There is a paper by the same author where identification of an analog compressor unit is performed. In practice, the Polynomial Hammerstein model implements a combination of static nonlinearities and linear filters with memory. So the resulting identified nonlinear system is stateful.sadko4u wrote:CrocoDuck wrote:However, I think that this method doesn't allow to describe stateful system (like which compressor is).
What do we have? What do we need?
Moderators: MattKingUSA, khz
Re: What do we have? What do we need?
-
- Established Member
- Posts: 1067
- Joined: Mon May 12, 2014 7:11 am
- Has thanked: 15 times
- Been thanked: 36 times
Re: What do we have? What do we need?
Do we have something similar to this?
Acoustic feedback plugin.
https://www.softube.com/index.php?id=af
Acoustic feedback plugin.
https://www.softube.com/index.php?id=af
- funkmuscle
- Established Member
- Posts: 2804
- Joined: Mon Jun 02, 2008 2:30 pm
- Has thanked: 130 times
- Been thanked: 32 times
Re: What do we have? What do we need?
oooooooooooooooooooohhhh yeeeeeeaaah!!! that is sweet!! maybe tramp can model that somehow with Guitarix or a gx plugin..rghvdberg wrote:Do we have something similar to this?
Acoustic feedback plugin.
https://www.softube.com/index.php?id=af
I had to move my guitar to the front of my studio monitors just to fake a feedback and still it sucked.
-
- Established Member
- Posts: 1067
- Joined: Mon May 12, 2014 7:11 am
- Has thanked: 15 times
- Been thanked: 36 times
Re: What do we have? What do we need?
Seems like tramp has some homeworkfunkmuscle wrote:oooooooooooooooooooohhhh yeeeeeeaaah!!! that is sweet!! maybe tramp can model that somehow with Guitarix or a gx plugin..rghvdberg wrote:Do we have something similar to this?
Acoustic feedback plugin.
https://www.softube.com/index.php?id=af
I had to move my guitar to the front of my studio monitors just to fake a feedback and still it sucked.
-
- Established Member
- Posts: 1067
- Joined: Mon May 12, 2014 7:11 am
- Has thanked: 15 times
- Been thanked: 36 times
Re: What do we have? What do we need?
https://www.meldaproduction.com/MEqualizer
This eq has a nice feature, harmonics control. Check it out, video explains it at 5:15
https://youtu.be/B8UyyC61ndA?t=312
This eq has a nice feature, harmonics control. Check it out, video explains it at 5:15
https://youtu.be/B8UyyC61ndA?t=312
Re: What do we have? What do we need?
Impossible? How does a kemper profiler work? While you might not be able to perfectly identify the system you can always make a model that reproduces measured responses. With a good idea of the significant nonlinearities of the system, you can get pretty close if not dead on. I'd expect a series of step responses or shaped noise bursts could at least give a good idea of what OTT is doing.sadko4u wrote: In common case it's impossible for non-linear systems.
_ssj71
music: https://soundcloud.com/ssj71
My plugins are Infamous! http://ssj71.github.io/infamousPlugins
I just want to get back to making music!
music: https://soundcloud.com/ssj71
My plugins are Infamous! http://ssj71.github.io/infamousPlugins
I just want to get back to making music!
- sadko4u
- Established Member
- Posts: 987
- Joined: Mon Sep 28, 2015 9:03 pm
- Has thanked: 2 times
- Been thanked: 361 times
Re: What do we have? What do we need?
I told that impossible is to reproduce original signal (before processing) from plugin's output it it's a non-linear device.ssj71 wrote:Impossible? How does a kemper profiler work?sadko4u wrote: In common case it's impossible for non-linear systems.
Profilers work by another scheme: they pass set of primitive signals to the input of the system and get the set of responses. According to some limitations/simplifications this set is enough to model the non-linear device.
LSP (Linux Studio Plugins) Developer and Maintainer.
Re: What do we have? What do we need?
I see. Sorry for the misunderstanding. My point really was that for Rob, if he wants to try to learn more about how OTT works, he'll have to do this sort of profiling as well.
