Roon Decoding MQA Listening Impressions

Use existence as reason why MQA bandwidth claims make sense. Seams about right and it stands up about as well as the other reasons for MQA.

Highresaudio and Qobuz to answer your question.

The reason I’m pushing back is that I dislike the ever-receding argumentation.
“MQA is no smaller in bandwidth than FLAC 18/96, and anyway ordinary high res is fine because bandwidth doesn’t matter because I have plenty of bandwidth to my home here in my first-world country, and anyway if music streaming services care about mobile users which are rapidly becoming the most important market and even if egress cost for the service is a major issue, look it works because Quboz and HRAudio exist, and even if they don’t exist yet in the U.S. they will soon, and even if they are tiny services with a tiny catalog, Apple and Spotify and YouTube and Amazon could afford it, and even if they choose not to, I hate MQA.”

The point about 18/96 is that a technical innovation is not enough, you have to drive success in the market by understanding the needs of all the stakeholders.
The point about bandwidth, saying that I can stream high res video, do you know how Netflix and those guys do that? They don’t stream from one datacenter, they set up content distribution centers all over the planet and pay enormous amounts of money for the egress bandwidth they need. Netflix used to pay many billions to the U.S. Postal Service to distribute disks, and when they switched over to internet that was a very expensive and risky move, and the only people who can play are the giants, and none of the giants consider the value proposition of even CD quality distribution profitable, I see no reason to believe they will jump from MP3 to 24/96.

So I think the chain of arguments have the plausibility of, “I can play video games because I don’t have any homework, and anyway I already did it, and anyway I can do it on the lunch break.”

Note that for me this is just about logical rigor, I have little interest in MQA myself. I have written that after evaluation, I chose a non-MQA DAC. And it’s been a long time since I chose music based on format or quality, I choose the music I like and get the best format available.

1 Like

I don’t see the reason to save bandwidth coz now almost everyone can stream full HD in their mobiles and 4K video via WIFI internet at home. We don’t have bandwidth issue here. Therefore having to stream music contents in the original master with 24/96k FLAC or even DSD are smooth sailing!

1 Like

Ok, that’s a reasonable position. It’s either right or wrong, from a business perspective, but it’s reasonable, defensible. You would have to defend the market segmentation — some people have caps or volume dependent pricing. The business may choose to ignore certain Market segments. Personally, I am more concerned with the egress cost for the service provider.

You may have to explain why none of the biggies go there. Maybe the cost is too high. Maybe they are lazy and stupid. Maybe they have concluded that their customers don’t care, they just want background music, which would mean that Roon and high res and MQA are niches.

But in any case, if you take that position, why talk about 18/96 being more efficient than MQA? If it doesn’t matter?

Redid my math, missed minutes so I originally thought it was realllllly cheap.

The difference between MQA and 24/96 for one stream 24/7 for 30 days costs tidal $2.69. I’m using AWS $.085 per GB for content delivery at their worst 10TB/month rate. (US/Canada).

24/96 is about 3 Mbps in a FLAC container.
MQA is about 2 Mbps in a FLAC container.

1 * 60 * 60 * 24 * 30 / 8 / 1024 *.085 = 2.69 (I don’t think I missed a number in the calc)

If we had an average user’s streaming hrs/day we could get more precise. Lets say an average 8 hours a day, that would be 90 cents. I’d pay 90 cents more per month for 24/96 over redbook. If someone wanted 24/192, $3.60/month extra. The provider could probably split the difference based on ratio of sample rate content in their catalog.

The bandwidth comparison between MQA and 18/96 FLAC dates back to an observation by Jussi in January 2016 as to why he wasn’t convinced by arguments in favour of MQA. The other point he made at that time was that the filters looked leaky to him. Archimago has confirmed the latter, and no one has really doubted the former.

Jussi’s point wasn’t that there was some existing available streaming service. Just that if bandwidth was an issue then there was a possible open source software solution.


Yes, that’s meaningful.

But the pricing for a service provider who buys network capacity is different from a consumer.
The service provider is charged by the network infrastructure provider based on 95% of the peak capacity he contracts for. Peak, not average. (That’s the tricky thing with being a resource provider, like AWS or Azure: you pay wholesale for the capacity you want to have on offer, and you charge retail for incremental resources. Like being a bank.) On the other hand, buying wholesale you get a better price.

The problem is, if you have 100 million customers and you want capacity for all of them to listen to the latest Beyoncé album at the same time, based on your (worst case) numbers, you commit to a few hundred million dollars per month. In practice you would gamble on capacity, like the airlines. We don’t know what the real prices are, but it can become non-trivial, for a big service. Even though MQA is not super-efficient.

