While I still had an active Tidal subscription a few weeks ago I was able to A/B Tidal's MQA vs Qobuz's 24/96 on a great new recording of Diana Tishchenko "Strangers in Paradise." About 2:54 into the 1st piece the difference can be heard when concentrating on the piano part. The HiRes sounds clean and sharp while the MQA becomes "watery" as if a slight pedal was used. This reminded me about an earlier experience with mp3, which I had used to rip some piano music to conserve disk space a few years back. I had no idea at the time that a lossy format like mp3 would affect piano music negatively. I have since re-ripped all my piano CDs to FLAC. 5 times more space taken, but hey, I happen to really care about sonic nuances. The recent experience reminded me that MQA, like mp3, is lossy. Quoting from TechRadar: "In hi-res competitor format DSD is 306MB while the MQA file is just 40MB. This compares with 73MB for the 24-bit/96kHz FLAC file and 142MB for the 24-bit/192kHz version." Thus MQA file size is just about 54% of HiRes FLAC size; there is no going around information theory: less information is less information. And for some music, definitely not all, it can be heard. Initially, when I started comparing streaming services in October 2019, I thought MQA was some great new music format; "Master Quality Authenticated" is a great name! It invokes associations with master tapes and mastering, and "authenticated" I took to mean something like "guaranteed." That's an effectively misleading naming. Wikipedia has a good article on MQA. I can now understand why some DAC companies like Chord, RME, Schiit, and likely more, will not support MQA. More popular brands like Bluesound and Fiio, do support MQA, as there is popular demand for native decoding of the format. If we spend thousands on dacs/amps/speakers/cables and discuss the tiniest sonic nuances that upgrades of them make, then digital format differences should not be overlooked either.
I found MQA not better than any other format, and found I preferred FLAC with fast rolloff filters to the apodizing filter demanded by MQA. Yes, it did make sounds like cymbals sound soft and squishy. Not even sure if it was the filter or MQA, but I found little benefit.
I would like to stress the fact that I heard the lossy nature of MQA only when critically comparing a modern classical recording involving the piano, an instrument particularly sensitive to reproduction imperfections. When listening to rock or electronic music I didn't hear any problems w/ MQA. It's like a lower resolution photo with some HDR, resulting in the ultimately pleasing experience. Good mp3 doesn't sound all that bad either... Lossy formats are good for the streaming services in lowering their bandwidth costs. What annoys me is how misleading the name "Master Quality" is -- for a lossy format.
I have fast internet. I ran both TIDAL and Qobuz for about 6 months. I just canceled TIDAL. TIDAL was $20/month. Qobuz $15/month. I have tried players from both. I use Roon and Audirvan players now from my PC. The payers all sound a wee bit different playing the same tracks. Some tracks TIDAL shaped the music in a more exciting way. I later found players and settings could do the same. I no just use downloaded FLAC, DSD and mostly Qobuz.
I did a Tidal trial and sampled Classical pieces. I was limited to my Bluesound Node2 for MQA, as my better DACs did not have MQA capability. I couldn’t notice any difference. Perhaps a better DAC would revealed something slight improvement for MQA, but given the ballyhoo that accompanied its rollout, I would have expected to hear some clear cut advance, and it wasn’t there. I canceled my TAS subscription and no longer read anything posted by Steven Stone. This people have lost all credibility for me
For me, the MQA adaptive filtering is what I think most "kills" the music. I notice it most on the decay of "sharp" notes, drums, some piano, etc. Given the typical recording chain today is 40+KHz (not 20KHz), and that most recordings are close microphoned, I think their claims w.r.t. "smearing" is highly questionable. Even with older material, you don't know the recording chain, so any "filtering" is a guess.
Maybe one day there will be sophisticated AI algorithms that can sift through music, pick out specific instruments, and "fix" whatever needs to be fixed (if it does), but MQA, even though adaptive, is brute force, and appears always on.
I do think the initiative to pick the best masters and ensure some sort of pedigree for them is good, but recording engineers who mastered the record have said flat out they were not involved in the approval of the MQA master, so what does that even mean?