1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
|
/*
Copyright (C) 2000 Jeff Tranter
tranter@pobox.com
(C) 1999 Stefan Westerfeld
stefan@space.twc.de
(C) 1999 Martin Lorenz
lorenz@ch.tum.de
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 program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
*/
#include <math.h>
#include "artsmodulessynth.h"
#include "stdsynthmodule.h"
using namespace Arts;
class Synth_STD_EQUALIZER_impl : virtual public Synth_STD_EQUALIZER_skel,
virtual public StdSynthModule
{
protected:
float _low, _mid, _high, _frequency, _q;
float tlow, tmid, thigh, tfrequency;
float a1, a2, b0, b1, b2, x_0, x_1, x_2, y_1, y_2;
unsigned long all;
public:
float low() { return _low; }
void low(float newLow)
{
if(newLow != _low)
{
_low = newLow;
calcParameters();
high_changed(newLow);
}
}
float mid() { return _mid; }
void mid(float newMid)
{
if(newMid != _mid)
{
_mid = newMid;
calcParameters();
mid_changed(newMid);
}
}
float high() { return _high; }
void high(float newHigh)
{
if(newHigh != _high)
{
_high = newHigh;
calcParameters();
high_changed(newHigh);
}
}
float frequency() { return _frequency; }
void frequency(float newFrequency)
{
if(newFrequency != _frequency)
{
_frequency = newFrequency;
calcParameters();
frequency_changed(newFrequency);
}
}
float q() { return _q; }
void q(float newQ)
{
if(newQ != _q)
{
_q = newQ;
calcParameters();
q_changed(newQ);
}
}
Synth_STD_EQUALIZER_impl() {
_low = _mid = _high = 0; _q = 0.5;
_frequency = 300;
}
void calcParameters()
{
/*
* _low, _mid, _high are in dB, transform them to tlow, tmid,
* thigh using:
* -6dB => 0.5 ; 0dB => 1 ; 6dB = 2.0 ; ...
*/
tlow = exp(_low * 0.115524530093324); // exp(p[LOW]*ln(2)/6)
tmid = exp(_mid * 0.115524530093324);
thigh = exp(_high * 0.115524530093324);
// _frequency is given in Hz, we need the w-value (and do clipping if
// it exceeds SR/2)
const float SAMPLING_RATE = 44100.0;
tfrequency = _frequency;
if (tfrequency > SAMPLING_RATE / 2.01)
tfrequency = SAMPLING_RATE / 2.01;
float w = 2 * M_PI * tfrequency / SAMPLING_RATE;
// Calculations:
float t = 1/tan(w/2);
float tq = t/_q;
float t2 = t*t;
float a0 = 1+tq+t2;
float a0r = 1/a0;
// and now the real filter values:
a1 = (2 - 2 * t2) * a0r;
a2 = (1 - tq + t2) * a0r;
b0 = (tlow + tmid * tq + thigh * t2) * a0r;
b1 = (2 * tlow -2 * thigh * t2) * a0r;
b2 = (tlow - tmid * tq + thigh * t2) * a0r;
// TODO: try if we need that here, or if we can change filter
// coefficients without setting the state to 0
x_0 = x_1 = x_2 = y_1 = y_2 = 0.0;
all = 0;
}
void streamInit()
{
calcParameters();
}
void calculateBlock(unsigned long samples)
{
all += samples;
if (all > 1024)
{
/* The _problem_: (observed on a PII-350)
*
* I am not quite sure what happens here, but it seems to be like that:
*
* If an ordinary signal (a mp3 for instance) is sent through the
* equalizer, and then no more input is given (zeros as input),
* the y_1 and y_2 values oscillate for some time, coming closer and
* close to zero.
*
* But before the reach zero, they reach the smallest negative number
* (or smallest positive, or whatever), and stay there
* (because 0.005*smallest_negative will remain smallest_negative).
*
* Since then, the CPU usage for all operations on these floats
* increases, (since handling of smallest_negative seems to be a rare
* case).
*
* The _fix_:
*
* We observe the value of y_1. If it's very close to zero (may be as
* well smallest_positive/smallest_negative), we set it to zero,
* together with y_2. This shouldn't significantly influence
* correctness of the filter, but effectively solves the problem.
*
* If you don't believe me, try without this fix and tell me what
* happens on your computer.
*/
const float zero_lower =-0.00000001;
const float zero_upper = 0.00000001;
all = 0;
if(zero_lower < y_1 && y_1 < zero_upper)
y_1 = y_2 = 0.0;
}
unsigned long i;
float tmp;
for (i=0; i<samples; i++)
{
x_0 = invalue[i];
tmp = x_0 * b0 + x_1 * b1 + x_2 * b2 - y_1 * a1 - y_2 * a2;
x_2 = x_1; x_1 = x_0; y_2 = y_1; y_1 = tmp;
outvalue[i] = tmp;
}
}
};
REGISTER_IMPLEMENTATION(Synth_STD_EQUALIZER_impl);
|
