Originally Posted by Swampfox
I'm sorry, this sounds like BS to me. Music compressed with 'lossy' algorithms destroys data. It can't be recovered.
At the absolute, that is true but perceptually, some games can be played.
As an example, algorithms like AAC+ model the sound at encoding time and and decode time, restore high freqency data completely eliminated at encoding (at bit rates < 64kbps) by "band replication." In other words, they look at the high frequencies that are still there (say at 10 KHz) and using different profiles add in high frequencies in unison with those. The perceputal effect is the music sounds closer to the original because it is no longer muffled due to lack of high frequencies.
At higher bit rates these techniques are not necessary but people have come up with various schemes to improve the perceptual feel nevertheless. Qsound and SRS have had schemes for these for many years as did Yamaha. In my last job, I would hear about schemes like this once or twice a year. None can undo or even predict what has happened in compression domain. So instead, they focus on simply improving the overall fidelity with say, better separation, more subjective bass (turning up lower frequency harmonics), etc.
Similar schemes exist for video for example in how film noise is filtered out and then put back in using profiling.
My personal reaction has been to notice the pleasant improvement initially just to be annoyed over time due to exaggerated compression artifacts. But for average consumer, that may not be as much as a factor, given how many people probably leave things like SRS on TVs on.