High quality audio morphing effect plugin, why it doesn't exist yet?

[whyterabbyt]

i really don't know what you're trying to say

okay, i'll try again, using the example ive already used.

do you think that a 50%/50% visual crossfade between an image of a sinewave and an image of a squarewave results something which correlates properly with a 50%/50% crossfade between a sinewave and a square wave?

if not, can you tell me what visual manipulation of those two images of audio data you think would correlate with the required crossfade between the two audio datasets?

but if you choose to believe that spectral processing isn't possible, you're free to do so.

nothing ive said correlates even vaguely with that belief.

[monas]

this is not about crossfading but about morphing, which means a gradual interpolation of two spectrograms, or as i previously said an interpolation of spectral features of two sounds. it is a method that does exactly that, but you are free to believe it doesn't work. i don't know what you mean by "visual" and "image" (waveform or spectrogram?) in that context and your example doesn't make much sense to me. i don't really see this discussion going anywhere. i give up.

[whyterabbyt]

this is not about crossfading but about morphing, which means a gradual interpolation of two spectrograms,

oh, ffs

did you read what I was replying to?? someone asked this:

Why should visual processing be misconception?
We are all used to optical representation of sounds, wav view, spectral view...
And why not use the same direction for morphing sounds?

#

Im replying to a question about why visual processing isnt the same as morphing sounds. Do you think visual processing is the same as morphing sounds, or do you agree with me when I say its not?

or as i previously said an interpolation of spectral features of two sounds. it is a method that does exactly that, but you are free to believe it doesn't work.

WHO THE f**k SAID I DONT THINK IT WORKS?

since when does saying 'crossfading pictures of two audio files is not the same as crossfading those two audio files' become a claim that 'spectral proessing doesnt work'.

and I already said I was saying no such thing. seriously are you just f**king trolling or what?

i don't know what you mean by "visual" and "image" (waveform or spectrogram?) in that context and your example doesn't make much sense to me.

you dont know what 'visual' means?

FFS.

i don't really see this discussion going anywhere. i give up.

i really wonder why.

[FabienTDR]

this is not about crossfading but about morphing, which means a gradual interpolation of two spectrograms, or as i previously said an interpolation of spectral features of two sounds.

...which is nothing else what simple cross-fading does: It linearly interpolates two spectra. Perfectly!

So, something is obviously wrong or missing from your definition of audio "morphing", because you're just describing a DJ moving his cross-fader.

Again, for the third time. There is a huge definition problem in this discussion. You can't solve problems you can't describe in words. And to me, it seems everybody has another "belief" of what morphing is. Not a particularly productive or solution oriented context to say the least...

Morphing seems to be a totally subjective fantasy thing. Similar to concepts such as "Squeezifiying", "Cross-Rumbling" or "Changenizing"...

Maybe we need a few PR/marketing pros to explain us what morphing is?

[monas]

this is not about crossfading but about morphing, which means a gradual interpolation of two spectrograms, or as i previously said an interpolation of spectral features of two sounds.

...which is nothing else what simple cross-fading does: It linearly interpolates two spectra. Perfectly!

no it doesn't. crossfading just fades volumes. morphing would interpolate the frequency and amplitude infos from where the mass of spectral energy in one sound is to where it is in the other. in this process it creates information in places where there hasn't been any information before. fading doesn't do that.

[tweiss2000]

Interesting discussion.

May I point to a point, we already discussed in these 11 pages, the definition.

An easy and in my view good definition for fading and morphing:
http://music.columbia.edu/cmc/musicandc ... /05_06.php

What I took out of the last discussion is, that I might loose sound quality, when I just make a screenshot out of a visual representation and use that screenshot for further morphing.

So I would need an all-in-one tool for audio morphing because I might loose to much audio information, if I would use a photo morphing tool.

[deastman]

this is not about crossfading but about morphing, which means a gradual interpolation of two spectrograms, or as i previously said an interpolation of spectral features of two sounds.

...which is nothing else what simple cross-fading does: It linearly interpolates two spectra. Perfectly!

So, something is obviously wrong or missing from your definition of audio "morphing", because you're just describing a DJ moving his cross-fader.

Again, for the third time. There is a huge definition problem in this discussion. You can't solve problems you can't describe in words. And to me, it seems everybody has another "belief" of what morphing is. Not a particularly productive or solution oriented context to say the least...

Morphing seems to be a totally subjective fantasy thing. Similar to concepts such as "Squeezifiying", "Cross-Rumbling" or "Changenizing"...

Maybe we need a few PR/marketing pros to explain us what morphing is?

I've already described what morphing actually is, twice in this thread. It is not simply a linear interpolation between waveforms or between "spectral features". It is the combination of that function and the warping of source and destination sounds/images over time to align features.

Aligning feature sets is like adding warp markers to two sounds in Ableton and gradually pushing and pulling them over time. Some parts will gradually get played faster, some parts will gradually get played slower. The important part is that the source sound is warped over time so its important features gradually start to match those in the destination sound. The reverse warping is applied to the destination sound, and at a 50% morph, the two meet in the middle. This warping happens in addition to a crossfade or "spectral interpolation" between the two sounds.

Metasynth is not capable of doing this. Nor is it capable of resynthesizing a video stream. It can "interpolate" between two static spectrograms, but it cannot warp those spectrograms over time.