Anyway, we either think bandwidth matters, or not. The reasoning follows.

1 Like

The reasoning about MQA encoding looking like random noise and therefore not being compressible is compelling. This is a well-known issue with encryption.

But that raises the question, since FLAC 18/96 exists already and is free, why hasn’t Jussi’s observation been picked up by any service provider? I don’t know, but I can imagine two possible reasons:
a) Streaming services are concerned about the compatibility of 18/96 data with prevailing DACs. Even though the standard may allow 18 (I assume, based on Jussi’s writing), it seems possible that DAC or DAC chip designers have taken shortcuts based on the “knowledge” that only 16, 24 and 32 bit depth are used. Don’t know if this is true, but it’s a difficult thing to test (constant flood of new chips released). So maybe the streaming services would not be willing to gamble.
b) Nobody is driving adoption of this approach. Jussi certainly seems to other things occupying him. An approach doesn’t succeed in the marketplace based on a white paper, it takes an enormous amount of work, technical and business and evangelical, to the various communities (hardware, publishers, streamers, consumers, regulators…). Years, usually.

No, the point is real: you can compress all the actual musical content in a dithered 24/96 file so that unlike MQA, it doesn’t throw away any of the actual bits with musical information during compression. It will be about the same size as the musically lossy MQA file made from the same 24/96 original. So the point is that even for saving bandwidth during streaming, a proprietary format like MQA is totally unnecessary.

That’s beside the point that there are already hi-res streaming services that don’t use MQA. So it really isn’t needed.

Why is MQA making some inroads and not an open source format? That’s easy. The record labels see it as a vehicle for future profits based on a proprietary closed system. Flac streaming doesn’t give them that.


It is great to see an explanation of why bandwidth matters from a different perspective to that of spoiled rich westerners with broadband piped to their front doors. I work adjacent to that industry and know what effort goes into delivering those speeds (which are often driven by marketing and not necessarily capacity) so any limiting of bandwidth use is welcomed, particularly in real time streaming applications. Now a big issue here is the continuous assertion that you could achieve similar efficiencies to MQA using unrestricted open source methods. But nobody has done so commercially, and the reason is that there is no will in the industry to do so.

Maybe because it’s only identified as an issue by someone trying to market their closed system that has licensing end to end.

Link that to what I said earlier and consider this. There was no will in the industry because there was no benefit to them for all of the work. But MQA offers them quite specific benefits both now and possibly in the future in other areas. MQA offers a joined up solution to a lot of people. The only people who don’t benefit directly is us.

1 Like

And to continue the discussion, the lack of demand is obviously the reason why the work isn’t put in to a solution. MQA likes to mention the vast mobile frontier but that’s simply not a market for high res. IMO, its limited to persons that want to feed higher end desktop headphone rigs from their device on hand. Wifi would be available in almost all those scenarios to alleviate bandwidth issues for the end user. Feeling a need for high res audio streaming in a persons daily commute or their vehicle is at best silly.

1 Like

Well the truth is the mobile arena is where MQA has to be successful. If they spend once a year, they would outspend us even if we put our hands in our pockets once a week. And the future isn’t WiFi or Broadband. It is 5G, tailor made for mobile. Mobile is becoming a market for a lot of things and the generations after us will use them to access everything including music. They are wired differently to us, they see value in very different things so while you don’t think there is a market for what MQA offers there, they may well see things differently.

I love the potential of 5G but I only see it getting me to my very limited and overpriced data cap in an hour instead of a day. :pensive: Streaming 256Kb aac isn’t viable with Canadian data caps never mind 8 times larger MQA.

I like the way MQA critics talk about the issues as if they haven’t occurred to the people at MQA and they are blindly carrying on as fools… lol I personaly expect they (MQA and associates) are fully aware of all the issues involved and feel it worth continuing. So, who’s right?
Phrases like Musicaly Lossy also don’t help as with MQA you get the studio sound and if that’s lossy? bring it on… (Hard hat on for another barrage) I’m happy…

No, that is not how an 18 bit 96 kHz FLAC encoding scheme would work. The files would be 24 bit physically. But they would be 18 bit logically – with the six LSBs zero padded. In compression, FLAC would take advantage of that zero padding redundancy. In decompression, FLAC would restore the files to 24 bit – with again the six LSBs zero padded.


Ah, yes, I remember now.
So then, maybe only the lack of advocacy.
Or lack of demand.

Mods have deleted some posts which departed from the guidelines and responses to those posts. Please try to keep the discussion about the ideas and not other posters.


Is it OK with you if I base my opinion on how it sounds to me, on my equipments? Just.checking, as I’m not sure if this qualifies as facts. Thanks.