_ssj71
music: https://soundcloud.com/ssj71
My plugins are Infamous! http://ssj71.github.io/infamousPlugins
I just want to get back to making music!
music: https://soundcloud.com/ssj71
My plugins are Infamous! http://ssj71.github.io/infamousPlugins
I just want to get back to making music!
- sadko4u
- Established Member
- Posts: 987
- Joined: Mon Sep 28, 2015 9:03 pm
- Has thanked: 2 times
- Been thanked: 361 times
Re: What do we have? What do we need?
I thought to build profiler that will pass sines to the input of the system and get response (FFT) on the output.ssj71 wrote:I see. Sorry for the misunderstanding. My point really was that for Rob, if he wants to try to learn more about how OTT works, he'll have to do this sort of profiling as well.
Then, by doing FFT of the signal and combining responses, we'll probably get something insteresting. But it's just an experiment, currently I can't guarantee that this will sound well. Also such schema is very CPU-intensive.
LSP (Linux Studio Plugins) Developer and Maintainer.
Re: What do we have? What do we need?
You should totally give a go to the Ant Novak method. It is very simple and, as you noticed, pretty much works as linear system measurements techniques. Just convolve the swept sine output with the inverse filter, window out the higher order responses for negative times, apply a matrix transformation and you got all the Hammerstein Kernels you need. Hammerstein, Wiener and Hammerstein-Wiener models are very general: if your system is not chaotic and does not introduce sub-harmonics and it can be modelled by differential or integro-differential equations then it will be modelled by a Hammerstein, Wiener or Hammerstein-Wiener model (eventually of infinite order).sadko4u wrote: I thought to build profiler that will pass sines to the input of the system and get response (FFT) on the output.
Then, by doing FFT of the signal and combining responses, we'll probably get something insteresting. But it's just an experiment, currently I can't guarantee that this will sound well. Also such schema is very CPU-intensive.
In my dissertation I measured audio amplifiers and used the resulting Hammerstein model to filter audio signals. I made a double blind listening test comparing synthesized signals with signals recorded straight from the amplifiers. I failed to find significant perceptual difference.
I think I will drop an email to my supervisor to ask if I can publish the thesis somewhere online. If I am not wrong the uni has copyrights for a couple of years.
There are many other methods that involve higher order statistics, MLS or similar. They attempt to identify the Volterra Kernels. They are insanely complicated and computationally intensive.
- sadko4u
- Established Member
- Posts: 987
- Joined: Mon Sep 28, 2015 9:03 pm
- Has thanked: 2 times
- Been thanked: 361 times
Re: What do we have? What do we need?
I think, if you have a good DSP basis, we can work on it together.CrocoDuck wrote: You should totally give a go to the Ant Novak method. It is very simple and, as you noticed, pretty much works as linear system measurements techniques. Just convolve the swept sine output with the inverse filter, window out the higher order responses for negative times, apply a matrix transformation and you got all the Hammerstein Kernels you need. Hammerstein, Wiener and Hammerstein-Wiener models are very general: if your system is not chaotic and does not introduce sub-harmonics and it can be modelled by differential or integro-differential equations then it will be modelled by a Hammerstein, Wiener or Hammerstein-Wiener model (eventually of infinite order).
LSP (Linux Studio Plugins) Developer and Maintainer.
Re: What do we have? What do we need?
I think that would be cool! I will send you a PM, so that I don't hijack this thread.sadko4u wrote: I think, if you have a good DSP basis, we can work on it together.
-
- Established Member
- Posts: 1067
- Joined: Mon May 12, 2014 7:11 am
- Has thanked: 15 times
- Been thanked: 36 times
Re: What do we have? What do we need?
O wow!CrocoDuck wrote:I think that would be cool! I will send you a PM, so that I don't hijack this thread.sadko4u wrote: I think, if you have a good DSP basis, we can work on it together.
It's kinda heart warming that a thread I started could result in a collaboration on a new kind of plugin (I'm guessing here, because I didn't understand 99% of what you talked about).
Keep us informed how and what you are doing.
-
- Established Member
- Posts: 188
- Joined: Fri Jul 31, 2015 11:28 pm
- Has thanked: 14 times
- Been thanked: 8 times
Re: What do we have? What do we need?
Ardour as an LV2!rghvdberg wrote:We really need a mame plug-in. Play mame roms in your DAW.