Up till this comment all was looking good
. I have a feeling the step I quoted may been a mistake though. Did you compare what Dirac did on its own to what it sounded like after your extra tweaking?
Why I suspect this wasn't the way to go is the distinction I made somewhere above between getting the smoothest response at a single point, and the volume average that Dirac deliberately targets. Dirac
could have got the response you ended up with but 'chose' not to...
What you enjoy listing to most is the ultimate deciding factor though
.