There is very interesting discussion on the neighbouring thread (http://www.northernsounds.com/ubb/No...ML/000190.html) about surprising amount of RAM one needs when going orchestral (to all of us kind of secret before we bought the GSt and tried). Hey guys, what about polyphony? You think 96 voices (talking GSt 96 here) is a lot? Ready for a bit of maths?
Here we go:
First we need to have a look at legato. If you create (in a sequencer) a legato passage, and make sure that notes do not overlap, Gigastudio still uses voices for both notes on transition.
Second thing to remember is the crossfade feature of the Gigastudio. If you use for instance the XFD versions of instruments in AO these use up twice as much voices of your polyphony for a single note.
Third thing is that, at least in AO, the strings use twice as much of polyphony for a single note (i.e. each note of XFD strings needs 4 voices) than other instruments.
Now, let\'s put this into perspective.
Imagine, that your strings are playing parallel runs (just look into Mahler\'s III. Symphony, I.mov, Nr.51-53) and you use the XFD versions. Only for this 5-voices section you need 5*4*2=40 voices of polyphony!
Let\'s say now, we want to do Mozart\'s Prague Symphony. Fairly moderate orchestration for 6 WWinds, 4 Brass, 5 part Strings and Timpani, i.e. only 16 voices (I pretend the strings don\'t play chords for the moment). However, if you look into bar 131 of the first movement you find two chords where all the instruments are playing and 9 of them (incl. 3 Strings) legato. If you count the polyphony needed on the transition between the two chords (still having the XFD versions) you get to number 65. GSt can still cope but that\'s only Mozart. What about my III. Mahler, scored for more than 40 instruments? Er,....
Of course, I can use the non-crossfade versions, possibly mono etc., there are also ways to work around this by recording sections of the orchestra separately and so on. I am not complaining! After all the GSt enables me, what I would not be able do otherwise.
With my examples I just wanted to illustrate that the orchestra is not only RAM intensive but also polyphony hungry. Hope you managed to follow me.
Good one Ondrej
Nice clear answer for any of us who are used to a bit of synth jargon. I\'ve always put a high priority on polyphony when looking at a piece of gear I want to arrange on. Who cares how many channels and patches the thing has if you can\'t play more than a few parts it before it chokes?
There\'s no question that crossfade/multi velocity/ stereo/ overlapping samples can make 160 voices look like a bottleneck.
I have a feeling that most people also don\'t understand that, even if they aren\'t playing certain notes, Gigasampler loads the beginning of every single sample which is available within a particlular instrument into RAM. So if you load up your big stereo velocity switched percussion instruments patch just to play the cowbell, you may actually be sucking up an inordinate number of the available sample locations.