Ableton Forum

i think the term "convoluting" applies here: how do you mix 2 sound sources, say a drum track and a vocal track, so that the resulting signal is a combination of the 2: it's not really drums, nor voice, but you can tell that there are drums and voice involved...
is this easily doable with live (if at all).
i guess with reaktor it shouldn't be a problem, but i don't own it.
vokator sould do the trick too, but i don't the result to sound too much like a vocoder.

any ideas/suggestions on the topic?

Convolution is an operation in the frequency domain, ie you need the sound spectrum over time for the audio sources. Basically you modify the frequency spectra of one of the sources with the spectra of the other.

It is a very costly operation in terms of CPU hit, but it can certainly be done in real time.

There are plugs such as ConvoBoy and SIR that convolutes one real time source with a static frequency/delay matrix (i guess) called an impulse response to do really good reverb simulations, but convolution can also be used just to mangle sounds.

SIR: http://www.knufinke.de/sir/index_en.html

A lot of impulse responses:

http://www.xs4all.nl/~fokkie/IR.htm

Edit: There are other ways to blend sounds like you suggest. Vocoding might be one of them. Search for for "sound morphing".

peeddrroo wrote:i think the term "convoluting" applies here: how do you mix 2 sound sources, say a drum track and a vocal track, so that the resulting signal is a combination of the 2: it's not really drums, nor voice, but you can tell that there are drums and voice involved...
is this easily doable with live (if at all).
i guess with reaktor it shouldn't be a problem, but i don't own it.
vokator sould do the trick too, but i don't the result to sound too much like a vocoder.

any ideas/suggestions on the topic?

Fast transform Fourier Synthesis - the old Emax SE and Emax II could do this but the computation took a while - results are similar to a vocoder but not quite the same.

are you using plogue bidule? It's got the FFT hookup and the spectral jimmy jam.

kyma

Prosoniq Morph:

http://products.prosoniq.com/cgi-bin/re ... l&refno=42

does exactly what you are looking for, i think

Yes, FFT may do jobs like these, but results are to my taste often more "electronic" than "organic" (talking about my own tries with building them...)
A more straighforward thing for the drum/voice might be to let the drum amplitude shape the voice and/or the other way around. Some FFT could extract the loadest frequency and control some filter by that, imposing the main frequency to the filtering of the other sound etc.
// C

you can do it with BIAS Peak

also the E-mu samplers had a function called Transform Multiplication
maybe the PCI cards have it too?

Delaydots' Spectral Morpher does exactly this (convolves one sound against another, in real-time) - it comes as part of the Spectral Designers suite. And it's 5$ cheaper than Plogue Bidule (but with Plogue you can build your own convolution wotsit)

http://www.anarchysoundsoftware.co.uk/h ... hythms.htm

something alike.