I, too, am getting the 24/44.1 and 24/48 FLAC issue. One TIDAL MQA album that used to unfold to 24/192 MQA is now playing as 24/48 FLAC (Grateful Dead, Cornell 5/8/77) and sounding worse in the process.
Power-cycling the DS and relaunching the mconnect Control app fixes the issue temporarily, but that is not a long-term elegant solution.
Any ideas as to what causes this issue, and is there a potential fix in the works? Thanks.
EDIT: After I power-cycled the DS, TIDAL MQA played properly, but it was very temporary. I was clicking around the new MQA releases using the Control app and, after clicking on my fourth or fifth MQA album after power-cycling (I believe it was Father John Misty’s new record), the dreaded 24/44.1 FLAC issue re-emerged.
I am still getting it - only about 20-30% actually decode - and it’s always the same ones - switching on and off does no good. It says MQA on both the DAC and the Direct Stream. I confirmed they were encoded correctly by someone with a DS Junior trying them and they work perfectly for him. Don’t know anyone with a bridge 2 and a DS to check that out.
I have tried all sorts of things - switching on and off, three different devices running the Mconnect Control app, nothing changes anything. I am forced to conclude there is something wrong with the bridge 2 at my dealers and/or the software.
I have a number of options - but the one I favor at the moment is sending it back to the distributor to get it working - although I will chat to my dealer about exactly what I do… I personally haven’t paid for it yet and wont until its working, but I know my dealer has and - well its a bit hard on him.
@bhobba: If it says MQA on the DAC then you are getting MQA. When TIDAL Masters files do not decode properly, the DS display and Mcontrol app would say FLAC instead of MQA.
I think people may want to try Roon again - latest build on ROCK (their own OS) because at least for me, it now sounds at least as good, or possibly slightly better than either HQ Player or DLNA - not a significant difference mind you. … the sound quality (in my system, with my configuration) is better with Roon than DLNA or HQ player.
@bhobba: If it says MQA on the DAC then you are getting MQA. When TIDAL Masters file do not decode properly, the DS display and Mcontrol app would say FLAC instead of MQA.
Latest update - the DSJ guy got back and confirmed some of the stuff I mentioned did play at 44.1/24 MQA. The other stuff must have been dumb luck to not be like that.
Mystery now solved well almost - Lemonade decodes to 96k on the Tidal app itself and he mentioned the same with using Audirvana. They must be doing up-sampling of some sort.
Now very very happy - except of course what the bejesus is going on in Tidal and Audirvana.
Thanks for your reply Paul. Good to know that there is an upgrade path for the Bridge III for the jr. Hope it can be a user upgrade kit versus a return to Colorado. I’d hate to be without the DSDJr for 2 weeks but if it is necessary will do. Love the jr so far. Amazing build quality and fantastic sound with Huron!
Do you NEED to use MConnect or will Bubble UPnP work the same way to try mqa tidal payback?
You can use BubbleUpnp to stream tidal with the Bridge II, however only in 16/44 without MQA support. Bubble is my favorite control point. Nice interface with lots of functionality and access to your local stored files (NAS), google Music, google Drive, Dropbox, Box, OneDrive, Amazon Cloud Drive (and Qobuz if you're in Europe ;-). Rock solid.
Mconnect control and mconnect control HD are developed by Conversdigital and (my thoughts…) the Bridge II in essence is the Conversdigital CDMCM-210 mconnect Hardware Module which supports DLNA and MQA:
Bill, we previoulsy discussed this article in considerable detail beginning here with a cite to Darko’s initial summary of the article.
Thanks for that.
It was so fascinating I stayed up until 5am - really mesmerized by it.
It never occurred to me that post ringing in the decoding is inaudible - but it is. It’s the encoding preventing pre-ringing from brick wall filters that seems the key.
I wonder if Ted can chime in and confirm the kind of up-sampling he uses - if its minimum phase or not - I suspect it is - but only he will know. It may be in the link you gave though.
If it is then it explains a lot.
I have an Auralic and it works - but it doesn’t use the MQA filters. It too sounds very good - but some careful listening is needed - when me and the golden ears out my way get some time - which these days seems never.
