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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457
|
class:: SMS
summary:: Spectral Modeling Synthesis
categories:: UGens>Analysis
Description::
An implementation of the sines+noise model first described by Xavier Serra in his 1989 PhD thesis; an input sound is analysed in terms of sinusoidal components by a peak tracking phase vocoder. The error between the sinusoidal reconstruction and the original signal (the residual) is then modeled by a noise model of filtered white noise. The sines part and the noise part are separately resynthesised, allowing independent transformations.
For technical details see:
Xavier Serra and Julius O. Smith (1990) emphasis::"Spectral Modeling Synthesis: A Sound Analysis/Synthesis System Based on a Deterministic plus Stochastic Decomposition"::. Computer Music Journal 14(4): 12--24
note::
this plugin assumes block size of 64 and is optimised for 44100 sampling rate (internally, it uses 1024 point FFTs).
::
Classmethods::
method::ar
argument::input
Audio rate input to be analysed
argument::maxpeaks
Absolute maximum number of allowed peaks to be detected in the spectrum
argument::currentpeaks
Current number of allowed peaks to be detected in the spectrum
argument::tolerance
Search area for matching peaks; within tolerance spectral bins
argument::noisefloor
Minimum magnitude for a candidate peak (measured as spectral magnitude)
argument::freqmult
Resynthesis parameter to change frequency; currently causes a gross multiplication of frequency of all sinusoidal components
argument::freqadd
Resynthesis parameter to change frequency; currently causes a gross addition of a frequency to all sinusoidal components
argument::formantpreserve
Even if changing the frequencies of sinusoidal partial tracks, re-impose the original magnitude spectrum so as to keep the formants (spectral envelope preservation). 0 is off, otherwise on (there is a small performance hit).
argument::useifft
Use IFFT based resynthesis, which is lower quality, but substantially more efficient
argument::ampmult
amplitude multiplier for internal compensation for window power loss within algorithm. Usually leave as default of 1.0.
argument::graphicsbufnum
Will fill a user provided buffer with sines + noise data; the buffer must be size 1 + 513 + 5*(maxsines). The first entry will be the number of sines active for that polled frame. Default for this argument is -1, meaning do not write any status data. See the example with live plotting at the base of this help file.
Examples::
code::
//sine reconstruction left channel, noises on right
(
{
var in, fft, output;
in= SoundIn.ar(0);
output=SMS.ar(in, 50,MouseY.kr(1,50), 8, 0.3); //or freqmult: MouseX.kr(0.5,4)
output
}.play
)
//sum reconstruction channels to simulate effect of straight through
(
{
var in, fft, output;
in= SoundIn.ar(0);
output=SMS.ar(in, 50,MouseY.kr(1,50), 8, 0.3); //or freqmult: MouseX.kr(0.5,4)
output.sum.dup
}.play
)
//just noise residual
d=Buffer.read(s,Platform.resourceDir +/+ "sounds/a11wlk01.wav");
(
{
var in, fft, output;
in=PlayBuf.ar(1,d,BufRateScale.kr(d),1,0,1);
output=SMS.ar(in, 50,50, MouseX.kr(1.0,10.0).round(1.0),MouseY.kr(0.1,20.0,'exponential'), 1.0);
