I dont use that interweb expression LOL much/if at all, but I am about to – simply because this actually made me laugh out loud as I observed my brain & perceptions do a little twist inside out. If you have never heard this effect then you may be about to experience the same.
Synthesis via resynthesis is a fairly technical but straight forward process of taking a naturally occurring sound, analysing it & then re-creating it using many sine waves at different pitches. When it is done very high resolution (eg with a Kyma system) it effectively gives you the ability to ‘play’ the spectrum of a sound, but I have never considered applying the approach to dialogue, until now…
Matt Davis of the MRC Cognition & Brain Sciences Unit at Cambridge has a thorough article here about the process & how our brain interprets it, but to illustrate the cognitive effect clearly, listen to this:
sounds like R2D2 right? now listen to this:
ok, now play sine speech 1 again – notice any perceptual shift between now & ten seconds prior? I laughed out loud… sorry LOL’d
Heres a few other examples:
got it?
how about that one?
any different?
Wow, that’s fascinating. Shows just how easily people’s perception depends on what they expect to hear – and re-enforces my belief in the importance of blind listening tests if you want to have any credibility in claims of audio quality perception.
Funny, this also reminded me of my fundamentalist upbringing when ‘backmasking’ was all the rage. We were all warned of the terrors of rock music, because if you played certain songs backwards, Satanic messages could be heard. I remember being mildly skeptical at the time when listening to the so-called examples – surely if Satan wanted to encode hidden messages he’d want to be a bit clearer about it. Anyway, in light of this dramatic you-hear-what-you-want-to-hear demonstration it’s even more obvious how these poor, deluded fools could find the Evil One’s influence in amongst great music. I bet I could find “God sucks” in Handel’s Messiah played backwards if I looked hard enough.
very cool. i can imagine using this idea in a short film, where someone is trying to teach a robot to speak…or something like that.
any idea of some programs outside of a kyma system that does this?
the kind of apps to search for are ones that do spectral analysis
Apart from Kyma the best i have used is AudioSculpt, although you have to pay a yearly subscription to ircam to use it..
http://forumnet.ircam.fr/forum-studio.html?L=1
Metasynth
http://uisoftware.com/MetaSynth
SoundHack
http://www.soundhack.com/
theres also a very good article discussing apps (mac & pc) that let you convert sound <> images
http://emusician.com/software/say-with-pictures/
I am looking for help with determining that speech by parrots is similar to sine speech. The reason I propose this is because the majority of people cannot comprehend creative statements by talking birds.
Mike
Florida
http://www.ParrotSpeech.com