Advice on Sampling Velocity for Piano

[Genetic_Junk]

I've asked learjeff and he did it (his Rhodes) by looking at the peak levels for about everythird note. He suggested looking at the RMS would be a better analysis. Anyone else have suggestions for getting about five velocity layers out of piano for a soft sampler patch.

Your suggestions will be rewarded with a free soundfont or wav files of the result.

[Watto]

We're doing the same thing at the same time

From what I've read so far, you do one velocity layer across the keyboard at a time, and use your ears and (if possible) a meter.

[Genetic_Junk]

I think you are right (I'll be metering). It will be a long process of trial and error, but worth it in the end.

[Watto]

Indeed! We have very different sounding pianos, too, which is great for KVR

[93143]

I put together a multilayer prototype of the Nearly Upright (my wrecked piano set) a while back.

What I did in this case was just record fortissimo as hard as I could possibly press the note, which gave me a benchmark. I then normalized the loudest strike, raised the other ff samples by the same gain (thus accurately reproducing level differences in the real instrument) and matched the RMS values of the other layers to the normalized ff samples.

If you've got a way to automate this, it makes it far easier. It may still require tweaking (and I may just possibly change to a weighted peak/RMS technique) but it worked okay in the prototype; it sounds fine with standard velocity -> volume scaling and all the samples at unity gain.

If you have good finger control, or a mechanical beater, you don't HAVE to go all out with the fortissimo...

[Shane Sanders]

.....

[Watto]

.....

I never quite get that - does it mean you disagree with the previous person, or you think the thread is ridiculous...?

[Genetic_Junk]

well the thread is certainly not ridiculous and 93143 gave us useful info, so I think Shane being the nice person he is, was just bumping the thread.