When I normalize audio tracks mostly this works fine.
But in some cases I have the right gain at the beginning, but the volume goes down after a few seconds.
I tried the “Normalize (2 pass)” effect instead. This seem to work.
Is this the richt way?
And whats the reason for falling volume using the simple normalisation?
Yes, if you want to normalise audio that you aren’t streaming live (which you obviously aren’t if you’re editing it in kdenlive), and you want to normalise based on the entire track, then 2 pass processing is exactly the tool you want.
whats the reason for falling volume using the simple normalisation?
Because it uses an algorithm designed for live streams with no lookahead possible, so it has no idea what ‘normal’ really is for the entire track, it only knows a window of what it has already seen, and how different that is ‘on average’ from the level you requested, and then it tries to gently apply an adjustment based on the idea that what it has most recently seen is probably representative of what about to come. But if it’s not, then it will ‘incorrectly’ vary the level during persistently loud or soft sections, and if what you requested is different from the initial level it will take time to react to the fact that it is not just a transient difference.