My point about not using visual tools to do audio processing was not to imply that Metasynth and Iris aren't fantastic audio processing tools. They are. But you can't just "mash up" RE:Flex (the only real morphing tool for After Effects)with Metasynth and magically invent an audio morphing tool which yields Kyma-like results. Gee, if it was that easy, why didn't anyone think of this before?!?

Not to continue harping on Metasynth (because I think it's a fantastic tool), but just because you could use it to create a basic echo effect, does that mean it is the best way to go about adding a delay line to an audio signal? Challenging audio tasks such as morphing require purpose-built tools, not just a mashing up of tools borrowed from another medium and a quick domain transformation thrown in at the end. Trying to do this with visual tools not designed for audio processing (meaning RE:Flex), even under the best of circumstances, will not preserve the full spectral, noise, and phase components of a sound. Tools specifically built for morphing audio can preserve this information. I *think* this is what Whyterabbyt was getting at.

[FabienTDR]

morphing would interpolate the frequency and amplitude infos from where the mass of spectral energy in one sound is to where it is in the other. in this process it creates information in places where there hasn't been any information before. fading doesn't do that.

It is your *personal* belief of what the term means. One could imagine an unlimited amount of possible definitions.

"morphing" simply has no meaning in the context audio signals. It's a fantasy term. You can interpolate parameters of a instruments, but you can't do this with signals (as long you can't accurately define the technical process without using vague voodoo such a "spectral mass of energy", lol).

You're looking for a re-synthesis algorithm, combined with a parameter automated playback synth (or a few truly creative traditional musicians).

[monas]

morphing would interpolate the frequency and amplitude infos from where the mass of spectral energy in one sound is to where it is in the other. in this process it creates information in places where there hasn't been any information before. fading doesn't do that.

It is your *personal* belief of what the term means. One could imagine an unlimited amount of possible definitions.

"morphing" simply has no meaning in the context audio signals. It's a fantasy term. You can interpolate parameters of a instruments, but you can't do this with signals (as long you can't accurately define the technical process without using vague voodoo such a "spectral mass of energy", lol).

You're looking for a re-synthesis algorithm, combined with a parameter automated playback synth (or a few truly creative traditional musicians).

this isn't voodoo terminology
in spectral processing that's a term that everybody actually understands, just as morphing is, as it was mentioned many times before in this thread. please look up the references, i won't post them again. just that you don't have much experience with spectral processing doesn't mean you have to say it's all BS. why not be open minded and learn something?

@ deastman
throwing in the term "warping" really doesn't help the definition chaos
when i said morphing means "spectral interpolation" i didn't mean the job is done when you do that. that's just how it BASICALLY works. just like pitch shifting in frequency domain BASICALLY means scaling up the fft output. but to perfection it to sound right it takes half a life time of coding (look at celemony).

[deastman]

@ deastman
throwing in the term "warping" really doesn't help the definition chaos

Good point!

[monas]

Interesting discussion.

May I point to a point, we already discussed in these 11 pages, the definition.

An easy and in my view good definition for fading and morphing:
http://music.columbia.edu/cmc/musicandc ... /05_06.php

What I took out of the last discussion is, that I might loose sound quality, when I just make a screenshot out of a visual representation and use that screenshot for further morphing.

So I would need an all-in-one tool for audio morphing because I might loose to much audio information, if I would use a photo morphing tool.

thanks for posting the article! this should clarify things for some people.

here's a simple, free, no-programming-involved way to morph sounds in Csound:

install Csound, and run the editor called csoundQT that comes with it. in the menu bar go to Examples-McCurdy Collection-FFT-pvsmorph.csd
now hit the RUN button, you'll see a simple GUI with two morph sliders for amplitude and frequency parts. there's pre loaded soundfiles, but you can also load your own. both sounds will be looped.

the sound quality of the pvs opcodes is excellent, but it is a very rudimentary version of morphing. a simple interpolation which is "blind". it won't detect where in the spectrum the significant features of a sound are and you don't have an option to set any "markers" either. also you can't really align the files the way you want. but it can still give nice results.

right now i am working on an old project to get additive resynthesis of arbitrary sounds happening in reaktor core. if i get there this should make morphing in reaktor possible and a lot more as well.

[secretkillerofnames]

right now i am working on an old project to get additive resynthesis of arbitrary sounds happening in reaktor core. if i get there this should make morphing in reaktor possible and a lot more as well.

Nice! Good luck!

[pdxindy]

My question is same what you made: why aren't developers such as Camelspace, Fabfilter or Waves interested producing a high quality audio morhing plugin. I don't see this just a niche application at all - new innovations could be included to the morhping idea. H.

They aren't because most likely they think the work to create it is more than the expected return... You have the opinion that it isn't a niche product, but that does not mean you are right...

[pdxindy]

I think the best solution is to use a synthesizer that can morph presets, and then become very good at programming it

Hmmm no I'm afraid a simple preset morph would only volume-cross fade between the two presets, and not truly and harmonically morph by interpolation into each other.

it does not have to only cross fade, it can morph parameter settings too... but that does not mean the result is interesting/useful

[pdxindy]

True. Alchemy is a very capable and flexible synth, yet I also haven't heard any audio examples of one audio file smoothly morphing into another over time, like the Kyma audio demos show.

The thing with the Kyma examples... it is also possible to have thousands of tries with different sounds and you pick out the few that are actually interesting...