All systems are imperfect. That is they cannot perfectly reproduce the signal fed into them. Therefore the sound that comes out will be judged either 'better' or 'worse' at reproducing that signal. The systems that are 'better' will be improving the sound of the (poor) recording.
Therefore a system could be designed that would process poor recordings to sound like good ones. But the changes made would render the performance different from the original recording.