<html>
<head>
<style><!--
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
font-size: 12pt;
font-family:Calibri
}
--></style></head>
<body class='hmmessage'><div dir='ltr'>Hi List,<br><br>I was wondering if someone might be able to help with a performance issue I am having with ChucK in respect to the use of FFT/IFFT.<br><br>I've written a Pulsar Generator and I've added a convolution function to the pulse train of the generator. I found a previous thread that I've used as a basis for the convolution (see the posting below). The main change was to replace the following line:<br><pre><i>adc => Gain input => FFT fftx => blackhole;</i><br><br>with this one:<br><br><i>SinOsc sin => WinFuncEnv env => Pan2 pan => Gain input => FFT fftx => blackhole;</i><br></pre>At rest, the JACK DSP Load is about 2.2%. When I run the Pulsar Generator at a reasonable rate, the DSP Load jumps to about 3.5% or so without convolution. When I add the convolution function as below, it quickly jumps to a 100% usage with steady Xruns. The sample sizes I am using for the SndBuf statement below are somewhere between 18000 and 28000 samples.<br><br>Is there a strategy anyone could suggest to achieve greater performance of the convolution function below or are there alternative strategies for implementing convolution in general? I'd appreciate any advice.<br><br>+++<br><h1>[chuck-users] Daniel: Convolution (FFT version)</h1>
<b>Perry R Cook</b>
<a href="mailto:chuck-users%40lists.cs.princeton.edu?Subject=Re:%20Re%3A%20%5Bchuck-users%5D%20Daniel%3A%20Convolution%20%28FFT%20version%29&In-Reply-To=%3C1484f078-b1dd-48bf-b974-796b4b7a3c69%40suckerpunch-mbx-7.cs.princeton.edu%3E" title="[chuck-users] Daniel: Convolution (FFT version)">prc at CS.Princeton.EDU
</a><br>
<i>Tue Nov 25 19:09:31 EST 2014</i>
<ul><li>Previous message: <a href="https://lists.cs.princeton.edu/pipermail/chuck-users/2014-November/007808.html">[chuck-users] Daniel: Convolution
</a></li><li>Next message: <a href="https://lists.cs.princeton.edu/pipermail/chuck-users/2014-November/007809.html">[chuck-users] chugins won't compile
</a></li><li> <b>Messages sorted by:</b>
<a href="https://lists.cs.princeton.edu/pipermail/chuck-users/2014-November/date.html#7807">[ date ]</a>
<a href="https://lists.cs.princeton.edu/pipermail/chuck-users/2014-November/thread.html#7807">[ thread ]</a>
<a href="https://lists.cs.princeton.edu/pipermail/chuck-users/2014-November/subject.html#7807">[ subject ]</a>
<a href="https://lists.cs.princeton.edu/pipermail/chuck-users/2014-November/author.html#7807">[ author ]</a>
</li></ul>
<hr>
<pre>FFT version.
Most efficient, lots of delay. Could chunk up
and factor, overlap-add for less delay. This is
the basic idea tho.
// FFT convolution with static impulse response
// by Perry R. Cook, November 2014
// upsides: as efficient as it could be, save for
// constructing a specific fft convolution chugin
// downsides: minimum delay is length of impulse response + buffers
// fix: break into pieces and overlap add
// Other fix: see filter version using my FIR Filter chugin
// our fixed convolution kernal (impulse response)
SndBuf s => FFT ffth => blackhole;
"CelloBodyShort.wav" => s.read; // whatever you like (caution of length!!)
2 => int fftSize;
while (fftSize < s.samples())
2 *=> fftSize; // next highest power of two
fftSize => int windowSize; // this is windowsize, only apply to signal blocks
windowSize/2 => int hopSize; // this can any whole fraction of windowsize
2 *=> fftSize; // zero pad by 2x factor (for convolve)
// our input signal, replace adc with anything you like
adc => Gain input => FFT fftx => blackhole; // input signal
IFFT outy => dac; // our output
fftSize => ffth.size => fftx.size => outy.size; // sizes
Windowing.hann(windowSize) => fftx.window;
// <<< s.samples(), fftSize >>>;
windowSize::samp => now; // load impulse response into h
ffth.upchuck() @=> UAnaBlob H; // spectrum of fixed impulse response
s =< ffth =< blackhole; // don't need impulse resp signal anymore
complex Z[fftSize/2];
1000 => input.gain; // fiddle with this how you like/need
while (true) {
fftx.upchuck() @=> UAnaBlob X; // spectrum of input signal
// multiply spectra bin by bin (complex for free!):
for(0 => int i; i < fftSize/2; i++ ) {
fftx.cval(i) * H.cval(i) => Z[i];        
}
outy.transform( Z ); // take ifft
hopSize :: samp => now; // and do it all again
}
Regards,<br>Mitch <br></pre><br>                                            </div></body>
</html>