Strange behaviour of hairpin
I present a sample MSCX in which a diminuendo hairpin with a velocity change of 10 unexpectedly increases the velocity by +15 at its onset and then starts reducing it until reaching +5 over the velocity before the beginning of the hairpin. The expected behavior would be to start reducing the velocity until it ends -10 relative to the velocity before the hairpin.
This happens in measure 18. The first 7 measures are an attempt to replicate the phenomenon in a different passage in which I included the same overall hairpin pattern. I didn't succeed: the rendering was as expected.
Trying to see what is happening I examined the MSCX but wasn't able to detect anything strange. I exported to MIDI and re-exported to MSCX in order to be able to "measure" the true velocity of each note. I include both files.
Any idea of what's going on?