The other issue that seems to get overlooked is that software decoding is using a generic DAC profile and that’s Ok but the full MQA is an ‘End To End’ process where the DAC and rendering characteristics and properties are known and accounted for.
So however good software decoding is, there is more to come from dedicated hardware. Also, as always the more expensive systems should produce better results than, say an Explorer2. Of course the theory of diminishing returns kicks in too and we all have to decide at what point to ‘Stop chasing the dragon’. For some of us our wallets may very well decide for us lol
Time machine… 10 years time we will be laughing at this… Chris
Yes, indeed. 1.3 feature set is truly amazing. Sure is nice to support a company such as Roon!
Guy, def let us all sit and enjoy the music and a beer. That is so important and as you note above, the stuff that comes out decoded does actually sound pretty terrific. I note on this forum that even some of the uber-naysayers are engaging in listening and are possibly getting past the “expert analysis” conducted in various places.
The research that lead to MQA was done in places like Cambridge Uni and was nothing to do with Hifi and record labels. It is proper science. That it has found an application in the wider world is fantastic and just what good science tends to do.
The comment above about all the stuff above 96k just being upsampled is wrong on quite a few levels, but, Anders pretty much already responded to that.
The kind of rhetorical question I have for you though is …What do you think when you see people asking for Roon to handle 16/44.1 conversions to DSD512. For some folks it is a “must have”.
Just compare for a second.
Blowing up CD data to enormous data levels with absolutely ZERO additional content is a curious thing to do. Not saying people shouldn’t do it but I don’t see anyone poking holes in it like they do with MQA. Needing more powerful cores devices for the extra DSP, better network bandwidth, better rendering devices downstream, etc. I reflect on the outrage that MQA might need something (e.g a new DAC) to make it work.
Compare that to the process of creating an MQA version whereby A2D corrections can be made, better A2D maths conversion is made including all the analogue signal timing that otherwise get smeared away, the packaging and unfolding, the restoration of those higher rate timing queues , etc etc
Check out the Black Keys latest album. It doesn’t unfold to hi-res but the MQA cut is terrific. Much better than the regular CD.
5 Likes
Amen indeed!!
1 Like
I am no software nor hardware engineer but don’t understand what is the difference between computer software and hardware software ie firmware that makes hardware more than just a box of components.
Perhaps I have been seduced by Miska’s any computer’s hardware is much more powerful than any DAC’s to take advantage of software/firmware.
Or is unfolding a MQA file able to deduct the exact definitions of your DAC compared to mine. Logically I cann’t understand why hardware “unfolding” would lead to better SQ but am I too simplistic? Just trying to get my head around MQA.
I think you need to do a proper read up on MQA.
Software decoding as defined by the way Tidal do it and similarly Roon propose to do it will only unfold the first stage of MQA and have no knowledge of the DAC you are using. This means that however good it is, it will be compromised with reference to Full MQA decoded.
In an MQA DAC they do use software but the properties of the DAC are known and the resultant output will be superior. You will also get the full decode of the particular track you are playing.
MQA is an Analog End to End process that delivers (via a digital pipeline) the authenticated studio Analog sound as defined by the Artist, Studio or copyright owners. (As I understand things).
The use of a generic DAC profile in software and the lack of full unfolding cannot deliver the maximum potential of MQA but even unfolded MQA sounds a lot better than CD in my experience on my system.
A rising tide seems to be raising all ships… 
Thank you Nick for the wonderful thoughts of yours, I tend to agree what you said. All of us are passionate about music, that’s why we have endless quest to try different ways to get the best possible sounds we like. Music in my opinion is something that you like but not necessarily others will, it is more of a personal taste.
There are times we need answers to how it works, what are the good and bad. Manufacturers should not be afraid to disclose them. We, the consumers can therefore made with a sound decision, otherwise it creates a lot of confusion. I’ve came across numerous blogs and posts about mixed feelings since its inception two years back.
The good thing for me is I’ve a vast experience in hi-res music playback since the inception of SACD in beginning of 2000. It was so dam good I immediately brought myself the flagship Marantz SA-1 SACD player and it lasted me for almost eight years! I got real taste of what hi-res recordings are all about while many are still playing back CDs. It is only a few years back hi-res downloads in PCM and DSD have really taken off.
In years to come, streaming will play an important part on how music is being playback at the end users. I believe MQA is going to fill that gap.
1 Like
Re: Software decoding
AFAIK, there is no technical obstacle preventing decode, application of arbitrary DSP, and rendering (with a DAC-specific profile filter) occurring in software. At the moment though, software decodes are limited to the first unfold (96k and generic DAC profile).
