Tidal Master vs Apple Lossless

[Davesrose]

I have an Apple Music subscription through my phone provider. I've noticed they have a lot of albums now that are lossless and more being released in Atmos 3D surround. Not sure if all they're doing with lossless is going to a previous master and using a different compression codec, or indeed doing a new mix. I'll listen to music over my main speaker system through Apple TV (or Airplay to my devices from my iPhone). The only main difference I hear with sources with my speakers is what level the levels are set for speaker channels and subwoofer (and most times comparing a digital streaming movie with my UHD disc is the lossless disc has higher levels).

I'm computer agnostic: I have a Mac, Linux machine, and 3 Windows 11 machines. I've never had issues with Apple formats through iTunes on Windows machines. I think all OSes are also similar with your perceived sound quality is more dependent on the audio driver settings. Though I would say the Mac is a little less intuitive if you're needing to get into more advanced device settings (instead of going to sound settings, to go to Audio MIDI Setup app). Switching between the same song with one service and the other might still sound different if you're listening to one on a web browser (and its audio player) vs an app and its settings.

So as far as services, I have Apple because it's cheaper with my cell plan. The main advantage I noticed with it is more albums coming out in Atmos: some sound good in surround, some are too artificially moving instruments all around. All services are being competitive in getting a huge catalog now. Before getting my Apple subscription, they would let you search their catalog, play some songs, and sample the others. Amazon does that too: if you're already a Prime member, they actually have quite a few things open to Prime members (but also pushing full catalog with Unlimited Music). I expect Tidal will let you search their catalog if there might be an issue with an esoteric record label being on one of the catalogs. At least so far, I don't think music is going to be like streaming TV/movies. Where it just was Netflix, Amazon, Hulu carrying many networks and movies...then the studios/networks wanted to get in and move content to their own streaming services.

 

[Steve999]

I had some problems years ago (maybe 5 years ago?) with Apple AAC distorting on my Windows PC when I used Itunes. I’m guessing it was some kind of digital clipping that arose from how I had Windows configured and how that played with ITunes. I never could put my finger on the problem.

Nowadays I don’t have that problem. I don’t have problems with distortion with any kind of streaming. On my Windows PC Apple wants me to use Apple Music instead of Itunes so I do that.

I have a bias between Apple and Tidal. I expect Apple to get things technically correct, and I think Tidal and its MQA games are not worthy of my trust, or my money.

I have multiple streaming services, Apple, Amazon, Spotify, and Youtube Music, because I can and I use them every day and I love music. They all have unique aspects to them that are very worthwhile to me, and all but Spotify come bundled as accessory services from the tech giants, But I don’t have Tidal.

 

[GoldenSound]

Im pretty sure it’s not placebo effects but comparing the master versions from Tidal they sound better than the equivalent lossless version of Apple Music. What do you guys think?

Tidal Masters may sound different because they are applying DSP to them.

MQA in most instances is applied as a post-processing option to the final song and so the Tidal Masters are not actually what the artist/label originally provided. Whereas Apple lossless and other lossless services are.

If one sounds better to you, use that, but if you want what the Artist created/the actual master, use lossless

 

[antanast]

Tidal Masters may sound different because they are applying DSP to them.

MQA in most instances is applied as a post-processing option to the final song and so the Tidal Masters are not actually what the artist/label originally provided. Whereas Apple lossless and other lossless services are.

If one sounds better to you, use that, but if you want what the Artist created/the actual master, use lossless

I was researching this more and then found your videos on the whole MQA story you showed me the way to Qobuz my friend

 

[bigshot]

They don't advertise that they are adding processing, do they? What sort of filter are they applying?

 

[GoldenSound]

They don't advertise that they are adding processing, do they? What sort of filter are they applying?

They are deliberately vague about what they're doing.
Previously they claimed that it was lossless but with a better compression ratio until that was publicly disproven, now they're claiming it's 'better than lossless' and provides 'deblurring' of transients.

They have a whitepaper and patent but their patent relies on a rather hilariously unexplained 'touchup to lossless'.

I made a video on the topic, and later a second response video, both of which go into the effects (both intended and adverse) of MQA and where their marketing does/does not hold up

 

[bigshot]

I get that MQA is a scam. My question is, if you say they’re applying some sort of filter, what kind of filter is it? And how do you know it isn’t just different mastering and not a filter being applied by the DAC?

