What morphing plugins are there and pros cons?

[Neon Breath]

Many years ago Kyma was the king of morphing. These days .. well, it still seems to be the only option for realistic sound morphing.

Indeed. All the pseudo morphing plugins are more like weird vocoder with a cheap sound, very very far away from the genuine deep spectral morphing of Kyma. The closest I found so far is the morph in Alchemy, when using spectral resynthesis. But still weak and very limited. All the rest of the plugins (I’ve tried them all) sound pretty bad.

[fairlyclose]

I can get interesting material from Zynaptiq's Morph. The others I have tried MMorph and Transmutator were not attractive to me. Shame that Kyma is effectively discontinued and/ or they haven't ported the software to PC

[scott_free]

Too bad Kyma needs a dedicated hardware platform which is not for sale anymore.

A new hardware platform (Pacamara?) is coming soon.

[Timfonie]

That's good to know
It totally deserves to be continued.

[CinningBao]

I've been looking at this stuff for a decade or so and yes, all the plugins are just doing some boring phase-vocoding or FFT stuff. Nothing comes anywhere close to the Kyma engine..

Except CDP, which is as old as time itself. I've performed convincing morphs between vocals and an accelerating motorcycle and a few other things, but it isn't easy to use. It's free, and takes a while to get to grips with. But I don't know of any other audio software which will extract a formant over time, and allow you to apply it to another signal from which the formant has been removed. That's where the 'good stuff' is.

https://www.unstablesound.net/cdp.html

does other mental granular and waveset manipulation. One day someone will come along and put front end on it that handles all the file management stuff you have to do to make the most of it. Wasn't Xenakios doing something? I hope one day it comes to fruition.

[fairlyclose]

I've been looking at this stuff for a decade or so and yes, all the plugins are just doing some boring phase-vocoding or FFT stuff. Nothing comes anywhere close to the Kyma engine..

Except CDP, which is as old as time itself. I've performed convincing morphs between vocals and an accelerating motorcycle and a few other things, but it isn't easy to use. It's free, and takes a while to get to grips with. But I don't know of any other audio software which will extract a formant over time, and allow you to apply it to another signal from which the formant has been removed. That's where the 'good stuff' is.

https://www.unstablesound.net/cdp.html

does other mental granular and waveset manipulation. One day someone will come along and put front end on it that handles all the file management stuff you have to do to make the most of it. Wasn't Xenakios doing something? I hope one day it comes to fruition.

I would have thought Zynaptiq did some sort of subspace decomposition. That tends to be where their maths lies.

[CinningBao]

zynaptiq's Morph is just a remake of some old prosoniq tech with a couple of variations; clever FFT/phase-vocoding stuff.. they do have some interesting patents kind of like multi-layer FFT stuff so they can be sample-accurate rather than the minimum granularity of the analysis block (which is what happens as soon as enter the frequency domain)

None of the real-time stuff can do formant extraction, or can understand the rhythmic content over time, which is where the magic is. They're all trying to do 'momentary' analysis and treatment of tiny little bits of sound. The best way to do it is with software which can map out the characteristics of the sounds, and gives the user various options to interpolate smoothly between that data, and that means it (or at least the analysis stage) has to run off-line.

[Unaspected]

Wasn't Xenakios doing something? I hope one day it comes to fruition.

Thank you for your posts above - I was in the dark here.

https://www.youtube.com/watch?v=i7FjwnSddSI

Thanks for the video! Unfortunately this extension plugin will remain in beta in perpetuity but hopefully it is useful in its current state. There's a better CDP frontend in the making but I've been busy with other stuff lately so it hasn't progressed enough to be released to the public in any form yet.

That was six years ago.

EDIT: Ah. People can't seem to find the script... Maybe someone here has a link or download.

[Xenakios]

EDIT: Ah. People can't seem to find the script... Maybe someone here has a link or download.

I have removed all Reaper related downloads from public distribution. But if the stuff happens to be floating around somewhere, you can of course attempt using it. Fully at your own risk.

