I stand corrected. But I do think he speaks about it as if he and his approach is the norm, and sadly he isn’t.
Yes, it does sound so good. It was a long time coming and worth the wait. 
Surely you can do better than that? I said I couldn’t find many and David Elias is probably the easiest to come across.
Hardly a resounding endorsement of MQA.
I can’t do better than that – and don’t call me Shirley.
AJ
1 Like
Classic!
Have you all considered that MQA in the conversation may reduce the urge to mix to MP3?
I still don’t like end-to-end. But if there is an industry push that actually forces more audiophile-friendly mastering, is that worth anything?
Or will it not do that because it’s still going to be listened to mostly through bluetooth earbuds?
Not really as you have to mix to regular PCM first before the conversion to MQA. No need for that last step.
Process wasn’t my point. What I am saying is that if engineers “mix to mp3” they are optimizing sound for that format, which has severe limitations. Doesn’t sound good on Magnepans.
If MQA is part of the discussion, then the format has higher capabilities and it is marketed as being audiophile grade sound, and might that cause engineers to modify what they mix to, which might be closer to what audiophiles prefer?
They mix and master to pcm as that’s what all professional DAWs work in. Conversion to MP3 post is just that.
OK let me clarify. I meant “mixing FOR mp3.” I didn’t literally mean that masters are mixed directly into mp3 format.
I was intending to refer to this statement:
But I misread it. Sorry for the confusion. This has nothing to do with the process of creating a master, but rather the guiding sound aesthetic.
“Mastering for the typical MP3 listener” would be the audio aesthetic, I think, to which you are alluding.
AJ
1 Like
The ear bud/car listening crowd, you can’t have enough compression scenario, sigh. MQA wants to penetrate that market so I don’t see an end result MQA file indusing a better master in this regard.
I think this is the fundamental misunderstanding that leads people to be making the wrong kind of comparisons with lossless.
I wonder if a better way to think about MQA is in the context of a modern DAC where it is typical to employ some level of oversampling as part of analog reconstruction including inter-sample peak recovery, and allowing the use of less aggressive band limiting filters.
When you look at what is actually happening inside a DAC such as an ESS Sabre 32 series (just as an example because there are plenty of block diagrams of this series on the internet), then while your 44.1K/16 FLAC or CD may indeed by handled in a completely lossless manner by the time it reaches the DAC process (ie its SPDIF or USB input), it is not necessarily handled in a lossless manner in the digital domain from that point on, so end to end, it is not digitally lossless anyway and never was since DACs stopped leaving as much to chance in the variances of analog electronics.
Early DACs relied entirely on analog electronics (and probably the incidental slew rate behavior of the DAC output) to perform the band limiting and yield the actual analog signal. This of course is then subject to the manufacturing variances of the components being used.
Over time as DSP has become available more cheaply and as what actually needs to happen has become better understood the job of analog reconstruction has moved much more into the digital domain and the actual digital to analog output has become higher precision (16->…->32 bit) to better use the digital processing. This process typically includes up-sampling.
So when you look at CD -> final DAC output stage - is digital really digitally ‘lossless’ any more? It has been up-sampled, mathematically processed etc. And does it even matter if the aim is to ensure consistency from studio master to consumers analog output rather than consistency from stream source to input to DAC process?
My take on MQA is that it is not about what happens in the digital realm from streaming service to streamer device (before going to the DAC), but rather what happens between studio master and the resulting analog signal from the consumers DAC, so basically they are trying to standardize the handling of the digital representation of an analog audio signal all the way to the final output stage of a modern DAC - including standardizing some of the DACs internal DSP where the analog reconstruction processing occurs.
This to my mind is why the whole discussion of ‘is it lossless’ (in the digital domain) is moot. The real comparison to be made is ‘can it help to yield a more consistent and accurate analog signal from your DAC that compares more favourably to a mathematically ideal DAC when given the same studio master?’ I think that when they original claimed it to be ‘lossless’, this is what they were focussed on - ie they believe it to be far less lossy from analog source via studio master to analog output then was previous the case with digital distribution formats and players etc.
