The question is who is quoting whom for what purpose and in what context. For instance, just because an audio engineer writing something on MQA decides to cite an article from the Journal of Neuroscience doesn’t mean this engineer has read and understood the said article. Quotations and reference lists don’t make you an expert, let alone a qualified neuroscientist…
1 Like
Let there be no doubt that I’m actively ignoring this part:
This, however, looks like a claim:
Repeating my original question, what exactly are they taking seriously, and how? What is the neuroscientific basis for MQA? What have they studied, and how do they apply their knowledge?
3 Likes
You should contact them for a definitive answer… but I remember when Apodising was introduced to Meridian and they were saying, as I understood it, that you can measure all you like but in the finish, you have to listen to the final result. How does it sound…
1 Like
You don’t neuroscience for this kind of “insight”, do you?
1 Like
I don’t think they should claim to use neuroscience in their development without saying what that means. Otherwise, it’s no different from all the times “quantum” this or that is being bandied about by manufacturers who know that people don’t know anything about quantum mechanics except it being Very Clever™ and Don’t Ask, You Wouldn’t Understand™.
But then again, they claim that MQA is lossless, so…
1 Like
Well, some clearly do and take it a lot further… This insight seems obvious now but it clearly wasn’t then …
1 Like
This paper from Stuart and Craven discusses the ideas behind MQA. It’s a free download. In the first half, Bob Stuart discusses some of the ideas from neuroscience and makes an attempt to provide a substantial bibliography.
2 Likes
In the general case, yes. But you don’t know Bob Stuart then. He’s known for studying psychoacoustics from university onward, and in the last decade, neuroscience. We have no one in the audio field who would qualify as a neuroscientist (actually not true, I can think of two people with neuroscience degrees), but there are handfuls of people with extensive reading/involvement knowledge, and there are some psychoacousticians whose knowledge of audio can be incomplete. There is also work in Japan relating neuroscience (EEG studies) with audio perception, some of which is cited in the Stuart paper.
Neuroscience research is beginning to be recognized as having important insights to offer in audio, that may influence thinking, for example about listening tests 
2 Likes
Thank you for this, interesting read. Here’s what they take away from neuroscience:
-
If you just encode the stuff humans can hear, you can get away with a smaller file size (page 262).
Comment: Any modern lossy codec does this very well, perhaps better than MQA. -
Hearing is quick and reconstruction via linear phase is maybe not as quick (page 262).
Comment: Many DACs, and Roon, let you choose minmum or linear phase filters at your discretion. MQA forces minimum phase because Bob has decided that your neurons like that best. But neurons don’t speak, they don’t prefer anything and they are useless unless they are in a head connected in some way to an ear or two.
The paper shows that the pre-ringing is going to be very low in volume. Neuroscience can’t decide that it’s audible, but double-blind listening tests might. In fact, this paper can only be thought of as preliminary until they actually implement the processing method and have people listen to it double-blind. Have they done that?
What the paper really drives home is that MQA is really at its best when it’s implemented as a complete system from recording to playback. Otherwise, it’s lossy. Or ‘degraded about as much as the sound is degraded by travelling through air’:
We suggest that for the current music archive, an efficient distribution channel-coding may be based on spline kernels that provide a music-appropriate coding, this being paired with complementary reconstruction at playback. Resampling should preferably be avoided, except within the same sample rate family and performed using the “hierarchical” methods described here. The aim is to ensure that degradations introduced by the analog–digital–analog signal path are comparable to those of sound passing a short distance through air.
1 Like
dear Chris,
The audio world needs real “standards” to make sure all people (not only me) can enjoy music, record their music, and stream music in the best quality possible…
These audio standards have already been made, and WAV is one of them
It’s used for generations in the professional audio business, digital cinema, etc.
WAV It is THE authority master format. 99.9999% of all music recording now in the world are made in this format, just because it works, it’s uncompressed and delivers the best quality.
But there are always “other” people that think that they know “better” and they come up with something new, something were nobody is waiting for…
MQA is one of them, it delivers minimum quality improvements over what we already have WAV, FLAC, DSF.
The only real intention of MQA is to make profit, as a company, nothing more.
