|
2012-04-26 21:19:16 +00:00
|
|
|
else if (soft_knee && (this_db <= db_curve[2].db))
|
|
|
|
comp_curve[i] = fp_factor(fp_mul(
|
|
|
|
((this_db - db_curve[0].db) / 6),
|
|
|
|
db_curve[2].offset, 16), 16) << 8;
|
|
|
|
|
|
|
|
/* interpolate along ratio slope above the knee */
|
|
|
|
else
|
|
|
|
comp_curve[i] = fp_factor(fp_mul(
|
|
|
|
fp_div((db_curve[1].db - this_db), db_curve[1].db, 16),
|
|
|
|
db_curve[3].offset, 16), 16) << 8;
|
|
|
|
}
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
/** comp_curve[64] is the compression level of a maximum level,
|
|
|
|
* non-clipped signal
|
|
|
|
*/
|
2012-04-26 21:19:16 +00:00
|
|
|
comp_curve[64] = fp_factor(db_curve[3].offset, 16) << 8;
|
|
|
|
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
/** comp_curve[65] is the compression level of a maximum level,
|
|
|
|
* clipped signal
|
|
|
|
*/
|
2012-04-26 21:19:16 +00:00
|
|
|
comp_curve[65] = fp_factor(db_curve[4].offset, 16) << 8;
|
|
|
|
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
/** if using auto peak, then makeup gain is max offset -
|
|
|
|
* 3dB headroom
|
|
|
|
*/
|
|
|
|
comp_makeup_gain = auto_gain ?
|
|
|
|
fp_factor(-(db_curve[3].offset) - 0x4AC4, 16) << 8 : UNITY;
|
|
|
|
|
2012-02-08 19:55:37 +00:00
|
|
|
#if defined(ROCKBOX_HAS_LOGF) && defined(LOGF_ENABLE)
|
2012-04-26 21:19:16 +00:00
|
|
|
logf("\n *** Compression Offsets ***");
|
|
|
|
/* some settings for display only, not used in calculations */
|
|
|
|
db_curve[0].offset = 0;
|
|
|
|
db_curve[1].offset = 0;
|
|
|
|
db_curve[3].db = 0;
|
2012-02-08 19:55:37 +00:00
|
|
|
|
2012-04-26 21:19:16 +00:00
|
|
|
for (int i = 0; i <= 4; i++)
|
|
|
|
{
|
|
|
|
logf("Curve[%d]: db: % 6.2f\toffset: % 6.2f", i,
|
|
|
|
(float)db_curve[i].db / (1 << 16),
|
|
|
|
(float)db_curve[i].offset / (1 << 16));
|
2012-02-08 19:55:37 +00:00
|
|
|
}
|
|
|
|
|
2012-04-26 21:19:16 +00:00
|
|
|
logf("\nGain factors:");
|
|
|
|
for (int i = 1; i <= 65; i++)
|
|
|
|
{
|
2012-04-29 21:31:30 +00:00
|
|
|
DEBUGF("%02d: %.6f ", i, (float)comp_curve[i] / UNITY);
|
|
|
|
if (i % 4 == 0) DEBUGF("\n");
|
2012-04-26 21:19:16 +00:00
|
|
|
}
|
2012-04-29 21:31:30 +00:00
|
|
|
DEBUGF("\n");
|
2012-04-26 21:19:16 +00:00
|
|
|
|
|
|
|
logf("Makeup gain:\t%.6f", (float)comp_makeup_gain / UNITY);
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
#endif
|
2012-04-26 21:19:16 +00:00
|
|
|
|
2012-02-08 19:55:37 +00:00
|
|
|
return active;
|
|
|
|
}
|
|
|
|
|
|
|
|
/** GET COMPRESSION GAIN
|
|
|
|
* Returns the required gain factor in S7.24 format in order to compress the
|
|
|
|
* sample in accordance with the compression curve. Always 1 or less.
