How I found out I can't tell mp3 (at 320 kbps) from Flac:
I always thought high bitrate mp3 is good enough for me, and used that as my main codec in my music library, in the times where disk space was more expansive. Around the time I started getting into audiophile headphones, I became curious to see if there is any advantage to listening to lossless. I had the HE4XX back then and those felt like a huge upgrade from the m50x I was using at the time. I really felt like I was not taking full advantage of my headphones when listening to mp3, so I got a HiFi Tidal subscription. I compared the music I had on mp3 to the lossless version in Tidal, and I suddenly felt like I'm missing a whole bunch. It was like the Lossless version was clearer, more precise, with more depth and nuance. I started having real concerns that my mp3 library that I accumulated over many years was no longer relevant, even though it sounded fine to me up until that point.
So I decided to put this to the test. Bling testing in the
ABX test site didn't show that I could really tell the difference, but I was still convinced that I could hear it in music that I liked and was familiar with. So I conjured up a different blind test, aimed at finding out if mp3 or lossless sounds more enjoyable. The test included taking 15 albums that I knew very well, and selecting two songs from each album. The files began as lossless files (ripped from Tidal), and one song of each album (chosen randomly) was converted to mp3, and after that both of them were converted to WAV so I wouldn't know which is which. I ended up with 15 folders, each containing two songs (from the same album, meaning the same mastering), one of them is lossless flac, the other is MP3 at 320kbps, and both presented as identical WAV files. The only way to discern which one originated from flac and which one from mp3, was to look at them in the spectral view in Audacity.
The test itself was listening to two songs at a time, and deciding which one had a better sound quality to my ears. During the listening, in almost all cases I immediately thought I knew which one was the lossless version and which one was the mp3, and with a high degree of confidence. But later when analyzing the results, I found out that I was right in just 8 out of 15 cases – basically no different than randomly guessing. Not only that I couldn't really tell them apart, I also had false confidence about my ability to do so. Later on I repeated the experiment listening to Focal Clear, and again got a 50% correct guesses.
So my conclusion is that any difference I think I hear between high bitrate mp3 and lossless is purely imagined – just as it should be, given the mp3 codec was designed to completely fool human hearing.
Nevertheless, I still have subscription to a lossless streaming service (Deezer now), because although the difference between it and lossy is imagined, imagination is still a powerful thing. In spite me knowing that in blind testing I can't tell the difference, I still sometimes feel like I'm enjoying the lossless music more. And enjoyment is the end goal, really, even if it's pure placebo.