Can this be an issue with or DSLv5a?
Is it sometimes preferable then to slightly underfit at soft stimulus? I’ve noticed that for my fit providers can struggle to fit via the three levels using . It seems that you may be saying it would be better to compromise by underprescribing the softer sounds rather than the louder ones with the accuracy focus being around medium sounds?
You’re ignoring this member’s posts.
