Simulated toms in chiptunes

When toms (parts of the drum kit) are simulated in chip tunes or electronic music (not necessary on the NES platform) this is usually achieved with a pitch slide downwards.

I wonder why this "works" at all. Technically toms, just like any other percussion instruments have a fixed, constant pitch. The pitch of a tom drum cannot vary because both the tension of the skin and the physical dimension of the tom remains constant. Having the pitch going downwards would mean either enlarging the tom or reducing the tension of the skin on the fly - two things which are obviously impossible.

So I wonder - why aren't toms simulated with a constant pitch in chip music ? That'd probably sound much more realistic - although the pitch should probably fall outside of the normal notes so that it is obvious it is part of the rythm and not part of the melody ?

Ideally I'd simulate a tom with a very short white noise, added with a long sine wave of constant frequency and exponentially decreasing intensity. This can't be done on the NES obviously - but having the frequency slide downwards to simulate a tom seems obviously wrong technically - even if for god knowns why reason sometimes it works well in practice.

Like all musical instruments, each note has many frequencies - some dominant, some harmonic, some dissonant. When you hit a steel rimmed drum, the steel itself rings out and then dies off, but the drum head vibrates (at a lower frequency) for longer. I suppose a better pseudo-drum would be a quick high pitch followed by a longer low note with slowly decreasing volume.

Real-world physical toms do NOT have a constant pitch. The tension the drum head is under is a function of the amplitude of the strike (i.e. drumheads are not ideal springs of the sort that comply with Hooke's law), so the period of the fundamental frequency will change.

Here's the spectrogram of https://en.wikipedia.org/wiki/File:Tom_drum_8_inch.ogg (8192 sample window, y=log(f))
Note the several semitones of detuning in the first 1/3 second.


Additionally, as dougeff pointed out, the harmonics of the initial strike are much more broadband than a single (non-white noise) oscillator can emit: a rapid pitchshift imitates that.

I don't know if there is an NSF/GBS etc player that shows you the notes or the spectrum as they are playing, so maybe this thing I'm working on might help.

It's a set of PCE ROMs that play music from popular PCE/Turbo games, and you can disable channels, change the playback speed, etc. High note pitches are to the left, low pitches to the right.


http://www.chrismcovell.com/data/Visual ... t-upd1.zip
http://www.chrismcovell.com/data/Visual ... -upd02.zip

I'm sure there are some Tom hits in there to learn from.