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

[Neon Breath]

For those who find that Kyma's morphs sound bad in the earlier example I posted, I suggest you to listen to these commercials created by the sound designer Pete Johnston, all featuring audio morphs with Kyma:

http://www.bantusound.com/SoundMorphing ... gPage.html

I haven't found any effects (or even soft synths known for performing some morphing like Alchemy) sounding as smooth and effective as these examples.

Maybe it's just shitty mp3 files, but not finding these samples particularly high quality. Radio quality, sure. High quality, not so much. Clever sound design, though.

Maybe. But still way higher than any other effects or soft synths recommended in this thread so far though.

[mandolarian]

For those who find that Kyma's morphs sound bad in the earlier example I posted, I suggest you to listen to these commercials created by the sound designer Pete Johnston, all featuring audio morphs with Kyma:

http://www.bantusound.com/SoundMorphing ... gPage.html

I haven't found any effects (or even soft synths known for performing some morphing like Alchemy) sounding as smooth and effective as these examples.

Maybe it's just shitty mp3 files, but not finding these samples particularly high quality. Radio quality, sure. High quality, not so much. Clever sound design, though.

Maybe. But still way higher than any other effects or soft synths recommended in this thread so far though.

True enough, but that ain't much praise to my ears. Ping me when these tools evolve past the carbon mic phase.

[Neon Breath]

True enough, but that ain't much praise to my ears. Ping me when these tools evolve past the carbon mic phase.

Hehehe, exactly! Which return us right to my very first point: why no high quality audio morphing effect plugin yet? It's definitly a path still wild & quite unexplored.

[valhallasound]

True enough, but that ain't much praise to my ears. Ping me when these tools evolve past the carbon mic phase.

Hehehe, exactly! Which return us right to my very first point: why no high quality audio morphing effect plugin yet? It's definitly a path still wild & quite unexplored.

From the perspective of a guy who took his first computer music class in 1988, there has been quite a bit of exploration of sonic morphing. Most of the exploration took place during the 1990s, to such a degree that the word "morph" brings back memories of the Clinton era, that Snoop Dogg video from 1993, OS9, and horrible green rackmount boxes with the word "phatt" in the title. There aren't many morphing plugins, but most of the morphing exploration took place before the time of plugins.

EDIT: http://en.wikipedia.org/wiki/Mighty_Mor ... er_Rangers

Sean Costello

[monas]

i think for now you'll just have to retreat to esoteric stuff like csound/kyma for this. it's the old problem: the industry is just too busy recreating analogue gear and putting new eqs/compressors/reverbs on the market.

maybe something will be in the next GRM bundle. (which will probably be 10 years from now, looking at how slow development goes there)


also, what just comes to my mind: have you tried Hadron from Partikkel Audio?

http://www.partikkelaudio.com

It is a granular plugin based on Csounds partikkel opcode. The basic version is for free, and it is what i think comes closest to the Kyma examples. I haven't explored it very much yet though

[deastman]

Image morphing is a combination of two simultaneous processes: crossfading from one image to the other, and warping the images so that features in one are aligned with similar features in the other.

Kyma does the same thing. Using the Tau editor, you add warp markers to specific features in one sound file, and then match those up with equivalent features in the other sound. This allows the sounds to stretch into position as the system gradually crossfades between them.

I actually think Celemony would be one of the best candidates for making a native morphing tool. They already have some of the best warping tools around, which is what really makes Kyma stand out from the competition. All that is missing is the correlation between features and a realtime performance interface. I'm sure it will never happen, but they could easily do it if they felt so inclined.

[Neon Breath]

From the perspective of a guy who took his first computer music class in 1988, there has been quite a bit of exploration of sonic morphing. Most of the exploration took place during the 1990s, to such a degree that the word "morph" brings back memories of the Clinton era, that Snoop Dogg video from 1993, OS9, and horrible green rackmount boxes with the word "phatt" in the title. There aren't many morphing plugins, but most of the morphing exploration took place before the time of plugins.

EDIT: http://en.wikipedia.org/wiki/Mighty_Mor ... er_Rangers

Sean Costello

Yeah well, I meant a path still unexplored in the realm of the plugins. It would be quite interesting to see a brand new & fresh morphing plugin created by a solid developer

[djscorb]

I agree with what some others have said. It would be pretty much impossible to make something like this in a real time plug in. The process needs human input and a look at the whole files in advance. Even with this you are unlikely to get a convincing morph between any arbitrary sounds.

Imagine you were trying to morph a 5 year old scottish girl's voice into a 70 year old American's. Even if you got them to say the same thing with the same timing, how would a real time plugin know when the sounds are in sync to allow a smooth morph? Even if it was in sync, the accents and dialects have very different vowell sounds, what happens when you morph between two formants? I bet you don't get a coherent vowell!

What happens if you morph between something consisting of wideband noise and a pure sine like bell tone. There is no common feature to morph and it just wouldn't sound like morphing.

There are so many cases where morphing wouldn't work, which suggests to me at least that any useful effect from it is going to be down to careful sound selection and some skill on the behalf of the engineer as he chooses the mappings, timing and pitch markers of such a process.

The best you could hope for is a plugin in which you load your two sounds in advance and can then set up what source features are mapped onto the target. You could use warp markers, pitch mappings and transission points but again, the order and rate of change of these mappings will work better for some sounds than others and only the engineer can determine what works best for the material or the effect he is trying to achieve.

I think the best way to get these effects is currently in synth parameter morphing. Synths are better at creating dog barks and mimicking real life sounds than software currently is at morphing between real audio sources.

