Forums

'Alchemical' Forecasting and Restorative Up-Sampling

Started by Winkie 11 months ago10 replieslatest reply 1 month ago158 views

Dear DSPRelated friends:

  • It's been a number of months since I prematurely tried your patience with ideas so far 'out-of-the-box' that I could be paying you for therapy just for listening to my ramblings.
  • But guess what: "I'm Back..." and ready to eat my tail, and risk your wrath, once again.

To review the ideas outlined in the title of this post, just visit: https://deserdi.xyz

Admittedly, a lot to wade through, but I hope it's worth your while.

- Cheers, Winkie. 

[ - ]
Reply by WinkieMarch 16, 2022

Dear DSP'ers:

Have some issues with the MATLAB code I have been working on. For example, with the "Noise Reduction" algorithm, which makes use of almost all of the routines and subroutines, I do have what appears to be a noise reduced signal (in this case of a scratchy 78rpm recording of Johnny Hartman doing 'Wild,' but it also has at least one time-shifted shadow signal of the same track superimposed.) 

I am currently looking into the up-sampling subroutine "spoon.m" - I may have been unintentionally adding to or stretching my IFFT-domain signal representation by one point or so in the middle, resulting in a half period (T) shifted shadow (all approximate guesses, for now). Working on this as able, and will post a note here if and when I can work the problem through (perhaps with your input.) 

-Winkie

[ - ]
Reply by Rick LyonsMarch 17, 2022

Hi Winkie.

In the text at the web link you provided are the words "The Bohmian Up-sampling algorithm {bup.m} can restore high-frequency components lost due to under-sampling, and also undoes aliasing errors introduced by under-sampling."

Can you post the MATLAB code in your bup.m routine here? I'd like to learn how you corrected aliasing errors introduced by undersampling. Thanks.

[ - ]
Reply by WinkieMarch 17, 2022

Well, I found what I think were a couple of mistakes, as I thought, in the 'spoon.m' up-sampling subroutine, which may have accounted for 'shadow' or 'mirage' signals appearing in my outputs, and I made the requisite changes. Not sure when I'll find the time, or, for me, the courage, to check the new output files, but I did make them available on the website. To be honest, I'm looking for supportive individuals to help with a sort of organic pseudo peer-review of this work, or at least the non-speculative, fact-checkable part of the work.  

And Rick, Re: your post above, the {bup.m} algorithm has a number of subroutines it calls, amounting to about seven pages of code, and it doesn't make sense to clutter this site up when all the necessary files are readily available via a Google-doc linked to the 'MATLAB Files Button' on the last 'MATLAB Files' page of the website. The individual routines must be parsed out and saved under the proper file name on the interested parties' computer, but this is do-able. 

The headliner routines then become {fork.m} - forward-forecasting, {bork.m} - backward-forecasting, {bup.m} - Bohmian (restorative) up-sampling, and {qear.m} - Quantization Error and Artifact Reduction. Please note, though, that all headliner routines as well as the subroutines: {forksub.m}, {borksub.m}, and {spoon.m} must be available to MATLAB for any of the units to work properly.

-Cheers, Winkie.

[ - ]
Reply by Rick LyonsMarch 18, 2022

Hi Winkie.

Yesterday when I first went to your web page (web link you provided) I didn't see any MATLAB code. At that time I didn't think about clicking the three horizontal lines at the upper right of the page. Tonight when I did click the three horizontal lines I saw that your MATLAB code was indeed available for review.

Because there was no "Comment" text in your MATLAB code I see now that I'll never be able to understand how it functions or what it does. So Winkie, I'm sorry to have bothered you. Good Luck.

[ - ]
Reply by WinkieMarch 18, 2022

Rick: When it comes to coding - when it comes to alot of things, actually - I am a bit of a scrub. Nice to know that comment text is on your 'wish list'. In one sense, I hadn't included that because I couldn't say where I was going before I got there. 

I just now talked to a 'coder' who works on the graveyard shift at the local gas station where I 'coffee up,' and he, in addition to clarifying the proper urban usage of the term 'scrub,' told me that it is generally easier to write code than it is to read it (and that's why clear comments are necessary) - something I wasn't completely aware of.

I am now in the 'contemplation phase' of adding comments to, and generally cleaning up the MATLAB code on the website, and it occurs that any comments should tackle both the theoretical level of mathematics, as well as the (small/medium) scale of commmand execution - perhaps color coded according to purpose.

Rick, I know it's hard to be jerked around by other people's process, and I hope I haven't lost your (or anyone else's) interest on this.

- Winkie

[ - ]
Reply by WinkieMarch 17, 2022

I should qualify these postings with the disclaimer that I consider myself as only a 'channel' for these ideas coming. I used to view the word 'channel' with open disdain only twenty years ago, yet my introduction to meditation by others more advanced than me, and a halting, but gradual increase in discipline have been largely responsible for these ideas taking form.

