Some questions around peak (?) detections used in loop tools

[Krakatau]

I was wondering on how the signal's attacks (or peaks, ...how can i defined these audio events correctly ?) in recycle or phatmatik pro, for example, are recognised and isolated by a slice marker inside the whole sample

And what comes to my mind is, would it have a sense for such engines ( peak detectors ) to make such an analyse in the entire frequency bandwith that it could to be capable to recognise some typical instrument's signature

I mean by "typical signature" something typical, for example:

of a wooden percussion

( like a wooden agogo, or other tuned perc like marimba, balafon, etc... )

or a metallic one

( triangle, vibraphone, tubular bells, etc...)

or even some signals that arent percussive-like with slow attack, ( but typical signature in the whole bandwith )

( a human voice, a flute, dunno...)


And deeper in the concept :

Would it be theorically possible to define a, so to say, variable slice position, depending of the frequency response of well-defined signal's attack ? (or peak)

...maybe some kind of resynthesis fonction could do such an accurate operation ?

Just want to know if it sounds like pure science-fiction to you or not ?

[citizenchunk]

one way to accomplish transient tracking is with envelope differentials. setup 2 envelopes--1 is a long averager (RMS, if you like), the other is a quick averager (again, you could use RMS). the differnce between the 2 (which is somewhat like a crest factor) is analyzed to determine whether a transient has occured. bingo bango, transient tracker!

don't know about the other stuff.

[MadBrain]

You mean recognise the input instrument? What use would that be?

[Krakatau]

no, not necessarily the instrument but BTW some type of spectral and transient common factors of an instrument's family

[Krakatau]

...and by recognizing different kind of impacts ( it dynamic response through the entire frequency bandwith ), adapting one way or another the discomposition in separate slices of the original audiofile consequently ( as well as it could maybe create the base of a slice's recomposition method, i may imagine ... )