More MQA Controversy

[RayDunzl]

My "philosophy" is that it is usually best to listen at a "realistic" volume - which could, presumably be the level you suggest for a string quartet at a distance, but quite a bit louder for a symphony, and even louder for rock music.

Do you have any measurements to quantify "realistic" at your place?

 

[Don Hills]

For me, "realistic" is the level that most of the music I listen to (rock/prog/metal) was mastered at (about 80 to 85 dB SPL). Bob Katz is the main driver of an initiative to bring the same consistency to mastering levels as already exists in film sound mastering.

http://www.digido.com/how-to-make-better-recordings-part-2.html

 

[Werner]

And, the technology, as I understand it, is multi layered, complex, very sophisticated and based on a lot of psychoacoustic research, the latter of which might be hard to track down independently.

It is not complex. MQA is conceptually very simple. Given a week time and MatLab with all DSP and audio toolboxes I could make a prototype encoder and decoder.

The MQA papers refer to lots of more or less recent psychoacoustic papers. When you look closer into these many don't seem remotely relevant. Many others seems to be in there just because they mention 'time'. And then there are the largely discredited Oohashi and Kunchur papers. What also stands out is that in MQA's reasoning the established acuity limits for interaural delay (which are indeed of the order of microseconds) are adopted for auditory perception at large. Which is outright wrong. Says enough.


MQA's true justification is in this:

Whether or not hires is audibly relevant, whether or not standard digital filters are detrimental, this is now moot because we have found a method that 1) allows to reduce filtering to a bare minimum and 2) allows to transmits this over a low-bandwidth channel.

All else is BS: the scientific justifications, the marketing, the badly-hidden attempts at DRM and at controlling the hardware market.

 

[Sal1950]

MQA's true justification is in this:

Whether or not hires is audibly relevant, whether or not standard digital filters are detrimental, this is now moot because we have found a method that 1) allows to reduce filtering to a bare minimum and 2) allows to transmits this over a low-bandwidth channel.

All else is BS: the scientific justifications, the marketing, the badly-hidden attempts at DRM and at controlling the hardware market.

So if I read you right do you feel the ends ( reduced filtering, etc)
justifies the means (DRM like and controlling the hardware market)?
As an open source supporter I'm having numerous issues with the entire MQA approach. Do we give up so much for a very small (if even audible) improvement in SQ.
The same reduction in file sizes are sure to be just around the corner from other sources and the advances in digital bandwidth needs makes it more irrelevant every day.

 

[Cosmik]

Do you have any measurements to quantify "realistic" at your place?

Well I've waved an SPL meter around while listening, but it's never going to be scientific - unless the recording is created extremely simply and they include a calibration tone and recommended SPL

But how it seems to me is that unless you get the volume at approximately the right level, you are not reproducing the same perceived frequency response as was intended, and nor are you 'feeling' the music in the same way.

Peter Walker of Quad said "There is only one correct volume level for any particular piece of music" which I think I agree with.

 

[Werner]

So if I read you right do you feel the ends ( reduced filtering, etc)
justifies the means (DRM like and controlling the hardware market)?

Not at all.

DRM and market control cannot be tolerated in the present (stimulating!) open source climate.

What I meant above with justification is that, even though it is not even certain that specific things matter, there is now a solution that sidesteps these issues entirely. One cannot be against this.

What has to be fought is the (false) notion that rigid end-to-end control is required. This is MQA's Trojan horse. And you know what's inside of that horse.

 

[Sal1950]

hat I meant above with justification is that, even though it is not even certain that specific things matter, there is now a solution that sidesteps these issues entirely. One cannot be against this.

Thanks for your response Werner,
Forgive me as I'm not to bright, but then what is this "solution that sidesteps" if it is not MQA juggernaut itself? That was to part that went over my head.
TIA

 

[Werner]

The issue is: are AA and AI filters detrimental? Some say yes. Some say no. Let's worry about filters. Then let's worry some more.

MQA sidesteps this: let's find the simplest possible filter that satifies specific criteria for the music at hand. Stop worrying.

 

[FrantzM]

Well I've waved an SPL meter around while listening, but it's never going to be scientific - unless the recording is created extremely simply and they include a calibration tone and recommended SPL

But how it seems to me is that unless you get the volume at approximately the right level, you are not reproducing the same perceived frequency response as was intended, and nor are you 'feeling' the music in the same way.

Peter Walker of Quad said "There is only one correct volume level for any particular piece of music" which I think I agree with.

I agree... I understand the need to put things in the most scientific context possible, it remains that if you were to listen to Metallica's " Enter Sandman"at 65 dB Max SPL, you would quickly know you're missing a lot. No way to enjoy Mahler #2 with max SPL of 65 dB at the listening position... THus there is a realistic volume, an approximation . I have read some people saying that their system is so resolving that they rarely go over 75 dB of max SPL ... I can only to that ...

