OK - so here's a really 'rough' test (I literally spent about 5 minutes on these samples on my PC this morning, and only just tested them on the kit now!) - I know it could be better, and I think I need to find an more suitable sample to test with.
I used HHX17 (which I'm pretty sure is a standard 2box sample, so OK to share here).
I used Lustark's tool to extract all the WAVs
I then edited all the 'bow' WAVs with a fade in to remove the initial stick impact sound.
I then reimported all the samples back into a new DSND file using the 2Box tool.
https://1drv.ms/f/s!AvEC4KT8mxCMqMk7KtdJ0eQXxDqEUQThis WAV file starts with the original sound, edge, edge roll, bow, bow roll, bell
You can hear on the rolls that the multiple impact samples sound different (in my opinion) to how an acoustic cymbal would sound.....
The second set of sounds are from the edited sample.
The edge hits and rolls are identical (i left these samples 'as normal' to allow for proper 'crashing' on the edge). However the bow sample give a slightly better swell effect - in my opinion. (And the bell is still the same)
What I didn't notice until now is that the bow samples are layered from 'ping' stick hits up to 'crash' type sounds - and that to me doesn't really sound correct on the swell. I may need to find a sample with just 'crash sounds'
I'm also thinking that (from memory - it's been a while since i hit a real cymbal!!) I think the swell rolls are typically played on the edge - so maybe that is the one to do the swell edits on.....
Anyway - as i said - a real 'quick and dirty' test to prove a concept, and I think there is something there that could be improved on.
I've put my edited DSND file up with the WAV file in case anyone else want to try it out.....