On Sat, 2006-10-28 at 03:12 +0200, Julian Rohrhuber wrote:
> you could use mbrola / praat.
> it would be a very nice way to do music sometimes..
Yes although if we don't have the need to replicate human speech we have
much wider options for the sounds we can make.
I could do with knowing more about semiotics here, but instead of having
a word that represent a sound, we can have the certain arrangement of
letters within the words represent the structure of a sound, and then
rearrange the letters to make a new sound. Speech synthesis is geared
towards translating words from one domain to another, so when you tell
it to synthisise kknnrnarlgn, it doesn't work. Or does it? My
experiences with the festival speech synth have been disappointing
anyway. Maybe all that needs to be done is removing some of the
limitations from the software...
But again considering the bol processor, if we aren't modelling human
speech we can do things like
kknnnr{aarghh,ooooooh}scshknk!??! , where the aarghh and ooooooh parts
happen at the same time and are combined. Similarly, we could also mark
pivot points to aide their placement in time.
"Extended schwitters" is a good inspiration here (all roads lead to
Hamburg):
http://www.hamburger-kunsthalle.de/archiv/seiten/en_huene.html
And (back to london) the quack project by Ade + co:
http://www.quack-project.com/
Must dash...
alex
Received on Sat Oct 28 2006 - 09:40:33 BST