The audio is changed whene upscaling down to 96 kb/s and ACC and moves forvard compaired to the old video and audio file, so whene muxing the better audio from the old file that was not upscaled, always a delay ms must be insert to sync the audio correct.
Artemis low quality progressiv, same fps
V 2.6.0
Unfortunately yes. There is no option to avoid that yet except extracting and re-fitting the original sound layer manually maybe with a third party tool like MkvToolNix.gui after processing.
Alright, why does them not write this out? This is something users by there self figures out in time, hard if u delete orginal video file and not now this in the first place, i dont notice first time.
Is it so on every model and setup ?
I dont understand why i happens?
I made this experience myself too and lost the original file already not expecting to have an incredible low sound quality fitted into my improved video after processing and it happens to any processing mode unfortunately. 
Its not good. Not all people can find the delay and use Audacity for exampel …
You need to first make sure your clip you want to UPSCALE is a CONSTANT FPS rate. If in doubt, re-render it without changing its size (720X480p etc) but at a real high bit rate (to ensure no loss), I use something real high when working with DVD’s like 9,000kbps and 23.97fps. Again, you need to know what the material FPS is and MAKE SURE its a fixed constant rate to do this. Then once TOPAZ is done with the video rendering at say 23.97fps and bumping it up to 1080p, you put the new enhanced video clip AND the old one into a video editor. Re-render the whole within as MP4 with the audio turned off on the new clip with enhanced video and the video turned off on the original. They will render perfectl as long as you use a fixed FPS rate and not variable rate. I get EXTREMELY good quality results. I prefer to re-render it, even though I might lose a tiny bit of video quality because it ensures better synch with the audio when its played back later.