I purchased one of Keith Eichmann’s power cables to check out months ago - its slated to be checked out with a very experienced friend next week - it took that long to get around to it.
Had a read of the discussion here - and of course the original post on computer audiophile - now for my 2 cents.
I have heard both the MQA from the Bridge 2 and MQA from a power supply upgraded Auralic Aries that does its own MQA decoding - probably what the guy in the article discovered - no need to worry about fancy Bessel function decoding ie reversing the encoding of the MQA to get the best approximation to the signal above 96k. Just to be specific consider a triangle encoding (the simplest Bessel function) where you have 192k and you reduce it to 96k the following way - you go 1/2, 1, 1/2. This has the effect of introducing a slow roll-off from about 20k and chucking every second sample away - but you don’t get that objectionable pre-ringing which we are very sensitive to. That creates aliasing components but that high in frequency its below the normal noise floor in the audible band anyway. The best approximation of the bit you though out is a simple linear interpolation so that’s what MQA does to get the 192k back - that reduces aliasing even further. They use more complex Bessel functions and correspondingly more complex ways to get the best approximation. OK so far - but you have to ask - why bother? That high in frequency is mostly noise so who cares if you aren’t exact. So why not use a quality ordinary up-sampling filter like Sox that does not change the phase and only has post ringing? The reason is it introduces more post ringing than the best guess from the Bessel function that encoded it. But is it audible that high up? Well its well known post ringing is far less objectionable than pre-ringing because after transients things decay anyway - they never do it before which is why your ear hears such quite harshly.
OK so far. Now having heard both the Auralic and PS Audio MQA I gotta tell you it really is hard to tell the difference. Some careful listening will be required here. But my observation is the Bridge MQA sounds slightly leaner while the Auralic is fuller - possibly from the bit of extra post ringing. But even the Auralic sounds a thin compared to ordinary 44.1/16 and some prefer that - personally I like the slightly thinner sound - its more real to me.
I know why PS Audio went for full MQA decoding ie the full unfold. But trust me from my experience with the Auralic the ordinary up-sampling the DS does is good enough - maybe even better. Certainly don’t worry it only goes to 192 and not 384 - it would IMHO make zero difference.
If I was designing a new audio format I would do it differently, I would entropy encode and use simple up-sampling to get the original sampling - but that’s me. I would use the bit stacking - but only down to 20 bits - the last 4 would hold the high frequency info. That combined with subtractive dither would effectively be pretty close to the original 24 bits anyway - and a much better compression scheme - optimfrog.
There are a lot of misunderstandings about sampling theory as well as differences of opinion about what matters most to the ear/brain.
If a signal is bandlimited and sampled at at least twice the highest frequency left then via a “proper” filter it can be reconstructed exactly. (There are details when you consider number of bits in the sample but that’s mostly irrelevant for this discussion.) Note that the sampling theorem doesn’t say “with screwed up phase” or “with preringing”, it says “exactly”. There is no quantization in time caused by sampling and then reconstruction so there are none of the imaginary problems with bad timing or weird effects in the high frequencies that many of the no-reconstruction filter crowd worry about. In fact it’s precisely the reconstruction filter that makes this so.
Anyway, there are a few gotcha’s: either the signal has to sampled for all time or the signal needs to repeat with a definite period to satisfy the sampling theorem. Obviously neither of these happen in practice, but we can get close enough to keep the “edge effects” down low enough to not be a real problem.
Another gotcha: people talk about preringing and show a square wave or impulse response with preringing - note that a square wave or impulse response isn’t bandlimited so one would expect at least aliasing errors when one uses them. If you band limit a square wave or impulse, guess what - it shows preringing. On the other hand if you take a signal that’s band limited to 1/2 the sampling rate and use a “proper” reconstruction filter you get back exactly what you put in - no preringing distortions, etc. (Any preringing you may see is present in the original bandlimited signal.)
Indeed it is true that preringing sounds wrong and postringing is almost entirely covered up/ignored by the brain (after all, things in nature ring after they are struck not before.)
