Convolution2: kernel size is truncated to fftsize

[joslloand]

Input kernel size is truncated to fftsize.

There should probably be a warning in the documentation that this is the case.

// -----------------------------------------------
// Test: Convolution2, kernel size is truncated to fftsize
//
// OS: OSX 10.8
// SC: 3.6.2
// SR: 96000
Server.default = s = Server.local.boot;

// number of frames for buffer (kernel)
~numFrames = 2**10; // 1024

// frames size of FFT
~frameSize = 29; // 512 - bad!
// ~frameSize = 210; // 1024 - good!

// set up buffer (kernel)
~buffer = Buffer.alloc(s, ~numFrames);
~buffer.zero;

// add an impulse @ gain
~gain = 0.dbamp;
~sampleIndex = ~numFrames - 1;

// create impulse at sampleIndex
~buffer.set(~sampleIndex, ~gain);

(
// inspect
"\nnumFrames = ".post; ~buffer.numFrames.postln;
"frameSize = ".post; ~frameSize.postln;

// test!
~synth = {
Convolution2.ar(
Saw.ar(110.0, -18.dbamp),
// PinkNoise.ar(-6.dbamp),
~buffer,
framesize: ~frameSize
)
}.play;
)
~synth.free;
~buffer.free;
s.quit;

[danstowell]

This OK? (patch attached)

2013/2/26 Joseph Anderson [email protected]

Input kernel size is truncated to fftsize.

There should probably be a warning in the documentation that this is the
case.

// -----------------------------------------------
// Test: Convolution2, kernel size is truncated to fftsize
//
// OS: OSX 10.8
// SC: 3.6.2
// SR: 96000
Server.default = s = Server.local.boot;

// number of frames for buffer (kernel)
~numFrames = 2**10; // 1024

// frames size of FFT
~frameSize = 29; // 512 - bad!
// ~frameSize = 210; // 1024 - good!

// set up buffer (kernel)
~buffer = Buffer.alloc(s, ~numFrames);
~buffer.zero;

// add an impulse @ gain
~gain = 0.dbamp;
~sampleIndex = ~numFrames - 1;

// create impulse at sampleIndex
~buffer.set(~sampleIndex, ~gain);

(
// inspect
"\nnumFrames = ".post; ~buffer.numFrames.postln;
"frameSize = ".post; ~frameSize.postln;

// test!
~synth = {
Convolution2.ar(
Saw.ar(110.0, -18.dbamp),
// PinkNoise.ar(-6.dbamp),
~buffer,
framesize: ~frameSize
)
}.play;
)
~synth.free;
~buffer.free;
s.quit;

—
Reply to this email directly or view it on GitHub.

http://www.mcld.co.uk

[joslloand]

Hello Dan,

Thanks for this! Sounds very plausible, thanks for such a quick solution. I've just spoken Josh, who has said he'll make a compile this evening and forward to me for a look.

I'll let you know when all is confirmed.

My best,
Jo

Joseph Anderson

http://joseph-anderson.org

On 26 Feb 2013, at 1:25 pm, danstowell [email protected] wrote:

This OK? (patch attached)

2013/2/26 Joseph Anderson [email protected]

Input kernel size is truncated to fftsize.

There should probably be a warning in the documentation that this is the
case.

// -----------------------------------------------
// Test: Convolution2, kernel size is truncated to fftsize
//
// OS: OSX 10.8
// SC: 3.6.2
// SR: 96000
Server.default = s = Server.local.boot;

// number of frames for buffer (kernel)
~numFrames = 2**10; // 1024

// frames size of FFT
~frameSize = 29; // 512 - bad!
// ~frameSize = 210; // 1024 - good!

// set up buffer (kernel)
~buffer = Buffer.alloc(s, ~numFrames);
~buffer.zero;

// add an impulse @ gain
~gain = 0.dbamp;
~sampleIndex = ~numFrames - 1;

// create impulse at sampleIndex
~buffer.set(~sampleIndex, ~gain);

(
// inspect
"\nnumFrames = ".post; ~buffer.numFrames.postln;
"frameSize = ".post; ~frameSize.postln;

// test!
~synth = {
Convolution2.ar(
Saw.ar(110.0, -18.dbamp),
// PinkNoise.ar(-6.dbamp),
~buffer,
framesize: ~frameSize
)
}.play;
)
~synth.free;
~buffer.free;
s.quit;

—
Reply to this email directly or view it on GitHub.

http://www.mcld.co.uk
—
Reply to this email directly or view it on GitHub.

[danstowell]

Fixed in commit e978406

[joslloand]

Thanks Dan! Updating the help clarifies this!!