I also think that if it was possible to do in real time we would already have it as the applications are broad. You could half morph a vocal to sound more like mariah carey etc. The abuse of autotune would pale into insignificance. Maybe it's not such a bad thing that we don't have it

[KungKrille]

Image morphing is a combination of two simultaneous processes: crossfading from one image to the other, and warping the images so that features in one are aligned with similar features in the other.

Kyma does the same thing. Using the Tau editor, you add warp markers to specific features in one sound file, and then match those up with equivalent features in the other sound. This allows the sounds to stretch into position as the system gradually crossfades between them.

I don't own CA Alchemy anymore but I think you add warp markers in there too.

[monas]

I also think that if it was possible to do in real time we would already have it as the applications are broad.

I couldn't disagree more. I think the industry is leaving a lot of possibilities unexplored.

I think it is totally possible to morph any two sound files with the right high quality resynthesis algos and a proper interface. some kind of manual mapping of spectral features would be necessary and you'd have to choose the right source sounds. of course it would take practice and skill to use it right, but that goes for ANY plugin.

celemony would definitely be a candidate for it, but the interface isn't for this kind of usage. they keep promoting their software for "creative sound design" as well, but the whole thing really is set up as a tool to fix up vocals. I've never found it very inviting for anything else.

also zynaptiq might be up to something:
http://www.zynaptiq.com/pitchmap/
it's very expensive stuff tough

Of course you can also morph things by interpolating parameters in physical modelling, but i think that's not really what this thread is about, unless you find a way to resynthesize an arbitrary sound in a physical modelling synth by just feeding it into it.

Alchemy is always mentioned, but i never tried it for those purposes. I keep hearing that the spectral resynthesis doesn't sound very good. can someone maybe post examples of alchemy morphing?

[Harry_HH]

I agree with what some others have said. It would be pretty much impossible to make something like this in a real time plug in. The process needs human input and a look at the whole files in advance. Even with this you are unlikely to get a convincing morph between any arbitrary sounds.

Imagine you were trying to morph a 5 year old scottish girl's voice into a 70 year old American's. Even if you got them to say the same thing with the same timing, how would a real time plugin know when the sounds are in sync to allow a smooth morph? Even if it was in sync, the accents and dialects have very different vowell sounds, what happens when you morph between two formants? I bet you don't get a coherent vowell!

What happens if you morph between something consisting of wideband noise and a pure sine like bell tone. There is no common feature to morph and it just wouldn't sound like morphing.

There are so many cases where morphing wouldn't work, which suggests to me at least that any useful effect from it is going to be down to careful sound selection and some skill on the behalf of the engineer as he chooses the mappings, timing and pitch markers of such a process.

The best you could hope for is a plugin in which you load your two sounds in advance and can then set up what source features are mapped onto the target. You could use warp markers, pitch mappings and transission points but again, the order and rate of change of these mappings will work better for some sounds than others and only the engineer can determine what works best for the material or the effect he is trying to achieve.

I think the best way to get these effects is currently in synth parameter morphing. Synths are better at creating dog barks and mimicking real life sounds than software currently is at morphing between real audio sources.

I also think that if it was possible to do in real time we would already have it as the applications are broad. You could half morph a vocal to sound more like mariah carey etc. The abuse of autotune would pale into insignificance. Maybe it's not such a bad thing that we don't have it

I think we are in this thread talking about more than one thing, just partly ovarlapping issues, but which, if we do not separate, can lead to not-that- fruitful-debate. Some (those who may know more about technology) try to figure out and describe some kind of "morhping theory" and they can list number of reasons why this complete final real-time morphing plugin is not possible to achieve. Others, such as myself, will just send a POSITIVE signal to developers, "please research this area", "we'd like to see this kind of application", I trust that by r&d manufacturers find new innovations/implications behind/beside the theory we do not know yet, and more importantly, new applications to the existing theory, if they just spend time with this app. dev. I think is rather useless in advance to judge the application impossible. As said, for me is enough, for starters, if I got a plugin like the Prosonic Morph for Windows (wich doesn't exist today), and as the next step maybe a new more innoavative interface and maybe better sound quality etc. etc. Very often the breakthrough innovations are more new application of the existing theory than any new sicientific findings - e.g. main innovation in tablet/finger computer is in the new kind of user experience, the touch screen technology has existed relative long time. H.

[Naampie2000]

You can do this with fft, morphing to the nearest neighbour.
Just would be a PITA.

[VariKusBrainZ]

Kyma does the same thing. Using the Tau editor, you add warp markers to specific features in one sound file, and then match those up with equivalent features in the other sound. This allows the sounds to stretch into position as the system gradually crossfades between them.

Are you sure its crossfading?
I have a Kyma and am sure its not.

List of features
In the Tau Editor you can:

•Time-align multiple files
•Match or transform the frequency, amplitude, and formant envelopes of files •Morph across two, three or more sounds
•Record your real time transformations on disk for later editing
•Create an entire library of synthesis patches at the touch of a button (Galleries)
•Test the mix and morphing combinations in real time as you are editing
•Analyze a new sound by simply dragging it and dropping it into the editor
•Morph without time-smearing or "robotic" comb filtering effects
•Edit with complete freedom, knowing that you have unlimited undos

[VariKusBrainZ]

This has been around for years

http://www.composersdesktop.com/
http://people.bath.ac.uk/masjpf/CDP/sndexs/exmph1mo.htm

I just noticed that Richard Orton has passed away

[valhallasound]

This has been around for years

http://www.composersdesktop.com/
http://people.bath.ac.uk/masjpf/CDP/sndexs/exmph1mo.htm

Trevor Wishart has written a book, Audible Design (published in 1994), that has LOTS of discussion about morphing between different sounds using the techniques available in CDP.

Sean Costello