Thanks Ben I’ll make sure I’m sipping my Baker’s 7 Bourbon to prepare for next stage Apex nirvana.
The Stereophile review of the Rossini Apex is up now:
Yes please.
5 Likes
Thanks for posting this @all2ofme.
Four things that came to my mind in reading this review:
(A) great incremental detail and explanation of the Ring DAC and the APEX changes, some of which I had not read elsewhere
(B) very high praise for the EMM Labs DV2 DAC, even versus Vivaldi, which surprised me
(C) I was pleased to see they used a $2,500 Roon Nucleus as the music server paired with $100,000+ of gear and even Nordost Oden cabling. Ostensibly price is no object in this system, and that says a lot. I also use Roon Nucleus as a source.
(D) I was confused, but I guess not surprised, that the reviewer listened to a lot of MQA, more precisely: “24/96 MQA”. Since it’s a known lossy codec, I have to admit I don’t even know what 24/96 MQA means. Are there different resolutions of MQA? Is high res MQA more or less lossy than low res MQA? Is anyone using Rossini or Vivaldi bandwidth or storage constrained? Stereophile appears to be still all-in with Meridian, and it’s a big knock to their objectivity, IMHO.
There are three physical file formats (ie the actual FLAC files) for MQA files right now:
1- MQA 16bit/44.1KHz - there’s no high res or embedded unfolding here, MQA is simply a form of mastering (call it “MQA beautification” - there is some processing done to the sound) and MQA authentication. Arguably these files are lossy and less that 16bit in depth (some bit depth is required to authenticate MQA)
2- MQA 24bit/44.1KHz - these are files where 'higher than 44.1KHz" information is encoded in the 8 least significant bits of the 24bit word - so the first 16bits are “bit perfect” and the higher frequency information is buried into the remaining 8 bits.
3- MQA 24bit/48KHz - very similar to ‘2’ but for the 48KHz family of sampling rates
MQA involves two stages of “decoding”:
A- “Unfolding” - this is where the 8 least significant bits get transformed into higher frequency data. For type ‘2’ this is transformed into 88.2KHz sampling, for type ‘3’ into 96KHz sampling. Type ‘1’ does not include this information and thus is not unfolded. I should really clarify here that files where the original master sampling rate was 44.1/48 are not unfolded to my knowledge - the 24bit word is all real 24bit data as is. This is a fairly common final master resolution as of late.
B- “Rendering” (aka upsampling) - the unfolded data includes a selection of the upsampling filter to be used (I think there are 32 choices) and what the “reference final sampling rate” is. So a file might say the final sampling rate is 96KHz or 384KHz - it’s just a label really, DACs will always pick a filter and upsample to their max rate regardless.
Although type ‘1’ does no unfolding (there’s simply no data to unfold with a file bit depth of 16bit) it might choose an upsampling filter, I don’t know.
Finally, MQA files should really be classified into a few different types based on how they were created, but the most important difference is “white glove treatment” vs the rest. In the “white glove” method, the MQA team went to the original master tapes (and these are analog masters in all of the cases I know of) and re-transferred them with great care to the reading of the analog information, and correcting for any possible issues the ADC they are using might have. This renders files that sound great, but you could possibly argue this has little to do with MQA and more to do with careful remastering.
“White Glove” files are FEW AND FAR BETWEEN. The VAST majority of MQA files are “dummy encoding” - the files get sent to MQA and they encode them without ever listening to them.
My experience listening to MQA: White Glove is almost always amazing - eg Aretha Franklin’s albums in MQA, which I purchased because IMHO they are the best sounding versions. All the others are a mixed bag IMO.
3 Likes
Wow, @miguelito you just explained MQA better than anyone at MQA (or Meridian) ever did! They should hire you for marketing and comms, though, I suspect they cannot afford your fees ; )
“The VAST majority of MQA files are “dummy encoding” - the files get sent to MQA and they encode them without ever listening to them.”
