the ellusive Audio Morph technology

[KingTor]

I've got a client looking for a sound that they think is simple and I can not figure out in my little brain how it is possible given any piece of technology that I've ever seen.

The idea is that you've got one voice like a William Burroughs kinda guy, or Henry Rollins, or whatever. And then you've got another voice, like Gwenyth Paltrow or Winona Ryder or something. And you somehow morph between those two voices. The idea being that at some point in the middle of the transition they sound exactly the same -- as though William Burroughs became Gwenyth Paltrow in mid-sentence. (I'm not sure I want to visualize that one. I think I'd rather see Henry morph into Gwenyth, frankly.)

I've fiddled about with pitch shifting and EQing to sound as similar as possible, but I think it needs something more extreme to match the voices more exactly in the middle. Anyone have any thoughts on what might work for this kind of thing?

[smashannon]

The only way I've ever heard it done is by having the talents read the lines in exact matching cadence and simply cross fading between the two...or cutting on consonants. Your client may be expecting the transition to last longer than what is possible. Pitch shifting will give a mechanical, unnatural effect. I dunno of a box or plug that does it!

Just a thought,

In the "morph area", where the different talent are reading the same lines,you might try using Vocalign to get the rhythm of the reads exact, and then try crossfading once they are tightly matched. I haven't done this myself,mind you, just thinking what I might do were I asked to do what you are describing.
This obviously isn't morphing, though. Anybody know how to really do an audio morph?

Good luck!

[nkf]

Jim Clarke wrote:
>>>Anybody know how to really do an audio morph?<<<

A possibility is to use a Symbolic Sound Kyma. It has spectral analysis and resynthesis. One thing you can do with resynthesised sounds is to morph them.
I had some very good results with it but it takes much time to fine tune everything if you don't want to loose too much sound quality.

Nirto Karsten Fischer

[KingTor]

groovy. thanks for the ideas, everyone.

I tried a test with me reading along to another person, and it's definitely going to take a more exact match than that to make the crossfade thing work. I think I'll try vocalign for that.

That Kyma thing sure looks like quite a tool, but probably not an option for this project. But I'll be checking it out at AES.

[Tweakhead]

You might want to check out the new plug in from Wave mechanics called Speed.

You can create custom pitch transformations over time. This might help to make the female morph to male and vice versa.

[KingTor]

thanks, tweakhead, I'd forgotten that Speed had that feature.

the funny thing is that men's and women's voices aren't that far apart in pitch -- the difference has more to do with overtones and frequency patterns and stuff that's harder to manipulate than pitch. hence the need for something like Kyma to really do it up right.

it's a good thing that the director of this one wants to use a lot of surreal effects and pitch changes, because that makes it easier to hide the fact that I don't have any true morphing technology.

Interestinging, I did a better job of lining up the voices by manually editing than I could get vocalign to do....

[Arno Peeters]

SonicWORX Studio is the tool you're looking for: you really need the manual though, but aside from some study it's very easy to use and produces great morphs.
Download the demo: http://www.sonicworx.de/html/download.html



------------------
Arno Peeters
www.tapetv.nl

[KingTor]

thanks, Arno. it's downloading now. hopefully I'll be able to figure out the demo well enough to do something with it.

Hey Arno,

Yes, thanks for the tip about sonicWorx. I'm going to check out the demo as well!