Is Roon supporting MQA? What are the pros and cons of MQA?

Well not really because you have to toss away the content put outside the audible band, you are left with 14-15 bits of musical data for a non MQA DAC.

At least that’s what @jussi_laako measured.

1 Like

Could we get a Roon Developer to comment on this since we obviously have several opinions but there is only one true fact–)) Roon currently doesn’t show MQA encoded Tidal content as lossy, so Roon at least doesn’t consider it lossy. I have no desire to waste time going through Tidal content to find stuff that doesn’t show up as MQA if it doesn’t make any difference (ie. it isn’t lossy at its base level). If it is lossy I’ll take the time to eliminate MQA content from my Roon dbms.

From my limited research on the internet it appears MQA is may be a little less than CD quality in its unfolded state, since they apparently use a few bits of the 16 bit part of the content to encode the compressed lossy MQA higher resolution content. Would be nice for someone who actually works with this and knows to answer this clearly.

Roon quality indicator is representative of it’s internal processing, not the source media.

Why don’t you keep the version that sounds the best to you?

And yet MQA clearly state better than a CD file. I think we have to use our ears here.

Lets see if any of the Roon team wades in. In the meantime you may want to look at this discussion for more
about what Roon considers lossy and what the “quality” indicators mean.

Generally speaking anything less than ie. CD quality ie 44/16 is considered lossy, anything above that is considered Hires.

By this definition MQA files are HiRes as even unfolded as they are generally 44/24 bit. However many don’t consider it HiRes unless is 88 or 96/24 or 176 or 192/24 and MQA is not that unti it is decoded. However MQA files are lossy in the sense that they do not preserve every bit of a 96/24 or 196/24 Hires file they discard some of the bits that contain no musical information, that is why some consider it lossy even though it is a higher resolution that CD. Without any decoding you still get CD resolution.

You also need to understand that MQA is more than just compression and that MQA files have been through a process that attempts to reduce problems in the original ADC conversion when the original digital file created which can improve the sound quality even without decoding. If you really don’t want MQA at all you should find non-MQA files as even if you turn decoding off MQA files will have an element of MQA processing in them and are not the same as the original CD or HiRes file even if they are sourced from the same master.

I find as do many others that usually the MQA file is better even without decoding, decoding improves it further. Many others disagree. Try it and make up your own mind.

From the horses mouth…for what its worth…

1 Like

Sorry, that comment is most likely prohibited by NDAs.

They are still trying to avoid showing the real numbers, by selectively picking samples that don’t have much high frequency content. None of those cases demonstrate unfolding with strong ultrasonic content and 24-bit TPDF noise floor, stuff that you can for example modern music produced in digital domain using software synthesizers. Or noise gated close-mic recorded cymbal kits.

Now give me some data for such, then we can see how MQA handles content that can go to 0 dBFS in 22.05 - 44.1 kHz band and how the noise floor in the 0 - 44.1 kHz band then looks like.

I’d also like to see decoded results of “fully unfolded” 0 - 176.4 kHz sweeps (DXD). And 7 kHz square waves bandlimited to 176.4 kHz bandwidth. How much there is content retained above 100 kHz and how the noise floor then looks like…


Sorry if this is a dumb question, but thought I read before for MQA to work properly does that mean I have to turn of Roon Convolution (I use the HAF Filters), and also Sample Rate Conversion?

Thanks - sort of what I was afraid of - when you get to the subjective lets use our ears to evaluate a technology such as a lossy format you lose any scientific basis. The industry spent many years developing MP3 lossy compression to the point where they felt peoples ears “couldn’t” hear the difference. Clearly for a large part of the audience they were correct. The more I learn about MQA it looks like this is more of the same - but applied at what is supposed to be a much higher quality level. Keeping its technology hidden and not letting the public industry fully test it and evaluate it does not play well with me. They are having success by having their technology being supported by manufacturers but it is unclear if this is because it is actually better or simply an advantage re: sales and marketing. Clearly it is not a lossless bit perfect format.

1 Like

It’s not a dumb question at all. And you are correct. For the MQA light to go on in an MQA DAC there can’t be any intervening non-MQA DSP.

Whether MQA would permit first unfolding in software to be then followed by non-MQA DSP has been a big issue for many Roon users.

As yet, no official announcement has been made by Roon beyond the fact that MQA is coming to Roon. The recent MQA announcement doesn’t add anything to that.

It’s been an extraordinarily long journey for all Roon users who are interested in MQA, and, at times, I lost confidence that anything would actually happen. I’m now confident that Roon and MQA are going to happen and that we will be pleased with the implementation.

I’d say that the large part of that audience didn’t care. When I’m at work listening to the radio, I don’t really care either. When at home however… and my friends who don’t care about listening to MP3, they do say they hear the difference and that it sounds almost live when I turn something on for them at my place. But after that, they still don’t care :slightly_smiling_face:

I’m sure there will also be people who don’t want lossy, no matter what it sounds like.

1 Like

Ha, it sounds like we have the same friends.

1 Like


Appreciate the quick reply.

Assume Roon will implement some sort of check or switch that would automatically turn off Convolution if it found an MQA file and stream it pure (or whatever MQA is), but revert to the filters when non-MQA. Realise this assumption is unknown, but would seem to make sense, as then it’s either MQA or Convolution, but not both during listening sessions.


i would think turning convolution off for MQA would be more detrimental to a persons system than the perceived value added of full MQA decoding.

Ultimately a full software decode prior to convoluton would be the cats meow.

Brian has confirmed that Roon will be doing first unfold:

1 Like

If you have read Stereophile part 2 review, you can see MQA doesn’t follow very much to a Hi-Res recording (DXD), in fact it is somewhat 20dB more noise overall from 10kHz to 30kHz, these noise is quite significant because it happens in the audio range.

In a separate discussion on Aurender A10 streamer, due to MQA ‘leaky’ filter, aliasing got reflected back to the audio range causing distortion.

All these abnormalities doesn’t look good at all for so called MQA ‘Hi-Res’. These are inherent limitations of the format which doesn’t solve anything but creates its set of own problems.

1 Like

I dont understand the science but MQA (through Audirvana) sounds incredible in my system. I prefer it actually. I am looking forward to Roon releasing a decoder.

1 Like

Thanks for confirming that, so it looks like Roon will follow Audirvana and Tidal desktop app, doing the first unfold, hopefully there’s an option to up-sample it with some filters to reflect the original recording’s sampling rate especially those beyond 88.2/96k.