Out.ar(0,Pan2.ar(output[1]));
}.play
)
//frequency multiplication and shift with formant preservation
(
{
var in, fft, output;
in=PlayBuf.ar(1,d,BufRateScale.kr(d),1,0,1);
output=SMS.ar(in, 60,60, 4.0,0.2, MouseX.kr(0.5,4), MouseY.kr(0,1000), 1.0);
Out.ar(0,Pan2.ar(output[0]));
}.play
)
//transient detection via Pitch hasFreq output
//could have freq input and transient detection input to SMS to control rendering
(
{
var in, fft, sines, noise, freq, hasFreq;
//in= SoundIn.ar(0);
in=PlayBuf.ar(1,d,BufRateScale.kr(d),1,0,1);
#freq, hasFreq= Pitch.kr(in);
#sines, noise=SMS.ar(in, 50,50, 8, 1.0, MouseX.kr(0.5,4));
Pan2.ar(sines*(hasFreq.lag(0.01,0.01)) + LPF.ar(noise,MouseY.kr(100,10000,'exponential')),0.0)
}.play
)
//alternative; only pass into SMS if not a transient region
(
{
var in, fft, sines, noise, freq, hasFreq;
//in= SoundIn.ar(0);
in=PlayBuf.ar(1,d,BufRateScale.kr(d),1,0,1);
#freq, hasFreq= Pitch.kr(in);
#sines, noise=SMS.ar(if(hasFreq,in, Silent.ar), 50,50, 8, 1.0, MouseX.kr(0.5,4));
if(hasFreq,Pan2.ar(sines + LPF.ar(noise,MouseY.kr(100,10000,'exponential')), 0.0),0.25*in)
}.play
)
//having fun
(
{
var in, fft, output;
in=PlayBuf.ar(1,d,BufRateScale.kr(d),1,0,1);
output=SMS.ar(in, 60,60, 4.0,0.2, MouseX.kr(0.5,4), MouseY.kr(0.0001,10000,'exponential'), LFNoise0.kr(2,0.5,0.5).round(1));
Out.ar(0,Pan2.ar(AllpassC.ar(output[0],0.05,LFNoise2.kr(10,0.02,0.025),0.5) + CombL.ar(output[1],0.1,0.1,2),0.0));
}.play
)
//having even more fun
(
{
var in, fft, output;
in=SoundIn.ar(0);
output=SMS.ar(in, 60,60, 4.0,0.1, MouseX.kr(0.01,10), MouseY.kr(-1000,1000), 0.0);
Out.ar(0,Pan2.ar(CombL.ar(output[1],0.02,0.02,0.5) + output[0]));
}.play
)
//testing IFFT resynthesis
(
{
var in, fft, output;
in=PlayBuf.ar(1,d,BufRateScale.kr(d),1,0,1);
output=SMS.ar(in, 300,MouseY.kr(1,300), 10.0,0.01, MouseX.kr(0.5,4), 0, 1.0, 1);
Out.ar(0,output);
}.play
)
//just sines
(
{
var in, fft, output;
in=PlayBuf.ar(1,d,BufRateScale.kr(d),1,0,1);
output=SMS.ar(in, 500,500, MouseX.kr(1.0,10.0).round(1.0),MouseY.kr(0.001,20.0,'exponential'), 1.0, 1.0,0,1);
Out.ar(0,Pan2.ar(output[0]));
}.play
)
//having even more fun, this time with IFFT
(
{
var in, fft, output;
in=SoundIn.ar(0);
output=SMS.ar(in, 200,200, 4.0,0.1, MouseX.kr(0.01,10), MouseY.kr(-1000,1000), 0.0, useifft:1);
Out.ar(0,Pan2.ar(CombL.ar(output[1],0.02,0.02,0.5) + output[0]));
}.play
)
//experimenting with ampmult
(
{
var in, fft, output;
in=SoundIn.ar(0);
output=SMS.ar(in, 200,200, 4.0,0.01,MouseY.kr(0.01,100,'exponential'),0, formantpreserve:1.0, useifft:1, ampmult:MouseX.kr(0.0,10.0));
Out.ar(0,output);
}.play
)
/////////////////////////////////////////////////////////////
//drawing out sine trails
b.free
//buffer for getting data back from UGen
b=Buffer.alloc(s,100*5*2+514,1)
//sine reconstruction left channel, noises on right
(
{
var in, fft, output;
in= SoundIn.ar(0);
output=SMS.ar(in, 100,100, 8, 0.05, useifft:1, graphicsbufnum: b.bufnum); //or freqmult: MouseX.kr(0.5,4)
output
}.play
)
//get data back to language
b.getn(0,1514,{|val| Post << val <<< nl; })
//
//frequency is (bin number/512.0) * pi
//1500*172 would be 258000 floats per second passed!
