I took again a look at the Gkar 1997 ROTJ track.
Strange is that at the point where it gets async the voices get much louder on the left and right channel.
Before this, the voices are just noticeable on the center channel.
So that is why it sounds like a can, what i mentioned in my first post of this topic.
The voices in the center channel are not sync with them in the left and right channel.

If i trim or stretch it, it gets async at a different point.
So i’m still not able to fix it.