There have been many properly conducted (double blind, level matched) ABX[1] tests. It's easy to test yourself: encode a lossless CD rip to various bitrates of

There have been many properly conducted (double blind, level matched) ABX[1] tests. It's easy to test yourself: encode a lossless CD rip to various bitrates of whatever codec you want to try, and use software like the ABX plugin for foobar2000, or similar standalone software to compare the lossless and lossy versions. You may be surprised at how low you can go before noticing much difference. Some sounds compress better than others, and the different codecs (MP3, AAC, Vorbis, Opus etc.) tend to have different problem samples, so take your time and play about. Of course the only point in lossy encoding is to save space (or bandwidth), so it makes sense to do it efficiently hence MP3 CBR320 is rarely the best option, despite it's popularity as a common "go to" solution. [1]: http://www.hydrogenaud.io/forums/index.php?showtopic=16295 http://www.hydrogenaud.io/forums/index.php?showtopic=16295