//plot sines (512) only (not plotting noise for now but could do as (128 bands))
(
var data, basedata, updatedata, paintdata;
var xperframe= 3; //5 values
var fps= 172; //172;
var drawaccum=8; //to avoid too much redraw, accumulate this number of FFT frames worth of data, then plot; 8 is good compromise
var updatesperwindow= 40; //22;
var frameperwindow = drawaccum*updatesperwindow; //176//must be divisible by drawaccum //fps = about 172 = 44100/256
var windowx = xperframe*frameperwindow;
var windowy=640;
var freqconvert= (512.0/pi); //(44100/1024.0)* // unit->m_nover2/pi
var freq1, freq2, amp1, amp2, freq, amp;
var temp;
var black, white;
var counter=0; //for scrolling around
var userview;
var drawdataready=false, drawpos=0;
var i, t;
//var drawupdate = drawaccum/
var bufferfetchwait = (1.0/172.0);
var drawupdatex = drawaccum*xperframe;
black= Integer.fromColor(Color.black);
white= Integer.fromColor(Color.white);
data= Array.fill(windowx*640);
basedata= Int32Array.fill(drawupdatex*windowy,{black});
updatedata=basedata.copy;
i = SCImage.new(windowx@windowy);
w = SCWindow.new("SCImage", Rect(0, 100, windowx+10, windowy+10)).front;
w.onClose_({ t.stop; i.free; }); // free the image when the window is closed
userview = SCUserView(w, w.view.bounds)
.relativeOrigin_(false)
//.focusColor_(Color.white.alpha_(0))
//.resize_(5)
.backgroundImage_(i) //, 10
.drawFunc_({arg view;
//image.drawInRect(view.bounds, image.bounds, 2, 1.0);
//userview.refreshInRect(Rect(10+(counter*drawupdatex),10,drawupdatex,windowy));
i.drawInRect(Rect(10+(counter*xperframe),10,drawupdatex,windowy),Rect(counter*xperframe,0,drawupdatex,windowy), 2, 1.0);
});
t= {
inf.do {
b.getn(0,1514,{|val|
var number = val[0].round(1.0).asInteger;
var drawshift= drawpos*xperframe;
//number.postln;
//updatedata=basedata.copy;
if(drawpos<drawaccum, {
number.do{|i|
var pos= 1+(5*i); //5 values
//freq1, freq2, amp1, amp2, phase
freq1 = (freqconvert*val[pos]); //.round(1.0).asInteger.min(511);
freq2 = (freqconvert*val[pos+1]); //.round(1.0).asInteger.min(511);
//freq1 = freqconvert*val[pos];
freq= (freq1+freq2)*0.5;
//warping
//freq= ((((freq1+freq2)*0.5)/511.0)**0.25)*511.0;
//[\freq1, freq1, \original, val[pos]].postln;
freq= 511-((freq).round(1.0).asInteger.min(511));
amp= (((val[pos+2] + val[pos+3])*0.5)*1024).min(1.0);
//[\freq, freq, \amp, amp].postln;
//goes across by rows then columns down page
//i*windowy+freq1
//(i*windowy+freq1).post; " ".post;
//Integer.fromColor(Color.white);
//was = white
xperframe.do{|i| updatedata[drawshift+i+(freq*drawupdatex)]= Integer.fromColor(Color.green(amp));};
//" ".postln;
};
drawpos=drawpos+1;
if(drawpos==drawaccum,{drawdataready= true;});
});
if(drawdataready,{
drawdataready=false;
paintdata= updatedata.copy;
updatedata = basedata.copy;
drawpos=0;
{
counter=(counter+drawaccum)%frameperwindow;
i.setPixels(
paintdata,
Rect(counter*xperframe,0,drawupdatex,windowy)
);
//drawupdatex
//w.refresh
userview.refreshInRect(Rect(10+(counter*xperframe),10,drawupdatex,windowy));
//i.drawInRect(Rect(10+(counter*xperframe),10,xperframe,windowy), Rect(counter*xperframe,0,xperframe,windowy), 2, 1.0); // only a section
}.defer;
});
});
bufferfetchwait.wait;
}
}.fork;
)
::
|