Blame tweiss2000 for summoning me into this thread
Neon Breath wrote:a true morph plugin that would allow a flowing river sound to slowly morph into a barking dog for example.
But that doesn't make much sense to do that though, it's like morphing a picture of someone's face into picture of a guitar, they have no features in common to morph between so it will just look/sound like a blend/crossfade with some deformations going on in between (granted there are different ways to blend sounds if we're talking about morphing, some that might be more interesting than just mixing audio sources), but nothing very coherent. You can morph between things that have something in common like harmonics, but that kind of morphing is just going to sound like some sort of crossfade with pitch-shifting (like in the Kyma video) and some granular thingie if you want to make the sound "freeze" while the transition is going on. You could always morph something with more features than just one set of harmonics though, and by separating those features and morphing them separately have something perhaps interesting.
cron wrote:I think morphing is more a skill than an effect. Obviously the tools play a part, but in the sound examples posted, you can hear that the humans in the recording are somewhat mimicking the sounds to be morphed. For instance, the harpist matches the rhythm of the dog barking. I don't think you could ever make a one-slider process that works in real-time with whatever you throw at it. The system needs some knowledge of what's ahead if it's going to produce a convincing morph trajectory.
The video is quite impressive. Sounds like some kind of pitch-aware granular process from the sounds of it. The morphs often sound like a sliding pitched buzz during the transitions as the granular stream moves from one pitch to the other.
I own CDP and the morph function in that doesn't seem to be pitch aware, so a lot of pre-processing is required to get a convincing result. I listen to Vox 5 by Trevor Wishart which contains some outstanding morphs, all created with an early version of CDP's phase vocoder algorithm in 1986, and it instantly reinforces my view that morphing is a compositional process. The only really great result I've had was morphing the last note of a vocal phrase into a gong in CDP. It wasn't very musically interesting sadly.
All the Kyma buzz I've seen over the last decade seems to be over this morphing effect. It sounds a bit cheesy to me. The clean morph from source to source is a novelty effect you can only really use once or twice before it gets old, and I'm not sure what you'd do with it musically. I'm sure it could be creatively abused, but then I've plenty of methods of spitting broken FFT grids already. Loads of unsubtle, sledgehammer narrative possibilities for people doing sound for film/TV though...
Amen to all of that! You can either dumbly transition between two harmonics using pitch-shifting and blending of the two sounds, or you have to design sounds so that they have more in common, just like you wouldn't morph between two pictures of faces if the faces are at different angles and one is grinning while the other one has their mouth and eyes closed.
Personally I think it's more interesting to morph two sounds into a third new sound than to morph from sound A to sound B over time, the problem with creating a third sound is you need to have sounds that have a lot in common, like two people speaking the same sentence. You could do that using two spectrograms of those two similar sounds and any morphing program like Morpheus (though in the case of voice it might be tricky to get frequencies to align plus if you want to morph noises (siblants, fricatives, plosives, breath etc..) you would have to separate them first and morph them separately) then output the image that is halfway between the two original images then resynthesise that new image in a program like... mmmh let's see, why Photosounder
is the only one I can think of right now!
. But you would lose quality from the resynthesis process anyway. In the case of voice morphing you could probably improve on that using vocoding (modulating an harmonic tone that matches to the frequency of the morphed tone over time with the morphed formants, all of that separated from the noise components of course).
By the way the part about harmonics is important because in an image it's okay if things get deformed a bit, in a sound you have to keep the harmonic ratios intact otherwise it will sound detuned, so in other words you can only morph harmonics using pitch-shifting and a sort of cross-fading. With this in mind I think it could be actually more interesting and simple to morph sounds that are entirely made of noise.
My conclusion is it would require pretty specialised tools and a very good idea of what you're doing to do good sound morphing. I for one wouldn't want to make such a tool.
Last edited by A_SN on Mon Feb 25, 2013 9:58 am, edited 1 time in total.