|
|
|
|
*/
|
2012-03-27 23:52:15 +00:00
|
|
|
static inline int32_t get_compression_gain(struct sample_format *format,
|
2012-02-08 19:55:37 +00:00
|
|
|
int32_t sample)
|
|
|
|
{
|
2012-03-27 23:52:15 +00:00
|
|
|
const int frac_bits_offset = format->frac_bits - 15;
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
|
2012-02-08 19:55:37 +00:00
|
|
|
/* sample must be positive */
|
|
|
|
if (sample < 0)
|
|
|
|
sample = -(sample + 1);
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
|
2012-02-08 19:55:37 +00:00
|
|
|
/* shift sample into 15 frac bit range */
|
|
|
|
if (frac_bits_offset > 0)
|
|
|
|
sample >>= frac_bits_offset;
|
|
|
|
if (frac_bits_offset < 0)
|
|
|
|
sample <<= -frac_bits_offset;
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
|
2012-02-08 19:55:37 +00:00
|
|
|
/* normal case: sample isn't clipped */
|
|
|
|
if (sample < (1 << 15))
|
|
|
|
{
|
|
|
|
/* index is 6 MSB, rem is 9 LSB */
|
|
|
|
int index = sample >> 9;
|
|
|
|
int32_t rem = (sample & 0x1FF) << 22;
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
|
|
|
|
/** interpolate from the compression curve:
|
|
|
|
* higher gain - ((rem / (1 << 31)) * (higher gain - lower gain))
|
|
|
|
*/
|
2012-02-08 19:55:37 +00:00
|
|
|
return comp_curve[index] - (FRACMUL(rem,
|
|
|
|
(comp_curve[index] - comp_curve[index + 1])));
|
|
|
|
}
|
|
|
|
/* sample is somewhat clipped, up to 2 bits of overhead */
|
|
|
|
if (sample < (1 << 17))
|
|
|
|
{
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
/** straight interpolation:
|
|
|
|
* higher gain - ((clipped portion of sample * 4/3
|
|
|
|
* / (1 << 31)) * (higher gain - lower gain))
|
|
|
|
*/
|
2012-02-08 19:55:37 +00:00
|
|
|
return comp_curve[64] - (FRACMUL(((sample - (1 << 15)) / 3) << 16,
|
|
|
|
(comp_curve[64] - comp_curve[65])));
|
|
|
|
}
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
|
2012-02-08 19:55:37 +00:00
|
|
|
/* sample is too clipped, return invalid value */
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
2012-03-27 23:52:15 +00:00
|
|
|
/** DSP interface **/
|
|
|
|
|
|
|
|
/** SET COMPRESSOR
|
|
|
|
* Enable or disable the compressor based upon the settings
|
|
|
|
*/
|
|
|
|
void dsp_set_compressor(const struct compressor_settings *settings)
|
|
|
|
{
|
|
|
|
/* enable/disable the compressor depending upon settings */
|
|
|
|
struct dsp_config *dsp = dsp_get_config(CODEC_IDX_AUDIO);
|
2013-05-23 17:58:51 +00:00
|
|
|
bool enable = compressor_update(dsp, settings);
|
2012-03-27 23:52:15 +00:00
|
|
|
dsp_proc_enable(dsp, DSP_PROC_COMPRESSOR, enable);
|
|
|
|
dsp_proc_activate(dsp, DSP_PROC_COMPRESSOR, true);
|
|
|
|
}
|
|
|
|
|
2012-02-08 19:55:37 +00:00
|
|
|
/** COMPRESSOR PROCESS
|
|
|
|
* Changes the gain of the samples according to the compressor curve
|
|
|
|
*/
|
2012-03-27 23:52:15 +00:00
|
|
|
static void compressor_process(struct dsp_proc_entry *this,
|
|
|
|
struct dsp_buffer **buf_p)
|
2012-02-08 19:55:37 +00:00
|
|
|
{
|
2012-03-27 23:52:15 +00:00
|
|
|
struct dsp_buffer *buf = *buf_p;
|
|
|
|
int count = buf->remcount;
|
|
|
|
int32_t *in_buf[2] = { buf->p32[0], buf->p32[1] };
|
|
|
|
const int num_chan = buf->format.num_channels;
|
|
|
|
|
2012-02-08 19:55:37 +00:00
|
|
|
while (count-- > 0)
|
|
|
|
{
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
|
|
|
|
/* Use the average of the channels */
|
|
|
|
|
2012-02-08 19:55:37 +00:00
|
|
|
int32_t sample_gain = UNITY;
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
int32_t x = 0;
|
|
|
|
int32_t tmpx = 0;
|
|
|
|
int32_t in_buf_max_level = 0;
|
2012-03-27 23:52:15 +00:00
|
|
|
for (int ch = 0; ch < num_chan; ch++)
|
2012-02-08 19:55:37 +00:00
|
|
|
{
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
tmpx = *in_buf[ch];
|
|
|
|
x += tmpx;
|
|
|
|
labuf[ch][delay_write] = tmpx;
|
|
|
|
/* Limiter detection */
|
|
|
|
if(tmpx < 0) tmpx = -(tmpx + 1);
|
|
|
|
if(tmpx > in_buf_max_level) in_buf_max_level = tmpx;
|
2012-02-08 19:55:37 +00:00
|
|
|
}
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
|
|
|
|
/** Divide it by the number of channels, roughly
|
|
|
|
* It will be exact if the number of channels a power of 2
|
|
|
|
* it will be imperfect otherwise. Real division costs too
|
|
|
|
* much here, and most of the time it will be 2 channels (stereo)
|
2012-02-08 19:55:37 +00:00
|
|
|
*/
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
x >>= (num_chan >> 1);
|
|
|
|
|
|
|
|
