Hello all,
I've been playing with image convolution filters, and wanted to take a look at the same ideas in audio. I was wondering if someone could point me towards some good resources. I've played a bit with the FFT tutorials, but they don't make too much sense to me yet, I think I need more background. Google hasn't been too yielding, perhaps since I'm not too sure what I'm looking for. Dodge's /Computer Music/ only has two references to convolution, baiscally saying it's really cool, and Soundhack uses it alot.
I put together a little patch to experiment from the FFT examples and my little understanding of convolution, but it's very staticky. Note that it's meant to have the audio fed in through an inlet. The hanning window on the left doesn't do anything (of course), but the one on the right, an attempt at a Prewitt edge finder, does something interesting but staticky.
Any sort of advice would be helpful.
Thanks much, Ian
On Thu, Feb 17, 2005 at 05:37:29PM -0500, Ian Smith-Heisters wrote:
I've been playing with image convolution filters, and wanted to take a look at the same ideas in audio. I was wondering if someone could point
i think the closest audio approximation to image style convolution (eg, where you apply a kernel to get effects like edge detection) is an FIR filter. it's covered by the same theory (edge detection for example is a highpass filter) only applied in one dimension instead of two.
pix.
pix wrote:
On Thu, Feb 17, 2005 at 05:37:29PM -0500, Ian Smith-Heisters wrote:
I've been playing with image convolution filters, and wanted to take a look at the same ideas in audio. I was wondering if someone could point
i think the closest audio approximation to image style convolution (eg, where you apply a kernel to get effects like edge detection) is an FIR filter. it's covered by the same theory (edge detection for example is a highpass filter) only applied in one dimension instead of two.
Yes, that's it.
I actually, by coincidence, found an example patch on the CVS "fft_convolution" that does exactly what the patch I posted earlier does, minus a mistake or two. It's copyrighted 'GG' (Guenter Geiger?). It seems to have the same problem of being very staticky as my patch. Does any one know what causes this? Is there a way to remove it? ([lop~] had little success and muffled things a lot)
Thanks for the pointers and the book and link, Frank.
-Ian
pix.
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://iem.at/cgi-bin/mailman/listinfo/pd-list
Hallo, Ian Smith-Heisters hat gesagt: // Ian Smith-Heisters wrote:
I've been playing with image convolution filters, and wanted to take a look at the same ideas in audio. I was wondering if someone could point me towards some good resources. I've played a bit with the FFT tutorials, but they don't make too much sense to me yet, I think I need more background. Google hasn't been too yielding, perhaps since I'm not too sure what I'm looking for. Dodge's /Computer Music/ only has two references to convolution, baiscally saying it's really cool, and Soundhack uses it alot.
See www.dspguide.com for a good online dsp book, and if you can find it, definitely check out F. Richard Moore's "Elements of Computer Music" for a very good introduction to the whole complex for computer musicians. IMO it is the best of its kind I've read so far, easily surpassing both classics by Dodge/Jerse and Roads, but requiring a bit more math knowledge - which actually is impossible to avoid in that topic.
Ciao
Frank Barknecht _ ______footils.org__
_ __latest track: fqdn _ http://footils.org/cms/show/38
-
Frank Barknecht -
Ian Smith-Heisters -
pix