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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270
|
// This file is part of VLevel, a dynamic volume normalizer.
//
// Copyright 2003 Tom Felker <tcfelker@mtco.com>
//
// This library is free software; you can redistribute it and/or
// modify it under the terms of the GNU Lesser General Public License
// as published by the Free Software Foundation; either version 2.1 of
// the License, or (at your option) any later version.
//
// This library is distributed in the hope that it will be useful, but
// WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
// Lesser General Public License for more details.
//
// You should have received a copy of the GNU Lesser General Public
// License along with this library; if not, write to the Free Software
// Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
// USA
// volumeleveler.cpp - defines the VolumeLeveler class
#include <sys/types.h>
#include <assert.h>
#include <math.h>
#include "volumeleveler.h"
using namespace std;
VolumeLeveler::VolumeLeveler(size_t l, size_t c, value_t s, value_t m)
{
bufs = 0;
SetSamplesAndChannels(l, c);
SetStrength(s);
SetMaxMultiplier(m);
}
VolumeLeveler::~VolumeLeveler()
{
for(size_t ch = 0; ch < channels; ++ch)
delete [] bufs[ch];
delete [] bufs;
}
void VolumeLeveler::SetStrength(value_t s)
{
strength = s;
}
void VolumeLeveler::SetMaxMultiplier(value_t m)
{
if(m <= 0) m = HUGE_VAL;
max_multiplier = m;
}
void VolumeLeveler::SetSamplesAndChannels(size_t s, size_t c)
{
assert(s > 1 && c > 0);
if(bufs) {
for(size_t ch = 0; ch < channels; ++ch)
delete [] bufs[ch];
delete [] bufs;
}
bufs = new value_t*[c];
for(size_t ch = 0; ch < c; ++ch)
bufs[ch] = new value_t[s];
samples = s;
channels = c;
Flush();
}
void VolumeLeveler::Flush()
{
for(size_t ch = 0; ch < channels; ++ch)
for(size_t i = 0; i < samples; ++i)
bufs[ch][i] = 0;
silence = samples;
pos = max_slope_pos = 0;
max_slope = max_slope_val = avg_amp = 0;
}
value_t VolumeLeveler::GetMultiplier()
{
value_t multiplier = pow(avg_amp, -strength);
if(multiplier > max_multiplier) multiplier = max_multiplier;
return multiplier;
}
size_t VolumeLeveler::Exchange(value_t **in_bufs, value_t **out_bufs, size_t in_samples)
{
switch(channels) {
//case 1:
// Exchange_1(in_bufs, out_bufs, in_samples);
// break;
//case 2:
// Exchange_2(in_bufs, out_bufs, in_samples);
// break;
default:
Exchange_n(in_bufs, out_bufs, in_samples);
}
if(silence >= in_samples) {
silence -= in_samples;
return in_samples;
} else {
size_t returned_silence = silence;
silence = 0;
return returned_silence;
}
}
void VolumeLeveler::Exchange_n(value_t **in_bufs, value_t **out_bufs, size_t in_samples)
{
// for each user_pos in user_buf
for(size_t user_pos = 0; user_pos < in_samples; ++user_pos) {
// compute multiplier
value_t multiplier = pow(avg_amp, -strength);
// if avg_amp <= 0, then the above line sets multiplier to Inf, so
// samples will scale to Inf or NaN. This causes a tick on the
// first sample after a Flush() unless max_multiplier is not Inf.
// hopefully this fix isn't too slow.
if(unlikely(avg_amp <= 0)) multiplier = 0;
// untested!
// The advantage of using floats is that you can be
// sloppy with going over 1. Since we have this nifty
// average_amp calculation, let's apply it to limit
// the audio to varying normally below 1. Again,
// hopefully this won't slow things down too much.
if(unlikely(avg_amp > 1)) multiplier = 1 / avg_amp;
// limit multiplier to max_multiplier. max_multiplier can be Inf
// to disable this.
