[GoldenSound]

So, do you all think that Audirvana will somehow retain MQA even though Tidal appears done with it? I have loved MQA.

Roon and Audirvana will most likely continue to work with MQA local files if you have them.

Tidal still has MQA files at the moment but they're slowly getting rid of them and replacing them with lossless versions.

There is the possibility that Roon and Audirvana may not be able to continue using MQA depending on what happens with the company once it goes into liquidation, as they have a licensing agreement whereby they actually pay MQA ltd per decoded track rather than a fixed fee or fee per unit/software license. These types of ongoing agreements may not be able to be upheld if MQA ltd no longer exists. But we will have to see. I imagine that most likely whoever ends up holding the rights to the MQA product would be happy to just take a fee to let Roon/Audirvana continue using it

 

[bhobba]

So, do you all think that Audirvana will somehow retain MQA even though Tidal appears done with it? I have loved MQA.

I love MQA, too. I have gone into how it works. For over 99% of content, it is just a tricky way of encoding lossless 96K - except above audibility has a shallow filter applied, i.e. at about 24k. Then, it throws away samples above the 96k sample rate. That creates aliasing products below the noise floor - hence inaudible but does not hurt the phase. All you would do is leave that alone - as I said, on nearly all material, it's inaudible. Then it adds and subtracts each alternate 96k sample. The added 48k signal is transmitted, but the subtracted bits are small and easily encoded. It is just a few bits, which becomes the bottom few bits of the 48k signal where it is below the noise floor, so it is again inaudible - technically, it is just part of the dithered noise most recordings have these days. All Tidal would do is transmit the MQA as 96k Flac without the tricky encoding. The very few that, when properly upsampled, you can hear a difference from the 196k or even 384k original will eventually be transmitted at that rate. Over time, the shallow filter will be removed, and we will get the original master - possibly through some new scaleable codec like SL6.

The bottom line is I am not too worried - it will all be sorted out without too much disruption.

 

[GoldenSound]

I have gone into how it works. For over 99% of content, it is just a tricky way of encoding lossless 96K - except above audibility it has a shallow filter applied, i.e. at about 24k. Then, it throws away samples above the 96k sample rate.

MQA is not lossless, it is a lossy encoding process.

Additionally it does not throw away samples above the 96k sample rate, that would be standard downsampling (assuming a filter was applied, if no filter was applied that'd be worse). The 'first fold' does sort of reconstruct content above 22.05khz but does so in a lossy fashion, leaving behind artifacting in the audible band and added noise.

and we will get the original master

Most MQA files are not produced from a different master, they were batch processed from the existing 44.1khz release. (And the ones that are from a hi-res master are not losslessly reconstructed. There usually is a lossless hi-res version available on other services like qobuz in these instances though, or now TIDAL)

 

[ANewtypeCentury]

I realize you're talking about SMSL, but I look at, say, Topping with the competing DX5 or DX5 Lite, and I just can't think of a single reason why I'd spend the extra hundred bones for the MQA version when (for my money) that's not something I want in the first place? But that is the rare example, it's not usually that cut & dried. Just have to ask yourself how much you value that feature, go from there

 

[bhobba]

MQA is not lossless, it is a lossy encoding process.

Of course, it is not losses - that shallow filter that cuts in about 24k means, by its very design, it is not lossless. In fact, I have seen the graphs of the effect of that filter. It was supposed to be so shallow that you get negligible post-ringing. Don't you believe it - it was there all right. The usual method of creating 96K was to put a 96K brick wall filter and then throw away the samples above that. But the filter they used was not brick wall alias components in the audible spectrum were created - they claimed however it was below the noise floor so caused no harm.

While I like MQA, they were rather good at, how to put it, stretching the truth.

Anyway, thanks for your excellent reply that allowed me to correct any misconceptions my post may have caused - it got my instant upvote.

Thanks
Bill

 

[bhobba]

Hi Again All

Ages ago, I found an excellent paper on MQA
https://www.audiomisc.co.uk/MQA/origami/ThereAndBack.html

In there, you will see measurements of the terrible post-ringing MQA introduces. This idea of the excellent time smear MQA has is bunkum - there, I said it - bunkum.

Ayre produced a much better filter IMHO than the MQA shenanigans:
https://www.ayre.com/wp-content/uploads/2020/08/Ayre_MP_White_Paper.pdf

Using the Ayre filter with what analysis of the original shows is the maximum musical content above the noise floor IMHO would produce better results, and with FLAC even smaller files.

Thanks
Bill