When playing the Tidal masters through Roon, I get 24 bit on my DAC. I tried using the Tidal app straight into my DAC, but although it sees it (I can set it in settings), it seems to refuse sending anything to it. Only goes through the Mac’s own speaker. Any ideas how to resolve this?
@danny2, not disputing your central point–that hardware decoding is better–but I do think one thing should be clarified.
You may have noticed–I did–that in order to use Tidal’s SW decoding, you need to select your DAC directly: “System Output” (or whatever the Windows or LInux equivalent is) doesn’t work. From the things I’ve read, I’m pretty sure that’s because Tidal/MQA NEEDS TO KNOW what DAC it’s dealing with so that it can compensate. (Yes, I believe it is possible to do it in advance, if you know what’s coming.) The MQA folks have been very clear about this all along: It’s an end-to-end technology. Indeed, the patent application includes very specific specifications about the impulse response of the end-to-end system. (It’s impressive: maximum of 6 sample widths, single lobe with 90% of the energy, one undershoot, no other ringing before or after.) The only way they can guarantee that is if they know what DACs they’re dealing with.
I have no inside information, but from what I’ve read this seems pretty clear: Whether HW or SW decoding is superior–and I think 1. you’re probably right and 2. some MQA-capable DACs may be better than others (although by design MQA-capable DACS should vary less in their sound than other DACs)–it’s not because SW decoding leaves the final conversion out of the equation.
Cheers,
Jim
but that raises a question for me about how the desktop Tidal app is dealing with that now. does that read the actual DAC it is sending to?
As I said I have no inside information. But from what I’ve read about the technology–and from the fact I mentioned, that you’re forced to select the DAC itself from within Tidal instead of leaving it to the operating system–I think that’s likely to be the case. MQA has a profile of your DAC, I’m thinking (and mine), which shapes the data it sends out, compensating for the time-domain effects of your DAC. Based on what I know, this is not a certainty, but it is very likely.
None us really know. You could be right, but personally I doubt anyone has gone to the trouble/expense of profiling large numbers of DACs - especially legacy DACs - for MQA.
My DAC is only a few years old, but isn’t in production any more, I just don’t see that any commercial entity is going to take the trouble to profile it for MQA-especially when you take into account that with each passing month or year there will be less of them in the market. In fact a while back MQA even said that the reason they weren’t doing software decoding was that it compromised the model, b/c they’d need to use a generic profile, and not an exact one.
The whole MQA thing is really pointed at the mass market and streaming, and not at audiophiles. So I’m skeptical they will write a DAC profile for a model that sells at most a few hundred or a few thousand units, as audiophile stand alone DACs do. And only a small number of those DAC owners are going to be Tidal customers.
The situation is different when a manufacturer wants to partner with MQA in certfiying a new DAC; and I’m guessing at some point MQA will also profile widely held models for playback like an iPhone or a Samsung Galaxy. or a mass market AV receiver. The difference is that those are units that sell in many thousands or even millions of units.
All that said, I hope I’m wrong and they do profile each DAC in SW.
I can’t say I can agree with this. I consider it is aimed at both. Firstly Audiophiles as they are the only ones who can initially access the mainstream via Tidal HI Fi at a premium cost in this market. (just look at this forum) The sound quality from my limited experience is remarkable and even the just curious will want a taste of it in time.
I also believe that mass market is in sights via smartphones and portable media players which raises the game across the board.
As for DAC’s, many high end audiophile DACs will be MQA enabled and new models developed.
This will stimulate equipment sales across the board also.
Just thoughts, Chris
I read in an MQA Q&A that a software decoder might start by using a generic D/A convertor “blur” profile. I expect that this is what Tidal are using. Better results could be obtained with DAC specific profiles.
I spent a good time yesterday comparing the sound of Meridian Explorer2 hardware decoding and Tidal App decoding on my Devialet D440 / ProAc D40R system and MQA decoded by the Tidal app sounded slightly better. This is so even though the Tidal software decoder only performs single stage unfolding ( to 88/96 k samples/s) whereas the explorer fully unfolds the music. Both sound better than undecoded 16/24 and something to enjoy in this stormy new year!
See Darko for the latest on MQA without an MQA DAC: MQA & Tidal – where are we now?
Here’s the short version:
At present, software - like Tidal - can only do the first MQA unfold, and there is no DAC side “de-blurring” as there is with an MQA HW encoded DAC. An MQA DAC does the full unfold (up to DXD rates if the file and the DAC support them) and adds in the special encoding for de-blurring written specifically for that DAC.
With software decoding, you DO get whatever deblurring was done on the recording/ADC side.
I’m guessing it will stay that way, in order to encourage users to buy MQA certified equipment.
There may be a generic DAC de-blurring applied after the decode (I honestly don’t know), but that will by no means achieve the same results as a DAC-specific filter.