if(unlikely(multiplier > max_multiplier)) multiplier = max_multiplier;
// swap buf[pos] with user_buf[user_pos], scaling user[buf] by
// multiplier and finding max of the new sample
value_t new_val = 0;
for(size_t ch = 0; ch < channels; ++ch) {
value_t in = in_bufs[ch][user_pos];
out_bufs[ch][user_pos] = bufs[ch][pos] * multiplier;
bufs[ch][pos] = in;
if(VLEVEL_ABS(in) > new_val) new_val = VLEVEL_ABS(in);
}
pos = (pos + 1) % samples; // now pos is the oldest, new one is pos-1
avg_amp += max_slope;
if(unlikely(pos == max_slope_pos)) {
// recompute (this is expensive)
max_slope = -HUGE_VAL;
for(size_t i = 1; i < samples; ++i) {
value_t sample_val = 0;
for(size_t ch = 0; ch < channels; ++ch) {
value_t ch_val = VLEVEL_ABS(bufs[ch][(pos + i) % samples]);
if(ch_val > sample_val) sample_val = ch_val;
}
value_t slope = (sample_val - avg_amp) / i;
// must be >=, otherwise clipping causes excessive computation
// TODO: maybe optimize - just save i, then compute slope, pos, and val only once later.
// maybe unlikely()
if(unlikely(slope >= max_slope)) {
max_slope_pos = (pos + i) % samples;
max_slope = slope;
max_slope_val = sample_val;
}
}
} else {
// only chance of higher slope is the new sample
// recomputing max_slope isn't really necessary...
max_slope = (max_slope_val - avg_amp) / ((max_slope_pos - pos + samples) % samples);
// ...but it doesn't take long and has a small effect.
value_t slope = (new_val - avg_amp) / (samples - 1);
// probably needs to be >= for same reason as above
// maybe unlikely()
if(unlikely(slope >= max_slope)) {
max_slope_pos = (pos - 1) % samples;
max_slope = slope;
max_slope_val = new_val;
}
}
}
}
// this code has been proven correct, but not tested much. ;-)
void ToValues(char *in, value_t *out, size_t values,
size_t bits_per_value, bool has_sign)
{
switch(bits_per_value) {
case 32:
if(has_sign) {
for(size_t i = 0; i < values; ++i)
out[i] = ((value_t)((int32_t *)in)[i]) / 2147483648;
} else {
for(size_t i = 0; i < values; ++i)
out[i] = (((value_t)((u_int32_t *)in)[i]) - 2147483648) / 2147483648;
}
break;
case 16:
if(has_sign) {
for(size_t i = 0; i < values; ++i)
out[i] = ((value_t)((int16_t *)in)[i]) / 32768;
} else {
for(size_t i = 0; i < values; ++i)
out[i] = (((value_t)((u_int16_t *)in)[i]) - 32768) / 32768;
}
break;
case 8:
if(has_sign) {
for(size_t i = 0; i < values; ++i)
out[i] = ((value_t)((int8_t *)in)[i]) / 128;
} else {
for(size_t i = 0; i < values; ++i)
out[i] = (((value_t)((u_int8_t *)in)[i]) - 128) / 128;
}
break;
default:
assert(false);
}
}
// note: no clipping, just wrap. I don't know how badly clipping will perform.
void FromValues(value_t *in, char *out, size_t values,
size_t bits_per_value, bool has_sign)
{
switch(bits_per_value) {
case 32:
if(has_sign) {
for(size_t i = 0; i < values; ++i)
((int32_t *)out)[i] = (int32_t)(in[i] * 2147483647);
} else {
for(size_t i = 0; i < values; ++i)
((u_int32_t *)out)[i] = (u_int32_t)((in[i] * 2147483647) + 2147483647);
}
break;
case 16:
if(has_sign) {
for(size_t i = 0; i < values; ++i)
((int16_t *)out)[i] = (int16_t)(in[i] * 32767);
} else {
for(size_t i = 0; i < values; ++i)
((u_int16_t *)out)[i] = (u_int16_t)((in[i] * 32767) + 32767);
}
break;
case 8:
if(has_sign) {
for(size_t i = 0; i < values; ++i)
((int8_t *)out)[i] = (int8_t)(in[i] * 127);
} else {
for(size_t i = 0; i < values; ++i)
((u_int8_t *)out)[i] = (u_int8_t)((in[i] * 127) + 127);
}
break;
default:
assert(false);
}
}
|