/** 1p HP Filters: y[n] = a*(y[n-1] + x - x[n-1])
|
|
|
|
* Zero and Pole in the same place to reduce computation
|
|
|
|
* Run the first pre-emphasis filter
|
|
|
|
*/
|
|
|
|
int32_t tmp1 = x - hpfx1 + hp1y1;
|
|
|
|
hp1y1 = FRACMUL_SHL(hp1ca, tmp1, 7);
|
|
|
|
|
|
|
|
/* Run the second pre-emphasis filter */
|
|
|
|
tmp1 = x - hpfx1 + hp2y1;
|
|
|
|
hp2y1 = FRACMUL_SHL(hp2ca, tmp1, 7);
|
|
|
|
hpfx1 = x;
|
|
|
|
|
|
|
|
/* Apply weighted sum to the pre-emphasis network */
|
|
|
|
sample_gain = (x>>1) + hp1y1 + (hp2y1<<1); /* x/2 + hp1 + 2*hp2 */
|
|
|
|
sample_gain >>= 1;
|
|
|
|
sample_gain += sample_gain >> 1;
|
|
|
|
sample_gain = get_compression_gain(&buf->format, sample_gain);
|
|
|
|
|
|
|
|
/* Exponential Attack and Release */
|
|
|
|
|
|
|
|
if ((sample_gain <= release_gain) && (sample_gain > 0))
|
|
|
|
{
|
|
|
|
/* Attack */
|
|
|
|
if(attca != UNITY)
|
|
|
|
{
|
|
|
|
int32_t this_gain = FRACMUL_SHL(release_gain, attcb, 7);
|
|
|
|
this_gain += FRACMUL_SHL(sample_gain, attca, 7);
|
|
|
|
release_gain = this_gain;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
2012-02-08 19:55:37 +00:00
|
|
|
release_gain = sample_gain;
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
}
|
|
|
|
/** reset it to delay time so it cannot release before the
|
|
|
|
* delayed signal releases
|
|
|
|
*/
|
|
|
|
release_holdoff = delay_time;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
/* Reverse exponential decay to current gain value */
|
|
|
|
{
|
|
|
|
/* Don't start release while output is still above thresh */
|
|
|
|
if(release_holdoff > 0)
|
|
|
|
{
|
|
|
|
release_holdoff--;
|
2012-02-08 19:55:37 +00:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
/* Release */
|
|
|
|
int32_t this_gain = FRACMUL_SHL(release_gain, rlscb, 7);
|
|
|
|
this_gain += FRACMUL_SHL(sample_gain,rlsca,7);
|
|
|
|
release_gain = this_gain;
|
2012-02-08 19:55:37 +00:00
|
|
|
}
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
/** total gain factor is the product of release gain and makeup gain,
|
|
|
|
* but avoid computation if possible
|
|
|
|
*/
|
|
|
|
|
|
|
|
int32_t total_gain = FRACMUL_SHL(release_gain, comp_makeup_gain, 7);
|
|
|
|
|
|
|
|
/* Look-ahead limiter */
|
|
|
|
int32_t test_gain = FRACMUL_SHL(total_gain, in_buf_max_level, 3);
|
|
|
|
if( test_gain > UNITY)
|
|
|
|
{
|
|
|
|
release_gain -= limitca;
|
2012-02-08 19:55:37 +00:00
|
|
|
}
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
|
|
|
|
/** Implement the compressor: apply total gain factor (if any) to the
|
|
|
|
* output buffer sample pair/mono sample
|
|
|
|
*/
|
2012-02-08 19:55:37 +00:00
|
|
|
if (total_gain != UNITY)
|
|
|
|
{
|
2012-03-27 23:52:15 +00:00
|
|
|
for (int ch = 0; ch < num_chan; ch++)
|
2012-02-08 19:55:37 +00:00
|
|
|
{
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
*in_buf[ch] = FRACMUL_SHL(total_gain, labuf[ch][delay_read], 7);
|
2012-02-08 19:55:37 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
in_buf[0]++;
|
|
|
|
in_buf[1]++;
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
delay_write++;
|
|
|
|
delay_read++;
|
|
|
|
if(delay_write >= MAX_DLY) delay_write = 0;
|
|
|
|
if(delay_read >= MAX_DLY) delay_read = 0;
|
2012-02-08 19:55:37 +00:00
|
|
|
}
|
2012-03-27 23:52:15 +00:00
|
|
|
|
|
|
|
(void)this;
|
2012-02-08 19:55:37 +00:00
|
|
|
}
|
|
|
|
|
2012-03-27 23:52:15 +00:00
|
|
|
/* DSP message hook */
|
|
|
|
static intptr_t compressor_configure(struct dsp_proc_entry *this,
|
|
|
|
struct dsp_config *dsp,
|
|
|
|
unsigned int setting,
|
|
|
|
intptr_t value)
|
2012-02-08 19:55:37 +00:00
|
|
|
{
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
int i,j;
|
|
|
|
|
2012-03-27 23:52:15 +00:00
|
|
|
switch (setting)
|
|
|
|
{
|
|
|
|
case DSP_PROC_INIT:
|
|
|
|
if (value != 0)
|
|
|
|
break; /* Already enabled */
|
2012-12-19 22:34:57 +00:00
|
|
|
|
|
|
|
this->process = compressor_process;
|
2013-05-23 17:58:51 +00:00
|
|
|
/* Won't have been getting frequency updates */
|
|
|
|
compressor_update(dsp, &curr_set);
|
2012-12-19 22:34:57 +00:00
|
|
|
/* Fall-through */
|
2012-03-27 23:52:15 +00:00
|
|
|
case DSP_RESET:
|
|
|
|
case DSP_FLUSH:
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
|
2012-03-27 23:52:15 +00:00
|
|
|
release_gain = UNITY;
|
DSP Compressor: Sidechain, Exponential Atk/Rls
This is an improvement to the current compressor which I have added
to my own Sansa Fuze V2 build. I am submitting here in case others
find it interesting.
