[dashezup]

You would notice that the main difference between lossy and lossless formats is that lossless format keeps all of the stuff while lossy formats partially or wholly drop 20kHz+ if you compare with the spectrogram of them.

And there are good reasons to listen to lossy ones instead of lossless ones, you probably can't hear 20kHz+ sound but it will reduce your headroom if it's there, this especially matters when you play it in high volume because the vibration of the 20kHz+ sound cause could cause audio distortion. MQA is the proprietary audio encoding which deals with such problem, although it seems to be a bit debatable.

Lossless formats are especially important for music storage/archive/remixing. But I really can't hear the difference between opus-128k (vbr), mp3-320k (vbr/cbr) and lossless ones.

I just encode lossless music to opus 128k to listen to when it's possible. opus is a very decent audio format, it's wildly used for VoIP. I wouldn't go any higher than 128k for opus because it's recommended in Opus wiki[1], and I've compared the spectrogram between opus 128k (VBR) and mp3 320k (CBR) and there are only very a few of differences.

[1] https://wiki.xiph.org/Opus_Recommended_Settings

[duped]

It's more like 17k, and the energy above 20k is almost so low it probably isn't adding much headroom. Not that you need it anyway.