Agreed. And I can’t get past the Golden Sound videos…
1 Like
Yes, in fact they do.
I have a couple of MQA CDs which contain 44.1/16 tracks that read as standard redbook tracks. However, when ripped, MQATagRestored, and streamed to the dCS system, they MQA render to various upsampled rates, and not necessarily the maximum rate supported by the DAC; for example, Rebecca Pidgeon’s “The Raven” MQA CD renders to 176.4/24, whereas Oscar Peterson Trio’s “We Get Request” MQA CD renders to 352.8/24.
The render rate is essentially a label. When you play any file at a given sample rate, the DAC will indicate that sample rate, but internally most (all?) DACs that upsample do so to the max rate of the chip.
Regardless, there’s fundamentally no information to unfold in an MQA CD, the most you can do is choose a rendering method, ie authenticate as MQA and choose the render filter setting. My gripe with all of this is that the data from this MQA CD has less than 16bit/44.1KHz resolution as some of the data bandwidth is used for MQA authentication and rendering choice.
I’m not sure if you mean MQA specifically or you’re talking about DACs in general, and if so, I think you’re talking about oversampling within the DAC chips, not upsampling. There’s a difference.
Yes, I’m well aware MQA CDs contain less than 16bits of actual musical content.
Other than the math used to interpolate/increase bit depth, what is the difference between upsampling and oversampling?
Understood, but that is my point: upsampling is a mathematical method to interpolate effectively. Oversampling is essentially a simpler method of upsampling. And as upsampling goes, the methods to do so are many.
Regardless, my point is very simple: MQA is choosing some upsampling method but it is simply upsampling rather than unfolding. Unfolding actually decodes higher frequency “true” data buried in the 8 LSBs.
You might have misunderstood. I was actually agreeing with your point that, yes, various 16bit MQA render (not decode) differently depending on the original source master (re-read my post).
1 Like
No upgrade anymore ? I am waiting mine…
What do you mean?
It’s a long time We do not see any member having received his upgrade. Normally, mine will come in 2 or 3 week.
By and large, I expect improvements to make bad recordings sound even worse, and good recordings sound even better… In my experience, the best systems make different recordings sound more different.
2 Likes
Yeah, I agree, but I don’t think they’re necessarily mutually exclusive outcomes.
If we assume that an upgrade is positive (an enormous assumption 
), then I reckon:
…a crap recording can be more enjoyable, perhaps because I’m hearing more of what’s there.
…an amazing recording can be more enjoyable, for the same reason.
And simultaneously, the two recordings might sound more different to one another than they did pre-upgrade.
What I’m not after (personally) is a system that only lets me enjoy perfect recordings. Lots of the music I love is terribly recorded.
A DAC is a rather unusual beast. It’s taking a digital bitstream and converting that into an analog form for downstream amplification. Seems to me that until dCS delivers a clearer explanation of all the areas where it made changes, and what these mean to the ultimate sonic characteristics we hear, we can merely guess at what we’ve now got on our hands.
It’s a bit like the parable of the blind men and the elephant (nice description here: https://en.wikipedia.org/wiki/Blind_men_and_an_elephant).
The improvements wrought by the upgrade represent a virtual cornucopia of changes; a sort of smorgasbord of flavors that emerge with every recording we listen to, and in all manner of unexpected ways.
At least in my system I’ve needed to get used to how utterly “balanced” the APEX represents each recording. What had been previously emphasized now blends into a more proportional sonic picture. It takes quite a few listenings of a song to “get it”. I suppose it’s like listening to a sound engineers remix, though I’ve never sat in that chair (a particularly refined 
- shaken, not stirred).
Actually, yes, agree. I have some really old mono recordings. As I’ve improved my system, they sounded more and more like “AM radio”, and yet you can get more into the music. A better system transmits the entirety of it better: the restricted dynamics and frequency extension limitations but also the nuances that make music be enjoyable.
1 Like
I‘m waiting now since August for a schedule and my dealer can still not give me a final date for upgrade!