Because all people have to BUY new MQA certified hardware to listen to their own music…
Luckily for all of us, MQA idea died earlier then I expected, just because nobody is waiting for a another format…remember VHS, Betamax? DVD+R, DVD-R? Blu-ray, HDVD? SACD,DVD-A?
People are really not waiting for another format, we want the best quality available
and that is: DXD WAV
Please join and accept DXD WAV as the ulimate Studio Reference Master format, and everyone is happy
1 Like
I take it that when you work you get paid wages, make a profit… That’s not a bad thing… Hi res Wav files make quality streaming impossible and the carbon footprint would be excessive if used ubiquitously. Many many recordings are only available as 44/16 due to the technology of the times and so MQA makes the very best of these. Also ultimately, MQA recording sound sublime on my system.
Tape transfers with MQA sound incredible, and modern recording in MQA do not even need such high sampling rates as the technology has moved on so much.
So, all the music there ever was is still out there, early recordings have been transformed and made deliverable in the highest quality and efficiency, also going forward we can do so much more wit less resources without audio compromise.
We are living in transformational times and MQA is a part of the future.
1 Like
No, stop with these ridiculous assertions.
Qobuz streaming is a direct counterexample to the first claim.
And the second claim about carbon footprint is little more than a red herring. But even it, too, overlooks an important counterexample. Every Roon core CPU works a bit harder to authenticate, unfold, and/or upsample MQA than it does to decode bog standard FLAC. How is that for your specious carbon footprint argument?
AJ
3 Likes
Well it matters to the owners of the streaming data bases as it will increase their carbon footprint and subsequent bills.
1 Like
Meaningless I think. Streaming will grow fairly massively in the future and will be the primary global source for A/V. Transmission and storage for audio files that are 4 to 8 times that of CD or MQA is significant in cost and carbon, including if transmitted over long distances. The cost of unfolding data is probably the red herring because you can think through the MIPS requirement to unpack a 192 kHz FLAC file versus a 48 kHz FLAC file. The MQA unfold and upsampling to 192kHz needs fewer operations (MIPS) than the 4x greater FLAC processing at 192 kHz. The MQA filters are efficient because they are short with many fewer taps than PCM filters, so again substantially fewer MIPS.
1 Like
Roon processing speed displayed in the Signal Path suggests otherwise. Provide evidence to the contrary.
MQA principals and astroturfers are grasping at straws now if invoking carbon footprint reduction as some kind of MQA moral good. It is a negligible issue.
AJ
3 Likes
Roon and MQA Ltd are the ones who can provide comparative data. If looking at Roon processing speeds, remember that Roon also does considerable network communication as well as data processing so it’s necessary to discuss this with developers.
The operational costs of streaming companies are not simple to assess, even in some of the web materials trying to analyze costs and payments to labels and artists. You’d need direct access to one of the companies to understand it. My thinking is opposite to yours on carbon, and I don’t see a basis for dismissing it as negligible.
3 Likes
Carbon footprint reduction is an admirable and necessary goal. But streaming audio is such a tiny drop in the bucket – a lightweight energy burden compared to that of streaming video, let alone power generation and transportation, which are the real carbon heavy anthropomorphic activities. No, the opportunity for significant reduction vis a vis “high res” audio streaming is minimal, and MQA is not the solution in that regard.
AJ
Oh, before we even go that far - how does the purported reduction compare to the carbon footprint of replacing all DACs that don’t include MQA decoding ?
(and recycling them, because nothing screams “I care about the environment” like “let me develop something that’ll send tens of millions of devices made of a mix of super-recyclable plastic, gold, and metal to a Ghanian or Chinese dumpster where they’ll totally not pollute anything”)
4 Likes
People replace DACs all the time and there is a natural obsolescence.
So new DACs have always been designed and sold regardless of MQA. I expect happy owners of MQA DACs will keep them for longer in the future anyway. But wether they do or not, there is a thriving second hand market and electrical goods are now widely recycled…
4 Likes
The companies themselves will have the best idea of this. The fact that they’re not exactly beating down MQA’s door suggests it’s not an obvious money saving proposition.
As for saving the planet, music streaming really isn’t going to make a difference as it’s not a significant fraction. There are plenty of studies of internet traffic, and its growth. Finding one that mentions music streaming traffic is a challenge.