3 Likes
Seems to me from the reading I have done that is the company line. However in real life the comparison has not been made. A+ 3.0 and ROON’s haven’t been released.
I feel enough research has been done in universities behind the principles. Bob Stuart and his team have developed a product that has been scrutinised enough to convince Warner Music to MQA their comple catalogue including material that has never even been digitised.
Major companies are developing hardware and software to support decodeing of MQA. Also it answers the problem of streaming High Res music to the masses.
Pseudo Science would not get out of the starting block on this scale. MQA is based on scientific evidence and the latest understandings in phsyco acoustics.
For myself. When playing New kid in town from the Eagles Hotel California Album CD mix and then swapping to MQA unfolded and SWMBO retorts, “bloody hell thats got more oomph” knowing nothing about what I was doing goes to show the tangible difference.
Before I hear, What Mix is it etc? Its the MQA mix that is approved as what is seen as the original preferred version. ‘Authenticated’.
Just thoughts. Chris
Consider the case of the Dragonfly Red as a renderer. It is doing the 2nd and 3rd unfold in the ESS chip by what looks like simply setting upsampling parameters. The ESS chip does not have anything MQA related, and the firmware update to the Dragonfly will simply just pass the upsampling data to the ESS.
So two points:
1- I don’t think there’s any secret sauce at all in those timing arguments, it is still upsampling - phase shifts (the same as timing in the frequency domain) are reduced with upsampling, but it is nothing we don’t already do!
2- The idea of the MQA stream setting upsampling parameters is innovative and interesting.
The remainder could be called “informed upsampling” but that’s about as much reverence as it deserves in my opinion.
My understanding from the information I have read is:
1- Software decoding is ALWAYS the same (to 2x rate)
2- Further “rendering” steps choose upsampling parameters/configuration based on the signal - this is where DAC-chip tailoring takes place simply because those parameters depend on the DAC. My interpretation of this information is there is no DSP of the signal other than each DAC’s native upsampling and a selection of parameters therein.
Step ‘2’ sounds interesting, but the meat of it is upsampling which we can already do, and will improve the timing as MQA argues regardless of whether you use the tailored configuration or not.
It is all software! So of course there is no limitation per-se. However, if you believe the details of the 2nd step of upsampling, it seems to be the case that the MQA stream contains information that is able to program specific upsampling parameters on a target DAC that fine tunes the upsampling in that DAC. This is interesting.
Example: Imagine music that benefits from minimum phase upsampling. The MQA stream would set the upsampling in the DAC chip - or could set the DSP in Roon if MQA told us how to interpret the data - to switch between linear and min phase upsampling automatically. This is kind of cool.
If the intepretation of this is correct, I can imagine eventually someone will backward engineer this information from the unfolded MQA stream to automatically set upsampling parameters in software DSP.
I think it is pretty ‘safe’ to do the first fold back through software decoding without having to the know the profile of the DAC. Since fold back occurred at max 96kHz, most DAC in the market are already 96kHz ready. This has been done in Tidal desktop software decoding.
The problem will surface on the second and third fold back. Not all DACs in the market are 192kHz and worse not even at 384kHz ready. If the software decoding have no knowledge of how far these DACs can accept the maximum sampling rate, it simply can’t playback at all. I’m not sure there’s a provision to scale back the fold back automatically or manually in the software settings but let Roon figure out how these can be done.
I believe Roon has added advantage that it does know what is the maximum sampling rate of the DACs and it can adjust automatically to the appropriate fold back to match the highest sample rate the DAC it can takes.
Guys, have more patience on this, see how Roon does the software decoding when it is ready to roll out, you will never know, Roon may able to a full decoding of MQA in the end!
Yes, Auralic has said they are developing their own proprietary up sampling algorithm to do just that. Check out the post on 6 Jan on their Facebook:
If we agree that what’s done above 96khz is upsampling plain and simple - possibly fine tuned by DAC specific parameters, but otherwise run of the mill upsampling - then there is no contradiction at all here. A DAC that only supports a max of 192 will upsample to that, one that supports higher rates will go higher, etc. Nothing new here.
Careful not confuse the word ‘upsampling’ and ‘fold back’, the later one does contains information unless you don’t mind get pick by someone in this blog😋
I will say my impression is there is NO information past the first unfold other than the DAC upsampling parameter tailoring I just referred to.
Also, I’m a tough guy. See my avatar.
1 Like
Isn’t this done because folks think their DACs are inferior to Roon in regards to upsampling?