Sampling pianos in fifths...why?


#1

I’m new at this. Is there a specific reason why we’re sampling following the circle of fifths, as opposed to doing every key?

Thanks!


#2

@tikihorea You can do it how you like, sampling in fifths just makes it easier than doing every single note on the piano. :slight_smile:


#3

Oh okay. Thanks bunches!


#4

Christian did explain why, in one of his vlogs


#5

I must’ve missed that one, then.
Thanks for the tip!


#6

I cannot remember which vlog it is, and I cannot look for it, at the moment, sorry!!!


#7

Circle of fifths gives you a fair number of samples up and down the keyboard, and requires relatively little time to do while still sounding pretty realistic.

Sampling consistently is a game of multiplication, every step you do you have to multiply by the rest. So if you sample all 88 keys, then you need to multiply that 88 by every velocity layer you intend to do. And every round robin. And every pedal layer. And so on.

If I wanted 2 velocity layers for my instrument, multiplying that by the like 13 fifths is a lot less time than by all 88 keys. And a lot less worries about making sure each note is consistent and so on.

That said you’ll want to balance the number of notes with round robins and velocity layers, to get a better representation of the piano, otherwise you’ll end up with something that maybe has a good range but terrible dynamic range (cough 2 velocity layers cough).


#8

I have a question about this too.

I know that Christian recommends sampling in intervals of 7 semitones for a piano.
But what about string instruments like guitars. Is an interval of 7 still appropriate for string instruments?

And is a good common number of velocity layers 4 layers, or do professionally sampled instruments have even more than that?


#9

Again, it totally depends on what you’re sampling and what your goal is. There are many different stringed instruments, and many ways to play them.

If you feel like the sound of the instrument isn’t going to change too much across its full dynamic range then you won’t need too many velocity layers. If you think that it’s got quite a complex timbre across its playing frequency range then maybe sample at a smaller interval, like whole steps or every third.

The most useful way to answer these kinds of questions for yourself is to think of it as a problem of possibility and compromises.
What are you able to do in building this instrument? How many microphones, how many notes, how many velocities, how many round robins, how many articulations… etc.

Then think: what compromises must I make? How much time and energy am I willing to spend? How much experience and skill do I have to build this instrument (complexity, scripting, …)? How well do I want it to perform(sample count, audio quality, script efficiency)?

If this still doesn’t answer your question, that might be a good thing. The professionals don’t all approach it one way, which is the reason why there are still so many different sample manufacturers. The best remedy is just to start sampling so you can get a feel for balancing the beauty and deficiencies of the instruments you create.

EDIT: one last thing too is to just go online and look at some “pro” libraries, developers usually list specs such as the things listed above. That might give you some help