 

[GoldenSound]

I get that MQA is a scam. My question is, if you say they’re applying some sort of filter, what kind of filter is it? And how do you know it isn’t just different mastering and not a filter being applied by the DAC?

Because you can inspect the files manually. There ARE some where the masters differ, same with regular lossless files, but generally the actual content is the same and can be confirmed via various methods, but it has 22.05khz-44.1khz content aliased back down into the <22.05khz band.

The first 'core decode' with MQA, be it either from the player software you use or the 'full decoder' products (they are the same thing) is doing some proprietary processing and does restore this content back into the 22.05khz-44.1khz band, BUT leaves artefacts and added noise behind so you end up with a damaged/lossy/imperfect result compared to a lossless copy.
The '2nd/3rd unfold' is actually just a fixed reconstruction filter and is not doing anything special. In fact I have a completely decompiled copy of the MQA XMOS code with these filter coefficients.

 

[bigshot]

Is it some sort of manipulation of harmonics, or alteration of response, or phase tricks, or something like that, or is it random noise and distortion artifacts that aren’t intended to improve the sound? Or are we just talking about differences that are masked like any other lossy codec?

Have you set up an A/B comparison to determine exactly what the difference sounds like? Is it clearly audible?

 

[GoldenSound]

Is it some sort of manipulation of harmonics, or alteration of response, or phase tricks, or something like that, or is it random noise and distortion artifacts that aren’t intended to improve the sound?

It attempts to 'fold' (Alias) content above the nyquist frequency down into the audible band, and then undo that at the core decoding step.
This works to a degree but inevitably cannot be done perfectly/losslessly and so ends up altering content that was never above the nyquist frequency itself in the first place, causing signal correlated issues that are a little tricky to describe simply as they are not really similar to any other typical processing or issues

 

[bigshot]

I can’t imagine there would be much up there to drag down below 20kHz. And I see how all that would be audible after doing that would be noise caused by error. Not a good plan.

The level of super audible frequencies is quite low, and I imagine that the noise produced is even lower, so the noise probably isn't audible. I love how audiophiles come up with theoretical improvements that wouldn't work even if they worked. Pure inaudible theory is enough to make them jump through hoops.

 

[GoldenSound]

I can’t imagine there would be much up there to drag down below 20kHz. And I see how all that would be audible after doing that would be noise caused by error. Not a good plan.

Yeah it certainly seems like a solution in search of a problem and not a particularly effective solution at that.
The main thing it was intended to solve was the problem of not enough licensing fees being collected

Unfortunately for them it seems that's not panning out too well either and the company is losing millions per year. They have an investor keeping them afloat but don't currently have a commitment for longer term support and as soon as that tap turns off the company will be insolvent. What happens to the branding/IP after that we'll have to see.

 

[gregorio]

I can’t imagine there would be much up there to drag down below 20kHz. And I see how all that would be audible after doing that would be noise caused by error.

It’s not dragged down, it’s aliases, images of the ultrasonic content mirrored below the Nyquist Freq. So it will only be noise if noise is the only thing above the Nyquist Freq.

The level of super audible frequencies is quite low, and I imagine that the noise produced is even lower, so the noise probably isn't audible.

Yes it is quite low but it’s not only noise and the aliases are only related to the Nyquist Freq, not the music/material. So even quite low level freqs might be audible if they’re uncorrelated/not harmonically related to the music. Although in practice I doubt there are too many recordings where this would be audible at reasonable listening levels.

G

 

[bigshot]

Wouldn't you expect super audible frequencies in recordings of music to be related to the music?

I doubt the difference between MQA and non-MQA would even be audible. It's processing the inaudible and presenting it inaudibly!

 

[gregorio]

Wouldn't you expect super audible frequencies in recordings of music to be related to the music?

We’re not talking about super audible frequencies, we’re talking about musical content that’s mirrored about the sampling frequency into the audible band.

I doubt the difference between MQA and non-MQA would even be audible. It's processing the inaudible and presenting it inaudibly!

No, if it’s causing aliasing it’s processing the inaudible into the audible frequency range, which may or not be audible depending where it is in the audible band, how high it’s amplitude and whether it’s masked.

G