modulate sample start point by velocity

[jemkeys25]

is it possible to modulate sample start time by velocity

say i sample a piano note.
I strike the key as hard as I can, let the note die to silence, mark the start of the sample as velocity 127, mark where the sample dies out as 0 velocity, layer in a hammer attack for all velocities 126 and below, and set into a program and voila, a piano sound with no crossfades, depending on the velocity played, the sample is triggered at different points, that were calculated automatically when point 127 and point 0 were specified.

[peter_schwartz]

It might produce an interesting sound on its own, but it won't sound like a piano over the entire dynamic range that you'd expect to be able to play. Even at that, it would only sound like a piano for that string at the loudest dynamic, and maybe stretched a 1/2 or whole step on either side. Then you'd have to sample the next string. And the next... But here's why they won't sound like a piano except for the loudest dynamic, give or take:

The balance and intensity of overtones of that loud-played string won't be correct for softer dynamics on that same string. Even the "clunk" of the hammer strike you want to layer on it would only work for certain loud dynamics levels; outside of that range it would sound highly artificial -- too clunky. Finally, based on the method you described, the sound of soft notes would be much shorter than loud notes.

I don't know if Kronos offers start time modulation based on velocity. Maybe Dan can answer that one. But you can definitely do that kind of thing with Kontakt or even EXS-24 (Logic's sample plugin).

[jemkeys25]

hey pete,

maybe not, but like you said, it could be an interesting sound on its own,

I wouldn't mind being able to try.

also the attack layer could be multi layered wtih say 8 samples velocity switched.

[peter_schwartz]

I'd say it's definitely worth a try to see what kinds of results you get. That's one of the things about sound design... you can try something that sounds good in theory, but even if the results aren't what you wanted, it could be an interesting sound all on its own.

Moving the sample start time is a trick I use a lot to get more mileage out of various sampled instruments, like flute. If I have a flute sample that has just the right vibe (vibrato, tone, etc.) but the natural attack is too slow, I'll push the sample start a bit into the natural attack. This usually produces a "click" from the sample, so I add a little artificial attack to mask it. I use EXS-24 and Kontakt for this kind of work.

As I mentioned before, Dan can confirm whether or not sample start is a modulatable parameter (I don't remember offhand). But I remember from the Oasys that there's a Tone Adjust control which'll let you read into the attack on certain multi-samples.

[danatkorg]

Wherever you can select a Multisample or Drum Sample, you can also select a Start Offset. For ROM and EXs, the Start Offset points are pre-selected; there can be up to 8 of them, although most Multisamples may only have a few, and others have none. (For RAM samples, the only alternate start point is the loop start.) It's not as flexible as an arbitrary start point, to be sure, but on the other hand the pre-selected ones should all be musically useful. You can select different start points by velocity in the same ways that you can select Multisamples by velocity: by assigning them to different velocity zones in the Oscillator, or (more obscure but more extensible) by placing them in a Wave Sequence and using velocity (or another controller) to modulate the position and/or start point.

Hope this helps!

- Dan

[Scott]

I strike the key as hard as I can, let the note die to silence, mark the start of the sample as velocity 127, mark where the sample dies out as 0 velocity

It would be an interesting experiment, if the hardware were capable of it, but besides the other caveats mentioned, you would need some way to determine where to place the 126 "start" points along the decay line. You wouldn't want to simply divide the total time (attack to silence) by 127, leaving the "start points" evenly spaced along the time line. Real pianos have very long decays, and the volume reduction is not linear, it seems to me that you'd end up with a disproportionate number of "quiet" velocities at the low end of the scale. (Also, is there necessarily a linear correlation between how far along the decay line you are and how much less velocity would be needed to "start" the sound at that level?)

layer in a hammer attack for all velocities 126 and below, and set into a program and voila, a piano sound with no crossfades

Apart from the fact that hammer attack itself sounds different at different velocities, I'd also have my doubts about how easily the attack would mesh with different points along the sample line without crossfading.

[jemkeys25]

they have longer decays starting with the longest at the lowest A ,and get shoter as you ascend up the keyboard, so as long as you strike every key as hard as you can, assign the beginning of the sample as 127, where the sample dies out as 0, then the keyboard can automatically distinguish where the rest of the velocity values belong along the sample in equal values, so you don't have to know where to put 126, will can be done automatic.

[Scott]

as long as you strike every key as hard as you can, assign the beginning of the sample as 127, where the sample dies out as 0, then the keyboard can automatically distinguish where the rest of the velocity values belong along the sample in equal values

That's what I'm saying may be a problem... dividing them into equal values.

Let's say the whole sample is 30 seconds long. Velocity 64 will end up at around the level of the decay at the 15 second mark. I wouldn't assume that necessarily corresponds to the level you actually want at that velocity. Hitting a key at half velocity does not necessarily produce the same level as what hitting a key at full velocity produces 15 seconds into its 30 second decay.

[jemkeys25]

your missing the point that higher notes have a shorter duration, note doration decreases as the pitch is raised, the lowest A being the longest, and the highest C the shortest, playing say at velocity 64, will give you half the time of both notes, somewhere in the middle of the sample, once you specify velocity 127, and velocity 0, the keyboard can automatically figure out the rest of the equal points from 1-126.
note duration doesn't really matter since every note on the piano has a different duration, as long as you try to strike the note as hard as you can at sampling, you should get good results.

you also need attack and release samples.

vibrating strings have to follow the laws of physics, striking a note at high velocity will give you a note on the louder side to start out,and will continue to vibrate at its specific frequency until the string loses all of its energy and dies out to silence, passing through, say it started out at 85bd,smoothly from 85db down to 0db.

[michelkeijzers]

If I remember right samples are loaded are streamed but the first part of each sample to be used is already available in some kind of RAM (virtual RAM). However, if the start location of a sample is variable there would be much more sample time 'preload' needed because it is not known beforehand which part of the sample to start with. This can result in a 'load' time between pressing a key and actually her it and this kind of lag is mostly not considered musically (more like unplayable).

So I think that this hardware limitation also can be the cause this feature is not feasible.

[Scott]

your missing the point that higher notes have a shorter duration, note doration decreases as the pitch is raised.

But that has nothing to do with what I was trying to say.

I mean, if a low note last 30 seconds (halfway point at 15 second) or if a high note lasts 10 seconds (halfway point at 5 second), you're still assuming that the level at the halfway point matches the level you want at half initial velocity, and I don't think that's necessarily true.

vibrating strings have to follow the laws of physics, striking a note at high velocity will give you a note on the louder side to start out,and will continue to vibrate at its specific frequency until the string loses all of its energy and dies out to silence, passing through, say it started out at 85bd,smoothly from 85db down to 0db.

But I believe that the decay of the sound is not linear, that's the issue.

(Though as others have mentioned, there are other technological impediments regardless.

[jemkeys25]

I would just like to have the option to try, it may be intersesting.

[jemkeys25]