soundfonts?

[fake]

Hi.

When creating soundfonts, how many keys should you sample and which ones?

Any help would be appreciated
Ta!

fake

[sonicfire]

i´m always sampling C, D#, F#, A and that for every octave
but why you want sample soundfonts?

[Jafo]

sonicfire has it right -- c d# f# and a (using awave studio too, eh?) will let you limit transpositions to one semitone.

you can get by with just two samples per octave, perhaps c and f#, but four is a very workable compromise.

if you really want the extra quality, and don't mind the extra resource drain, sample for every note, in three or four velocity layers. that's what i'm doing with the uiowa mis samples, and the results are excellent. i'm automagically cutting the samples into individual notes, and using awave studio to assemble (and loop) them into soundfonts. i should have most of them finished before the demo period runs out

cheers.

-- just another f*cking observer

[sonicfire]

jodo: nope, i tried the demo sometimes, but didn´t found a use for it (these days i didnt used soundfounts so much). but this sound interesting! awave does that "automagically"??

[fake]

i´m always sampling C, D#, F#, A and that for every octave
but why you want sample soundfonts?

Thanks for the info.

Just mucking about with chainer.

Stacking some vst's etc. Made a nice one with crazy diamonds and cheese machine, thought it would be less cpu intensive just using one sampler. Plus, I can now "grab" some other stuff. I can now create a soundfont from a fav rompler patch or drumatic patch etc

fake:D

[Jafo]

heh. awave does little automagically. i'm using various little programs with names like "waveknife" and "autoloop" to do the stuff.

i really ought to just use sfz, but awave's autolooper is so good and so convenient -- it works in batch mode, natch -- that i can't just go without.

-- just another f*cking observer

[Jafo]

Stacking some vst's etc. Made a nice one with crazy diamonds and cheese machine, thought it would be less cpu intensive just using one sampler. Plus, I can now "grab" some other stuff. I can now create a soundfont from a fav rompler patch or drumatic patch etc

fake:D

hmm, got some numbers for comparison? you're probably right, that using soundfonts of vstis would hit less hard than the vstis themselves. sure, you lose a lot of the dynamic character by doing this, but in many cases that'd be negligible.

-- just another f*cking observer

[fake]

Thanks Jafo.

Nah, no numbers, it was just a thought.

With my host MS 2005 there are no inserts for fx only sends and it has a very simple sampler that uses little cpu. I can now create an instrument with some eq/fx on allready and then use a send on that.

Oh and, I should have read the bit in the manual about the sampler.

It sets out:

4 zone, C1,E2,G#3,C5
8 zone, C1,G#1,E2,C3,G#3,E4,B4,F#5
16zone, C1,D1,F1,A1,C2,E2,G#2,C3,E3,G#3,C4,F4,A4,C#5,F5,A#5

Thanks all for your help.

fake

[lawapa]

Depending on the sounds I'm sampling and their freq content I'll start at C2 OR C3 and transposing up 3 to 5 steps. This usually gives me anywhere from 12 to 16 seperate files for my set For most softsynth sounds a jump of 5 steps works well. Since I almost never loop my sets. I'm usually doing atmospheric sounds that don't loop well and I've got the ram so I don't bother. I will loop for effect though. I then process through SoundForge and use Awave to build. SoundForge lets me see my edits in detail as I'm processing. You can use Awave to process, it's algorithms are quite good.

[P.T.]

Instead of going through the trouble of sampling Vsti's you might give Highlife a try.
It does the sampling for you.