 

[Thomas savage]

I agree... I understand the need to put things in the most scientific context possible, it remains that if you were to listen to Metallica's " Enter Sandman"at 65 dB Max SPL, you would quickly know you're missing a lot. No way to enjoy Mahler #2 with max SPL of 65 dB at the listening position... THus there is a realistic volume, an approximation . I have read some people saying that their system is so resolving that they rarely go over 75 dB of max SPL ... I can only to that ...

I agree..,
From me too... And another ... Finished by a .

 

[TBone]

I agree... I understand the need to put things in the most scientific context possible, it remains that if you were to listen to Metallica's " Enter Sandman"at 65 dB Max SPL, you would quickly know you're missing a lot.

"Say your prayers, little one
Don't forget, my son
To include everyone

Tuck you in, warm within
Keep you free from sin
Till the Sandman he comes

Sleep with one eye open
Gripping your pillow tight

Exit: light
Enter: night
Take my hand
We're off to never never land"

Luv that track, although my only CD pressing sounds too hot at times, Sandman ain't bad ...

... track 2 and 3 are slammed, however, and therefore don't invite > volume within my system ...

Not certain if less compressed / peak limited pressings exist ...

 

[Fitzcaraldo215]

The issue is: are AA and AI filters detrimental? Some say yes. Some say no. Let's worry about filters. Then let's worry some more.

MQA sidesteps this: let's find the simplest possible filter that satifies specific criteria for the music at hand. Stop worrying.

I do not see that they have sidestepped the filtering issues, at all. In the absence of a clear, unambiguous mandate for single, specific, industry-wide approach - there is controversy, as you say - MQA is offering their own approach, a proprietary one that causes alarms to go off and photon torpedo shields to go up in many heads. They have faced the filtering issue head on, extending it to the ADC as well as the DAC, unlike anyone else to date, AFAIK. Are they overreaching? Off on a wild tangent? Blowing nothing but smoke in its audible effectiveness? Perhaps. Time will tell.

The tricky part of your construct is "satisfies specific criteria for the music". There may be considerable disagreement as to what those specific criteria are. That is probably much more true today than it was when digital audio was younger. Much greater technological choice, much broader knowledge and much better measurement techniques are available today. But, there is no "perfect filter" to solve this elusive audio processing problem; they are all imperfect. Specific approaches therefore appear to diverge more than they converge, and absolute perfection will likely never be achieved. At the same time, though, my impression is sound gets slowly but incrementally better in general across the industry as techniques and understanding become more sophisticated, not schematically simpler.

I talked to a physicist once about his work on subatomic particles. His view was that a lot of what he and others did was to find ever more complex and sophisticated approaches to just extend the number of decimal places in the precision to which certain phenomena were known. That was as opposed to making new discoveries of entirely new breakthrough phenomena, though the search goes on for those to a much more limited extent. I guess he was hoping to be lucky enough to stumble across something new and big and win a Nobel Prize, but that was not happening in his own work.

Perhaps, someting similar is the best that filter designers can hope to do: further extend the accuracy and precision to the point of absolute inaudibility, though never to the impossible absolute of measured perfection. Are we there at absolute inaudibility across the recording and playback chain yet? I doubt that any digital audio filter designers think we are. Audiophiles might be satisfied, that is, until they think they hear something that "sounds better". That, or, enough marketers, reviewers and other audiophiles tell them someting "better" exists.

All the various filters out there play music. If they do not do that reasonably well, they are not implemented or they do not survive. It is much trickier to answer which approach is better, let alone "best". I do not think we are at the point yet that all DACs sound alike, myself. And, DACs with selectable filtering options seem to be increasing in number, though that option might be something of a gimmick.

So, I would like to hear for myself someday what Stuart, et al, have done here. I doubt that it is the gee whiz, colossal sonic breakthrough that Harley and others might suggest. But, I am open to the possibility it might be noticeably better than what we have. It is more radical and far reaching in its attempt than most others. I am trying to keep an open mind as I passively follow developments. It is not a life and death matter, but possibly it could offer a worthwhile improvement eventually.

Succeeding commercially in the marketplace, given audiophile and industry resistance to change, is another matter, however. And, no one will be forced to use it.

 

[Werner]

I do not see that they have sidestepped the filtering issues, at all.

By formulating a strategy for arriving at a minimal filter, they did sidestep the issue whether complex filters are harmful or not.

The tricky part of your construct is "satisfies specific criteria for the music". There may be considerable disagreement as to what those specific criteria are.

But they are very clear on this: the temporal extent of the filter has to be minimal, while the resulting aliasing in the audible band has to be below the programme's innate noise distribution.

Any competent mastering engineer now can apply the same strategy.

 

[Cosmik]

But they are very clear on this: the temporal extent of the filter has to be minimal, while the resulting aliasing in the audible band has to be below the programme's innate noise distribution.
Any competent mastering engineer now can apply the same strategy.

So is the idea that in the studio, everything is recorded at high sample rate and high bit depth? Is it then down to the mastering engineer to apply microphone* and content-specific processing to each recorded track in preparation for distribution?

