>I wonder why noone hasn't opened a thread for this incredible
>breakthrough in DSP yet..
>
>
>
>
>A guy from germany wrote an algorithm/software that is capable to
>extract single notes (including the tone, e.g. the harmonic content)
>from a multi-timbral recording and modify them in realtime.
>
>In his software it is possible to alterate the pitch7volume of a single
>tone even if multiple notes are playing at the same time.
>
>If someone would have asked me two days ago if this could ever be
>possible I would have said: NO, this is impossible.
>
>This guy unscrambles scrambled eggs!
>
>Listen/View it, and judge for yourself:
>
>http://de.youtube.com/watch?v=jFCjv4_jqAY
>
>Consider my pants shat! I'm sure I'm not the only one who can't beleave
>this.
Hi,
As impressive as it is, I wouldn't shit my pants.
In the 70s they separated Caruso from an orchestra.
The interface looks really convenient for what they are
doing. Anyway, It's an advertisement-clip on youtube .. so before
you have that thing in your own hands, I wouldn't draw
too many conclusions from that.
I don't want to diminish the work, nor the result of what
this guy has achieved. But segmentation in time and separation
of simultaneous events has been done before.
If we can trust the advertisement the quality and the
interface are really impressive.
It would be interesting to know the internals and
details of how the system does the separation and segmentation.
But since it isn't open-source, that might remain his secret for now.
So if you wonder why nobody has opened a thread about this,
what would you want to discuss about it, without knowing what
algorithms they use?
gr.
Bjoern