9

I'm trying to morph a voice into a second one, for an anticipation movie (there's a brief resume of the movie in the beginning of a previous question I answered about how to make futuristic cars here : Futuristic cars )

The protagonist can change his appearance, and choose his voice. So, he's in front of a mirror, speak and by changing parameters he change his voice to another : his alter ego in virtual reality.

In the movie, there's 2 actors, the real and the virtual. The first one's got medium voice, the second got very low-deep voice. During the shooting they just say "Test...Test...Test.... Test" (4 times), and the idea is to morph from one to another,

This is what I (unsuccessfully) tried :

  1. Perfect alignement of source material
  2. Of course crossfade doesn't work : we hear 2 voices
  3. Using Melodyne, I tried to progressively pitch down the first and pitch up the second using the formant tool to make them natural, and tuning them to be in unison.

No, Voices are too sensitive, we perceive that it's not natural...

Any Ideas ? I don't own Kyma and don't have time to learn it, but If a Kyma user want to try it, I can send the 2 files ;-)

If I can't find solutions then I will do the opposite : making them artificial but I thought it was a good (not so obvious) question.

Yohann

  • 1
    Great question - more like this please. – Jay Jennings Jun 3 '14 at 7:49
3

I would recommend to use an effect setup using vocoders. There was also a special plugin once available called Morph from Prosoniq that was designed especially for morphing. Unfortunately its no longer available. On KVR is a thread about similar products and solutions. http://www.kvraudio.com/forum/viewtopic.php?t=305298 I think another solution would be to abandon the idea and use one speaker who changes his voices, that will be more natural.

1

Not sure if this will yield good results or not but you could try using the word "test" from each voice as an impulse response in Altiverb (or similar). Example: Use the low voice "test" as an IR, then process the normal voice "test" through that at varying degrees of wet/dry mix. Through some experimentation you may find that you can achieve your morph over the course of two or three words.

1

I've recently stumbled across this software from the guys at SoundMorph called TimeFlux. Looks like an awesome and intuitive little tool that allows you to morph up to 4 different sound files in interesting ways. I've had no chance to use it myself, yet, as I just came across it this week but it looks like it could give you some very unique results and might be just the thing you're looking for. Not to mention, a very reasonable price range.

http://soundmorph.com/index.php?page=soundpacks&spack=timeflux

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.