[Xenakios]

Wasn't Xenakios doing something? I hope one day it comes to fruition.

I have given up trying to do the front end for CDP. The CDP programs are just too finicky with the parameter settings and input data. The processings fail very often because CDP thinks something is wrong with the settings. And even when they do accept the parameter values/input data, they tend to crash and hang a lot. It would just make for a very poor user experience to slap a modern GUI on top of that ancient code.

[Audiority]

I had a Kyma about 10 years ago (when I was working on launching Epic SoundLab) and its morphing is unparalleled, so far, mainly because you need to do A LOT of work between the samples you want to morph between. It's not like you select the sample and do the morphing. You have to perform several analysis on each sample, map the crucial points that will be used a morphing "anchors" between samples. It was hours of work to just morph between four samples. If you need that specific kind of morphing, that's the best option. If you want to just play with morphing, there are a other valid solutions around.

My 2 cents,
Luca

[kidslow]

Another approach, and one I have used extensively, is NI Kore 2, which allows you to take presets from a plugin or even group of plugins and morph between them, again not using crossfading but in this case interpolation of automatable parameters

Is that similar to what Oli Larkin's pMix v2 does? https://github.com/olilarkin/pMix2

[Unaspected]

EDIT: Ah. People can't seem to find the script... Maybe someone here has a link or download.

I have removed all Reaper related downloads from public distribution. But if the stuff happens to be floating around somewhere, you can of course attempt using it. Fully at your own risk.

That sounds ominous. If I chance across it I'll try it on my laptop rather than my main PC.

It's a shame that the project couldn't continue but I totally understand the need to identify realistic goals and prioritise.

Thank you for the feedback!

[CinningBao]

Wasn't Xenakios doing something? I hope one day it comes to fruition.

I have given up trying to do the front end for CDP. The CDP programs are just too finicky with the parameter settings and input data. The processings fail very often because CDP thinks something is wrong with the settings. And even when they do accept the parameter values/input data, they tend to crash and hang a lot. It would just make for a very poor user experience to slap a modern GUI on top of that ancient code.

Oh man, that is a real shame, but I don't blame you if it's that difficult to get a stable GUI. The DSP is delightful, much of which hasn't been seen outside the CDP framework, but, having used it, I can completely understand how much of a nightmare it might be trying to turn the manual process into something more usable.

Sorry to the OP for taking this slightly more OT.. Is there any distance between the file/folder layer of interaction with CDP and the DSP code 'which does the work'? I guess I'm asking, is there another way to approach making CDP a more user-friendly experience which might occur to another programmer? Some way of stablising the success of the operations, or converting the DSP layer itself into something more 'future-facing'?

[Xenakios]

Oh man, that is a real shame, but I don't blame you if it's that difficult to get a stable GUI.

is there another way to approach making CDP a more user-friendly experience which might occur to another programmer? Some way of stablising the success of the operations, or converting the DSP layer itself into something more 'future-facing'?

The GUI itself could be made to work stable, since the CDP programs would be run as external processes anyway. But the problem is more with all the error handling and reporting needed...

Also making the processings more usable tends to require managing lots of temporary audio files. For example the processes that only process as mono, need stereo/surround inputs split into separate mono files etc...Sometimes the temporary files can't be successfully deleted, and I found that a rather irritating issue. (Of course someone with more patience to deal with it all, could probably figure it all out.)

The DSP code in CDP is ancient style C code spread across dozens, if not hundreds, of code files. Combing through all that and trying to stabilize and modernize it all would be a massive effort. It would probably be easier to just write all the interesting processes from scratch as modern C++ code, also with hooks to enable direct realtime playback/preview. But this is not a job for me, I've shifted priorities to other areas in programming. Some of the CDP stuff would be nice have working in other contexts too, though...

There is the project by Logan McBroom that has rewritten/reimagined versions of some of the CDP things. But I haven't followed that much lately.