Just my 2 pence.
The mscaler is designed on a concept and a belief, that its elaborate interpolated upscaling using a million taps, increases perceived clarity, depth of image and so on.
I do not want to go into whether it is a valid concept or belief, suffice that it was designed with that in mind. Some call it magic, some call it voodoo, I call it barely detectable, but it doesn't matter.
The point is that in a lab, doing an objective test and measurement, it beggars belief that it could beat a straight wire! It is a DSP circuit, and no circuit is lossless, noiseless etc.
So I never expected fantastic lab results in the first place.
Amir is very correct about perception vs hearing. I say that mscaler works on the basis of perception, and that has got a lot of unscientific corners to it.
I accept the measurement results, indeed they are better than I had expected.
But I spent an hour with mscaler/Dave, playing my own material into my own headphones (things I was familiar with), and although at times I perceived a change in sound quality, though very minor, I am sure in a blind test I couldn't say which was which.