I recently noticed that my G5 seemed a bit low in velocity, so I thought I should investigate this and maybe correct the midi velocities through software. I could have googled if there's some smart way to do this but I didn't, that would be cheating
Solving the problem should have 4 steps:
1) Find a way to consistently strike the keys with a set of velocities
2) Record this to a midi file
3) Analyze the midi data to find the systematic variance
4) Create an offset value (or, if necessary, a tuning curve) for every key
5) Apply the correction
6) Re-measure and hope that this has had any effect.
So far I got to step three as follows:
1) I built a sophisticated device to strike the keys for me, between 12 and 20 times per key: https://youtu.be/561Sveu9IP8
2) Midi data was recorded with Reaper
3) Midi data was analyzed with R, see https://photos.app.goo.gl/iUhNU16JgiTXE7P39
I think the results are pretty interesting:
a) I got quite consistent results, so I'd say the key striking device works okay.
b) There is less variability for higher velocities but I suspect this might be to a significant extent due to mechanical inaccuracies in the key striking device
c) The between-key systematic variability seems to be larger than I had hoped with the median value showing differences of 10 steps or more between highest and lowest observed (C5 and A5)
d) The difference in response to the difference in physical intensities is more worrying (C5 vs A5) because this requires a tuning curve per key and not just an offset to correct.
Obviously I've only done this for one octave and I have no apparatus for the black keys yet but anyway, I thought there might be other freaks that find this interesting.
I will probably create tuning curves and apply them through Reaper if I can figure out how to do this. I also need to get data for velocities around 20 and just under 127.
Has anybody done this before?