5 Likes
When I used to do mastering, then mp3 was a pain just because I could occasionally have a great sounding master when played through my studio DAC and monitors or even on many other system with either the lossless audio file or even down-samplings of it (to 48K from 96k for eg), but when converting it to MP3 it could end up being afflicted by horrible peak distortion sometimes and thus we have to tweak the peak levels down a tiny bit more.
This could vary from track to track - sometimes we get away with the usual -0.3dB, sometimes it had to be dropped a lot lower. Of course this was back in the loudness war days, so everyone wants their master to be as loud and impactful as possible (urgh!) so you end up treading a very fine line along with a few prayer that the mp3 devices you tried it with are actually representative enough. This was basically the level of ‘optimisation’ I did for mp3. I could have also chosen instead to compress the bass harder, but personally I thought that tended to compromise the sound too much. Dance music with an overly thin bass just doesn’t work so well.
Not everyone felt that way of course. Also the processing tool we were using back then were not as good as those in use today where you can now get away with a louder sound with fewer nasty artifacts and a more dynamic resulting sound. Also music styles have changed a bit as well with pop and dance music tending to be less busy today.
The other aspect of ‘mastering for mp3’ was more about mastering for a noisy listening environment than mastering for mp3 specifically. A lot of music on radio/streaming is expected to be listen to in relatively noisey environments and so it needed to be louder. This meant reducing the real dynamic range, but at the same time attempting to preserve the illusion of dynamic range, so an average loudness may only be 9db (or even less) below peak possible signal level.
(I used to disparagingly call this ‘doggy style’ mastering to reflect my opinion of what it often did to music - we used to have metering standards K20, K14, K12 where each sets 0dB as a reference point at 20, 14 and 12dB respectively below full scale). Because I would be asked to master to much higher - around -9dB which would be equivalent to a K9 scale if it existed, and so K9 = canine = dog and hence ‘doggy style’ 
Today things have changed a bit, but some of this still holds - streaming services for eg often want to have an average loudness (measured in a new and better way these days) in the -12 to -18 or so dB range. Too much dynamics in a noisy environment leads to people turning it up too much and hence risking their hearing around high transient with headphones on and also many portable players just don’t have the headroom either, so it would probably just distort. Either way - not good.
This is basically what mp3 orientated mastering was about - making these kind of compromises for your expected target audience.
6 Likes
Thank you Adam for all of the details. Do you think mastering for MQA will mean any increased dynamic range?
I disagree. Lossless is not a moot point just because many DACs don’t do a very good job of conversion. Nevertheless the real problem with MQA is that creates quite audible and awful distortion.
No - not meaningfully measurable dynamic range. It is possible the end to end process may improve (or at least change) your perception of transients which may change your perception of dynamic range.
IMHO - formats mostly make a tiny difference relative to production (recording mixing and mastering) choices. A good production shines regardless of format from even 128k mp3 through to 192/24. A poor production still sounds poor on the studio master
MQA in terms of its aims is interesting and as a concept has potential to yield the last fraction of sound improvement with what they appear to be attempting to address. Whether or not they actually achieve their aims is a different question
Would MQA make more sense if the compression aspect of it was entirely separate to the ADC/DAC compensation aspects of it? Some people are happy with one aspect of it and not the other. For eg some people don’t like the DAC filter compared to what they normally have set, so having Roon do the decode and skipping the render entirely suits them. Others argue just that the compression is digitally lossy and therefore has no place next to lossless hi-res in a high bandwidth age.
IYO. I dont hear awful distortion when I listen to MQA,
1 Like
Same here. I just hear the best recordings I have ever heard. Try Ashley McBryde Girl Goin’ nowhere in MQA and tell me what’s wrong. It can make you tingle… something music is supposed to do… or Angel by Lucinda Williams…
If you have an specific examples of such on Tidal I would be interested to hear them.
If you mean raw undecoded MQA - then yes - that sounds awful to me, but I have not come across any problems with decoded MQA and generally it seems good - if the production quality is there in the first place. The biggest problem of course is finding a non-MQA comparison that you know is from the same master.
Still the ONLY thing we should be talking about in this whole ludicrous marketing exercise of smoke and mirrors.