I am not sure if these ideas are 'pure' yet, they may require reworking by the engineering, physics and mathematical communities, but my hunch is that, with respectful dialog maintained, they could be of service. If you agree, at least in principle, perhaps you could see your way clear to save a copy of the website as well as the MATLAB code available in Google-docs - and accessible via the 'MATLAB Files Button' on the last page of the website - to your hard drive.

With my mental health history, I can't rule out a measure of paranoia in my psyche, but the website is currently a critical-point in the vulnerability of these ideas "Coming into Being," and if we are dedicated to an impartial quest for truth in a world which plays politics, having the only egg in one basket may not be the way to go - especially when the difficulty is so easily remedied. That said I would like to declare that I consider these ideas, including the MATLAB code I wrote, public domain, though not the sound files I used in their presentation (the music files are not mine to liberate, after all.)

Sorry to be so bizarre, maybe I just feel a little 'weirded out' by some things that happened today. I beg you guys' pardon.

Sincerely, Winkie. 

[ - ]
Reply by WinkieMarch 28, 2022

A friend/mentor whom I consider a great writer told me that the writing on the website was jargon-freighted and sparsely written. He couldn't follow the 'thread' of what I was trying to explain. Extensive editing has, I feel, markedly improved the 'flow' of the whole website and made it not so irritating to jump into. 

The last changes to MATLAB code were made a week ago, and the website sound clips now reflect these changes, although I still haven't documented the code as per Rick Lyons' suggestion. The website makes some ambitious claims, but I'll let you decide if they are warranted. Anyway, sorry to use you guys as my editors as well as my audience... that kind of double duty must be pretty tough.

Cheers, Winkie 

[ - ]
Reply by WinkieMarch 28, 2022

I had saved the Bohmian Up-Sampling test files, 'C' and 'D', I think, with the wrong sampling frequency listed for playback. They are only 3+ seconds long, but the Soundcloud length was 7 or so seconds. As a result they were 'growling' rather than 'purring.' I believe I have corrected this technical mistake. 

Cheers, Winkie

[ - ]
Reply by WinkieMay 22, 2022
Alchemy reportedly involves recursive (and somewhat messy) operations, as the Microcosmic realm of thought is purified and brought into alignment with the "Idea," or pure Platonic Form, as it exists "in the ethers" of the Macrocosmic realm.

If we have finally brought our Vision of Fourier Forecasting and Recursive Up-Sampling to fruition, that has certainly been the case. We wandered so far afield into realms of unnecessary complexity in trying to make the Vision manifest, but each step seemed to be leading someplace vital and potentially important. Finally, after much thought combined with meditation and attentiveness to what seemed Macrocosmic "signals," both from human and non-human realms, the Visions coalesced into much simpler forms than we had been considering.

The website has changed somewhat, in line with suggestions from Forum Members, while still having MATLAB files which reflect the Vision's current embodiment, as well as sound clips which demonstrate inputs and outputs from the MATLAB code. URL remains: https://deserdi.xyz

Thanks for working with me on this, and your tremendous patience with me.
- Winkie
[ - ]
Reply by WinkieDecember 27, 2022

I've cried 'wolf,' or in this case, 'alchemy' so many times on this and the KVR Audio forum that I would not be surprised to have lost 95% of interested parties. Nevertheless, I felt I should report on the last six months' happenings. With regards to band-limited forecasting, the current version of the code uses the negative phantom signal of zeroed out frequencies, which appears wherever the time domain signal would otherwise have been null, combined with a simplified deconvolution technique, along with a few other little 'sleeve-tricks,' in the attempt to accomplish effective forecasting.

Further, this forecasting algorithm is used in a restorative up sampling (RUP) algorithm, wherein the original (N) signal values are laid out on the unit-circle of the complex-plane, coupled with their time-reversal representation, such that the IDFT of the combination is purely real. This IDFT is then forecast out to double length before taking the only real portion of its DFT (actually just to clean up some tiny imaginary remnants), from which the first (2N-1) signal points are harvested.

I am looking to the community to evaluate this work, as looking too closely myself at the results has been like placing a microphone too close to the amplifier speaker, at least in my psyche. "Necessary, but not sufficient" tests have been all I can stomach and use to guide me, at least without a community to provide the buffer of communication.

Band-limited forecasting algorithms seem already to exist in the peer-reviewed literature, and deconvolution was first used for forecasting in WWII, so I don't know why I have a bug up my butt concerning the recapitulation of this previous work, maybe because it could have the potential to add credence to the other writings on the website.

Thanks,
Winkie
https://www.deserdi.xyz

P.S. I have finally provided some documentation for the MATLAB code, sparse as it is.