If you have a problem or need to report a bug please email : support@dsprobotics.com
There are 3 sections to this support area:
DOWNLOADS: access to product manuals, support files and drivers
HELP & INFORMATION: tutorials and example files for learning or finding pre-made modules for your projects
USER FORUMS: meet with other users and exchange ideas, you can also get help and assistance here
NEW REGISTRATIONS - please contact us if you wish to register on the forum
Users are reminded of the forum rules they sign up to which prohibits any activity that violates any laws including posting material covered by copyright
Research ... capture EQ curve
Re: Research ... capture EQ curve
Hello Martin. Care to share this schematic wired up with the real time convolver? 
please and thanks.
please and thanks.
-
francoisreme
- Posts: 29
- Joined: Wed Feb 15, 2017 4:01 pm
Re: Research ... capture EQ curve
Hi Guys.
Martin thanks for this clean and great work !
I'm a bit confused about the convolution part. I'm sure this is maybe covered somewhere else but I can't fint any ressource on that.
I understand that you get an half spectrum from DC to almost nyquist freq (4096 bins) from the output of the FFT anlysis module.
From what I understand, we are macking a average on the squared magnitude, and we finally apply the square root juste before the output of the FFT Anlysis.
But what I can't get is the zero padding part. I mean I get the point of the zero padding because of temporal aliasing in the convolution. But I can't get how taking IFFT of an asymetrical spectrum (which gives complex temporal domain filter kernel), tacking only this IFFT real part, shifting by 1/4 of the size, and keep only the half can be equivalent to making the spectrum symetric before IFFT. I migh be one of these cool DSP hacks but I miss a few ressource to get this one !
does someone get a ressource ?
Also I re downloaded all .fsm from the realtime fft thread, I seems that Martin introduced the idea of meshed convolution in his convolution implementation in this post :
viewtopic.php?f=4&t=3879&hilit=reverb&start=20
But I really can't understand how it works ! Any of you guys got it ?
Thanks !
Martin thanks for this clean and great work !
I'm a bit confused about the convolution part. I'm sure this is maybe covered somewhere else but I can't fint any ressource on that.
I understand that you get an half spectrum from DC to almost nyquist freq (4096 bins) from the output of the FFT anlysis module.
From what I understand, we are macking a average on the squared magnitude, and we finally apply the square root juste before the output of the FFT Anlysis.
But what I can't get is the zero padding part. I mean I get the point of the zero padding because of temporal aliasing in the convolution. But I can't get how taking IFFT of an asymetrical spectrum (which gives complex temporal domain filter kernel), tacking only this IFFT real part, shifting by 1/4 of the size, and keep only the half can be equivalent to making the spectrum symetric before IFFT. I migh be one of these cool DSP hacks but I miss a few ressource to get this one !
does someone get a ressource ?
Also I re downloaded all .fsm from the realtime fft thread, I seems that Martin introduced the idea of meshed convolution in his convolution implementation in this post :
viewtopic.php?f=4&t=3879&hilit=reverb&start=20
But I really can't understand how it works ! Any of you guys got it ?
Thanks !