*the famous "de-blurring" filters..?

 

[Werner]

So is the idea that in the studio, everything is recorded at high sample rate and high bit depth?

Yes.

MQA relies on a quad-rate (or better) original recording, to be downsampled with the aforementioned minimal filter to dual-rate, to be folded into single-rate with the origami trick.

If it does not start from quad rate then the actual MQA flow cannot be applied. Other stuff can still be done, though, see below.

Is it then down to the mastering engineer to apply microphone* and content-specific processing to each recorded track in preparation for distribution?
*the famous "de-blurring" filters..?

There are no deblurring filters. Part of MQA's marketing is to make people believe they are onto something very special. They are not.

Oh, whenever the recording ADCs are well known one could identify their audible or inaudible shortcomings and try to compensate for these. But anyone could do so, inside or outside of the MQA system.

Take for instance http://www.2l.no/pages/album/120.html.

This describes the preparation of an original DAT recording (i.e. 44.1k or 48k) for MQA. But that process amounts to a remastering using knowledge of the recorder's ADC: was the frequency response straight? If not, re-equalise. Were there any unreasonable phase distortions? If yes, compensate. Do we dislike the sight of pre-ringing? Then drop an apodiser on the recording. Did the DAT in combinati0n with the particular music allow significant aliasing in the top treble? Cut out the top treble! Or maybe the music has so little treble that there was no aliasing, and also there would be little imaging upon replay. Then you could upsample the track with a narrow filter and deliver that to the customer. And so on. No magic. Just informed remastering.

 

[Cosmik]

Yes.

MQA relies on a quad-rate (or better) original recording, to be downsampled with the aforementioned minimal filter to dual-rate, to be folded into single-rate with the origami trick.

If it does not start from quad rate then the actual MQA flow cannot be applied. Other stuff can still be done, though, see below.



There are no deblurring filters. Part of MQA's marketing is to make people believe they are onto something very special. They are not.

Oh, whenever the recording ADCs are well known one could identify their audible or inaudible shortcomings and try to compensate for these. But anyone could do so, inside or outside of the MQA system.

Take for instance http://www.2l.no/pages/album/120.html.

This describes the preparation of an original DAT recording (i.e. 44.1k or 48k) for MQA. But that process amounts to a remastering using knowledge of the recorder's ADC: was the frequency response straight? If not, re-equalise. Were there any unreasonable phase distortions? If yes, compensate. Do we dislike the sight of pre-ringing? Then drop an apodiser on the recording. Did the DAT in combinati0n with the particular music allow significant aliasing in the top treble? Cut out the top treble! Or maybe the music has so little treble that there was no aliasing, and also there would be little imaging upon replay. Then you could upsample the track with a narrow filter and deliver that to the customer. And so on. No magic. Just informed remastering.

Many thanks, Werner. I was originally sceptical about the idea, thinking that if we just distributed the original high res recording, none of this would be needed (with a further layer of scepticism as to the necessity for high res at all). Then I began to hear references to "de-blurring" etc. and the "end-to-end" idea. It seems as though they have all the angles covered - if, in ten years' time, bandwidth is so cheap that any audio recording can be distributed in its native high resolution without any limits, they still have the end-to-end "authenticated" card to play.

The final audio 'format'? I think they are in a position to do it (even if it is just smoke and mirrors).

 

[amirm]

You know, this makes me think. The goal behind MQA is to nullify the effects of filtering at recording/playback. But such an effect likely is either inaudible or close to inaudible.

What we badly need is nullifying the differences between the speakers+rooms used for recording and playback. A standardized method of capturing the response of the recording room is what we really need. That correction would obviously be hugely audible and very important to finally address.

 

[Thomas savage]

You know, this makes me think. The goal behind MQA is to nullify the effects of filtering at recording/playback. But such an effect likely is either inaudible or close to inaudible.

What we badly need is nullifying the differences between the speakers+rooms used for recording and playback. A standardized method of capturing the response of the recording room is what we really need. That correction would obviously be hugely audible and very important to finally address.

Could that be achieved through software some how? Each studio having a custom 'equaliser ' applied to a universal standard...

They need total uniformity for the recording and mastering process obviously this can't happen from a physical parity between the different studios. But a software applied equaliser might work...

 

[amirm]

It is non-trivial to do perfectly but baby step would be just a frequency response sweep so that we know the timbre at the listening location of the talent. That can be stored as metadata, taking very little space compared to the music file itself.

 

[Thomas savage]

It's very annoying to have the massive disparity in HF energy in recordings, they seem to be affected by how damped the recording/ mastering environments were.

This should not be imo. Of course sometimes it might be artistic licence..

You then have the issue the recording engineers have to 'guess' a approximation of the listeners acoustic environment for playback... If they could operate with a 'known' value surely this would revolutionise recording in the studio and playback in the home...

We would all then have what was intended... After that it's just up to a matter personal taste, but at least it's intentional not random as is the case now.

Much more important than digital format and infinity sampling rates imo