What’s a proper reconstruction filter? An infinitely long filter with linear phase. Non linear phase will mess with the phase, well, non-linearly. Since filters can’t be infinitely long, what are the best approximations to one? Opinions differ, but I agree with the folks at Schiit Audio and Chord that the filter should be as long as possible and shouldn’t mess with the phase - we all use as long of a filter as is practical (well, we all use slightly impractical filters )
FYI, mconnect control app updated today on my iDevices. There is a reference to MQA/Tidal in the update notes. Might solve some of the problems listed here.
FYI, mconnect control app updated today on my iDevices. There is a reference to MQA/Tidal in the update notes. Might solve some of the problems listed here.
Oh man, thank you for the head up. Downloading now and keeping my fingers crossed!
Looks like “Tidal Master (MQA)” reference has to to with Qobuz only. Sounds like HiFi/Master was not an option for Qobuz prior? No references to any bug fixes. so will have to wait and see.
If you band limit a square wave or impulse, guess what - it shows preringing. On the other hand if you take a signal that's band limited to 1/2 the sampling rate and use a "proper" reconstruction filter you get back exactly what you put in - no preringing distortions, etc. (Any preringing you may see is present in the original bandlimited signal.)
I have been mucking about with the entropy encoder you mentioned XIFEO with interesting results. I converted a lot of DSD I had to DXD then ran it through the encode. Its shocking the maximum frequency musical content actually reaches - very rarely above 35khz - even the well recorded 2l recordings rarely do - only a couple go to something like 55kz. The thing to realize is put a brick wall filter above actual musical content and it does nothing - zero - no pre-ringing or anything.
Using that I can get dramatic reductions in the size of DXD files with no audible consequences, and if you use a really efficient compressor like Optimfrog you get files sizes often better than MQA. Up-sample that to DXD and - well I don’t think I could tell the difference.
For example take 2l-109 from the 2l site. Its max frequency is about 35khz so sampling at 88.2 will not affect anything. Do that and encode with Optimfrog and you get a file just a bit bigger than MQA. And its full 24 bits, no tricks or anything - all musical content preserved and indistinguishable from DXD. Most is like that - although something like Magnificat extends to 51k but with 5 bits truncated from the bottom as it’s just noise. In that case you would use the bit stacking trick of MQA on those bits and encode at 88.1 with the 5 bits containing the rest of the frequencies compressed - you would get something like 70-80mb which is less than MQA.
My conclusion nowadays is while I still very much like MQA its really just a mechanism for generating royalties - you can do as good or better in other ways that are open sourced.
Jeff of Arabica said
Looks like "Tidal Master (MQA)" reference has to to with Qobuz only.
That's merely impossible, Qobuz offers streaming 'native' hires and doesn't support MQA at all....depending on the account Qobuz always automatic delivers the highest resolution available. However the Qobuz app did get an update recently for better UI and extended search facilities. Probably this new one is integrated in mConnect through the update. Can't check at the moment as the mConnect update is not yet available over here.
btw…I’m almost convinced the nasty ‘delay…-message’ some people get is caused by network and/or firewall issues. How else come that most of us (incl. me) can play tidal mqa flawlessly (except the gapless issues I hope will be fixed soon…)?
Ambiguous/confusing at best. Regardless, their update information is vague and lacking. The app update doesn’t seem to have changed anything with respect to Tidal and still has some very annoying flaws. It is not a well designed app and still lacks some very basic/standard functionality - relative to Tidal streaming.
My conclusion nowadays is while I still very much like MQA its really just a mechanism for generating royalties - you can do as good or better in other ways that are open sourced.
I don't agree that MQA is "just"a royalty generating mechanism. Certainly, it is a for-profit venture, just like any business. But beyond that, I have no intimate knowledge of the business practices behind MQA, and quite frankly, I don't care. For me, it provides a service, and one that I value. Therefore I have no issue patronizing companies that adopt this technology.
And, while some may have the time/motivation/desire to jump through the hoops you mentioned in your last post to create an equal (or better) end product than an MQA stream, I do not. I have native DSD and other media/formats that I play for the highest digital quality - when quality is the priority. Lets not forget, MQA is still just a streaming protocol in which you cannot remove the value of convenience that many of us covet. So, when I want convenience, and access to massive amounts of music, I look to streaming. And for me, Tidal (and MQA) provides the highest quality level of this convenience - at least at the moment.
"
)