Just published: MQA Decoding Explained
[QUOTE=Comments From Bob Stuart]
-
The strength of MQA is that one file can be played back in a wide variety of situations by the customer.
-
The three presentations you discussed: No decode; MQA Core; Full decode are all previewed in the studio. Each is optimally set up for that presentation (with appropriate de-ringing).
-
MQA Core (which comes out of the soft decoder or digital outputs) carries the additional information necessary for an MQA Renderer (eg Dragonfly) or a full Decoder (eg MSB, Brinkman, Mytek, Meridian) to ‘finish the job downstream’.
-
Full software decode is not possible because the DAC must be known and characterized. MQA is an analog to analog process.[/QUOTE](my boldface - danny2)
So, yes, for the full MQA experience you need the MQA DAC.
I believe the second and third fold back apparently not exist in Tidal desktop app was it doesn’t know the DAC profile highest sampling rate. So to be at the safe side, it is limited to 96k which by far most DACs in the market is capable of doing so.
Roon has an advantage that it does know the highest sampling frequency rate of the DAC profile through driver software installation. So it is likely Roon will match the fold back to the maximum sampling frequency that the DAC is capable of accepting it. For example if the DAC can do 384kHz, than the third fold back is likely to happen.
For time domain smear correction is a bit tricky, while this happened at ADC side which has already been taken care of, the one at DAC end is challenging. Somehow, Roon needs to know the type USB input receiver and the DAC chipset then apply the correct ‘de-blurring’.
This is really sad that a full blown MQA can only be realized through hardware decoding… May be Roon can shed some light on how close it can attain the software to of the hardware decoding?
Don’t think so.
MQA has good business reasons to keep the “two tier” system of decoding in place:
the basic SW tier brings people into the MQA ecosystem and (hopefully for MQA) gets them to like/prefer MQA’d files.
The higher second full (HW) tier exists only for those manufacturers and HW owners who’ve paid royalties to MQA for the HW encoding.
So if you want the full benefits of MQA you need the HW decoding If they don’t do something along these lines, why would HW producers partner with them, and why would anyone pay to buy an MQA DAC? If only SW is needed for the full experience, they will anger all the MQA DAC owners, and ruin the HW market.
So far every indication by MQA - including direct quotes in several fifferent instances by Bob Stuart - is that full software decoding is not planned. Only in the wishful thinking of audiophiles has it ever been said that full SW decoding is planned.
3 Likes
I agree. There are a lot of business and quality issues at stake for MQA. I know DCS (Hardware/Software) had originally speculated that they would be able to use a SW solution to implement MQA in its Rossini and Vivaldi pieces. Now the word is that they may have to include hardware to implement MQA.
It is the second part which is the problem.
-
Even if the DAC is known, you have to have the MQA profile for the DAC. Where is that going to come from? Will it be free?
-
Is there even going to be a profile made for each DAC. MQA is not going to go back as Danny2 mentioned and redo old DAC profiles unless they were huge. At best, they will do a generic profile for DAC chips previously sold in large volumes, like ESS, or Burr-Brown. MQA applies its own filters at decode time part of the issue is that the DAC manufacturer would have to give MQA access to it’s proprietary filters in order for MQA to tailor the profile. Not a lot of companies will do that.
-
But, lets assume, you decode it in software. When it hits a DAC, say like the PS Audio, whatever filters and/or temporal de-smearing were applied in decode, aka MQA special black box sauce, will not survive the change to DSD and the proprietary filters applied. In this case, MQA is really just functioning as a different master version. This pretty much applies to any post processing, DSP or Convolution or secondary upsampling and filtering (like HQPlayer).
What kind of profile does MQA really want to know? Assuming to my best of knowledge, MQA does two things when it decodes at the DAC end:
- Fold back to the original sampling frequency
- Time domain smear correction (DAC end)
Anymore I need to know?
Obviously Roon knows what is your maximum sample frequency the DAC supports, using this information, it can adjust the appropriate fold back. Say, if MQA is mastered at 384K and the DAC max out 192k, then Roon will then apply the second fold back. If the DAC supports 384k, then it will apply the third fold back etc…
For Time domain smear correction, this will remain a challenge, obviously, different type of DACs have different type of digital filters but all have a default digital filter - Sharp roll off filter, the software can therefore apply a ‘generic’ footprint.
Your third argument is only applied to specialized DACs with or built-in DSP function, most DACs in the markets are simple straight forward design and software decoding can take care of it.
[quote=“MusicEar, post:138, topic:1375”]
Your third argument is only applied to specialized DACs with or built-in DSP function, most DACs in the markets are simple straight forward design and software decoding can take care of it.
[/quote]This is my belief also … based on pre-MQA launch discussions with Bob.
When the AQ Dragonfly MQA update comes out, will that turn the DF Red into a hardware decoder? Up to its 24/96 limit?
Yes … that’s correct. With a MQA DAC you would normal disable an software MQA decoding.