Features added to the existing compressor:
Attack, Look-ahead, Sidechain Filtering.
Exponential attack and release characteristic response.
Benefits from adding missing features:
Attack:
Preserve perceived "brightness" of tone by letting onset transients
come through at a higher level than the rest of the compressed program
material.
Look-ahead:
With Attack comes clipping on the leading several cycles of a transient
onset. With look-ahead function, this can be pre-emptively mitigated with
a slower gain change (less distortion). Look-ahead limiting is implemented
to prevent clipping while keeping gain change ramp to an interval near 3ms
instead of instant attack.
The existing compressor implementation distorts the leading edge of a
transient by causing instant gain change, resulting in log() distortion.
This sounds "woofy" to me.
Exponential Attack/Release:
eMore natural sounding. On attack, this is a true straight line of 10dB per
attack interval. Release is a little different, however, sounds natural as
an analog compressor.
Sidechain Filtering:
Mild high-pass filter reduces response to low frequency onsets. For example,
a hard kick drum is less likely to make the whole of the program material
appear to fade in and out. Combined with a moderate attack time, such a
transient will ride through with minimal audible artifact.
Overall these changes make dynamic music sound more "open", more natural. The
goal of a compressor is to make dyanamic music sound louder without necessarily
sounding as though it has been compressed. I believe these changes come closer to this goal.
Enjoy. If not, I am enjoying it
Change-Id: I664eace546c364b815b4dc9ed4a72849231a0eb2
Reviewed-on: http://gerrit.rockbox.org/626
Tested: Purling Nayuki <cyq.yzfl@gmail.com>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
2013-10-03 12:57:00 +00:00
|
|
|
for(i=0; i<MAX_CH; i++)
|
|
|
|
{
|
|
|
|
for(j=0; j<MAX_DLY; j++)
|
|
|
|
{
|
|
|
|
labuf[i][j] = 0; /* All Silence */
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Delay Line Read/Write Pointers */
|
|
|
|
int32_t fs = dsp_get_output_frequency(dsp);
|
|
|
|
delay_read = 0;
|
|
|
|
delay_write = (DLY_TIME*fs/1000);
|
|
|
|
if(delay_write >= MAX_DLY) {
|
|
|
|
delay_write = MAX_DLY - 1; /* Limit to the max allocated buffer */
|
|
|
|
}
|
|
|
|
|
|
|
|
delay_time = delay_write;
|
|
|
|
release_holdoff = delay_write;
|
|
|
|
limitca = get_att_rls_coeff(DLY_TIME, fs); /** Attack time for
|
|
|
|
* look-ahead limiter
|
|
|
|
*/
|
2012-03-27 23:52:15 +00:00
|
|
|
break;
|
2013-05-23 17:58:51 +00:00
|
|
|
|
|
|
|
case DSP_SET_OUT_FREQUENCY:
|
|
|
|
compressor_update(dsp, &curr_set);
|
|
|
|
break;
|
2012-03-27 23:52:15 +00:00
|
|
|
}
|
|
|
|
|
2012-12-19 22:34:57 +00:00
|
|
|
return 0;
|
2012-02-08 19:55:37 +00:00
|
|
|
}
|
2012-03-27 23:52:15 +00:00
|
|
|
|
|
|
|
/* Database entry */
|
|
|
|
DSP_PROC_DB_ENTRY(
|
|
|
|
COMPRESSOR,
|
|
|
|
compressor_configure);
|