Sample amplitude vs sample loudness... and timbre

[pljones]

I've been wondering about this off and on for a while. For years I've used Voxengo Leveler to get the peak amplitude of a set of samples I'm mapping. But, really, is that the right measure? Is that what my ears would decide? My main area of mapping is drums / cymbals. These respond to how hard they're hit and it's really "how hard was the hit" that "MIDI velocity" is meant to track. So you might find that a drum doesn't really get a much higher amplitude after a certain point - but, as the skin deforms more, the timbre changes. Unfortunately, I guess, "the timbre changes" might be different per head per tuning per shell... as a worst case.

So I'm going to ask for the impossible: is there a tool that will let me drop a bunch of samples (say at least 100 per kit piece to allow for layers, round robins and multi-mic'ing) and have it spit out at least three things: peak amplitude, "loudness" and "timbre", for some definition of the latter two terms?

[Rockatansky]

nvm

[enroe]

So I'm going to ask for the impossible: is there a tool that will let me drop a bunch of samples (say at least 100 per kit piece to allow for layers, round robins and multi-mic'ing) and have it spit out at least three things: peak amplitude, "loudness" and "timbre", for some definition of the latter two terms?

Sorry, I don't know a tool that can do that automatically
and in batch. It really seems impossible.

What I do is loading the wav into an audio-editor (audacity,
wavosaur or anything) and go the the "wav-statistics". Here
you can see:

1. peak ---> amplitude
2. rms ---> loudness

So these two properties can be figured out very easy. But

3. frequency-spectrum --> timbre

is very much dependent on your instrument. It can be
the frequency-distribution in the spectrum - but it is
up to you!

In case of hihats or cymbal you may realize that the
sound gets more aggressive if you hit stronger. The skin
deforms, yes - but mostly the timbre changes to a
stronger emphasis of certain frequencies. That means:
The deforming of the cymbal is equivalent to a
deforming of the frequency-spectrum.

I guess that you know already that this kind of deforming
or emphasis can be mimicked by the cutoff-parameter of the
sampler. The trick here lies in choosing the correct cutoff-
frequency.

[jancivil]

I think you're gonna have to use yer ear for criterion #3 there.

It's a pretty good truism that snares and cymbals peak out, you can basically choke all the tone out of a snare if you keep banging it as hard as you can and you can diminish the perception of tone and volume of cymbals the same way. You have to allow them to vibrate. (plus the people with the really big sound are playing to the microphones, not hitting harder)

Especially with snares, well, you've noticed there'll be a wide discrepancy, some are made more to smash and for phenomena such as 'will still ring pretty much'. Then how many layers are actually there, a lot of sampled snares aren't very dynamic anyway, so as you say you're asking the impossible for a batch analysis whatsit.