MQA is Vaporware

[Miska]

Apple was the first successful company to create true end-to-end encryption/DRM on a consumer hand-held device. This was a pre-condition from the labels (who were dealing with the Napster issue around the same time). This is why the original iPod had only the Apple proprietary connector (no USB).

 

And then already long ago removed DRM from the music content in order for the iTunes music store to really take off...

 

I have purchased new, non-DRM protected hires Pink Floyd, David Gilmour, Mark Knopfler, etc music content, no MQA either, don't need it.

 

I don't need "authentication", especially because I can already see from the few MQA hires files I have that some are plain upsamples of 44.1k content. I also especially don't need technologies that are designed to prevent things like digital room correction, digital headphone cross-feed/3D processing, etc.

 

Plus, comparing the original non-MQA versions to the decoded MQA versions, one can conclude that it is lossy and really obviously not the original. Even more, without decoder there's quite severe quality degradation.

[Miska]

I don't like use of the term "lossy" in the discussion because it is not the same as "severe quality degradation" but people think it is.

 

"Quality degradation" is of course a different ballgame, and use of those sorts of words I don't mind at all, because everyone is absolutely clear about what it means.

 

Yes, I agree that it is ambiguous in general terms.

 

In the context where audio codecs are usually categorized into two bins, lossy (MP3, AAC-LC/AAC-HE, Vorbis) and lossless (FLAC, ALAC). Where the definition is strictly whether the decoded content in bit-by-bit copy of the original. Sonically there are still differences too, AAC-LC is very likely to be sonically better than MP3. So in that respect, MQA belongs to the lossy-family. The tricky part here being that now if you buy MQA content you sort of have like "MP3-in-a-FLAC-file" although such comparison is not exactly accurate either.

 

Then another aspect is "quality degradation" which is less black and white and to some extent subjective. With MQA decoder employed, there is "quality degradation" while without MQA decoder there is "severe quality degradation" meaning that it is worse than carefully made RedBook version would be.

 

For example when I look at decoded MQA hires output compared to original hires version I can see some increase in noise/noise modulation and loss of high frequency harmonics. In many cases those high frequency harmonics precisely represent the waveform details of fast transients... Essentially hires becomes less hires. So what does the "master quality" mean? If I make a recording with my brand new awesome RME ADI-2 Pro at 768 kHz rate or DSD256, and make MQA version out of it, it is not what I heard at my studio when I played back the original. Instead it begins to roll off already around 30 kHz.

[Miska]

In my car, I listen to FM radio, Spotify and Tidal. In that order.

 

But the claim about streaming bandwidth requirements is non-problem. At home I get about 50/35 Mbps over 4G (mobile handset), while my subscription has theoretical max of 300 Mbps. And pretty much anywhere I go here I get at least 30/10 Mbps. Plenty enough to stream DXD FLAC and at home enough to stream 5.1 channel DXD FLAC...

 

With your 4G router that has better antennas I can get 100 Mbps download speeds.

[Miska]

That sounds like a comment from 15 years ago, when downloads were DRM-protected and labels tried to add copyrotection to CDs. The music industry was in panic mode.

 

Legal downloads took off after DRM was abandonned, and whether they are MP3, CD quality or studio quality doesn't make a difference in terms of "giving away crown jewels", since the main loss of income comes from people illegally sharing MP3s, which is still the dominant format. It just makes a difference for the tiny, economically irrelevant fraction of audiophiles.

 

And wasn't the MQA version supposed to be precisely the "crown jewels"? So with MQA we are getting less, just something "crown jewel -like" but still not the real thing?

 

MQA doesn't prevent copying or distribution of the copies. It just prevents creation of content and "full resolution playback" on anything else than hardware that has been approved and licensed by the MQA company to perform playback the way they want.

 

So technically it is much similar to SACD, but with a twist that you could freely and easily copy SACD discs. But you could play the DSD layer only on Sony-approved hardware, otherwise you'd get only the CD-layer. Also the content creation side is similarly controlled, you pay MQA every time a song is encoded for you. Or you purchase ~$20k encoding machine.

 

If manufacturing of hardware with MQA decoder ceases to exist, soon you cannot anymore play the content in full and you are left with less than CD resolution. I have bunch of HD-DVD discs I cannot play anymore. Now I'm supposed to buy the same movies and concerts again on Blu-ray. Which certainly will also cease to exist at some point. I'm not going to do that.

[Miska]

MQA can easily be turned into full-blown DRM. Here's how:

1. Make a subtle change to the format, creating MQAv2.

2. Make decoders supporting v2 verify your rights to play the content online.

2a. Apply online verification also to v1 content.

 

Sure, you can keep your old DAC and play your old v1 music, but any new purchases will need a new DAC or a firmware update. If I were Mr. Stuart, I'd have this planned from the start. Nothing they've said so far contradicts the existence of such a plan.

 

Yes, lot of the infrastructure is sort of there already. If they'd add (or already have) a device specific key in the decoder, then the content could be encrypted to be decoded only on that particular piece of hardware.

[Miska]

For one view at the MQA related technical aspects, there's pretty good three page write out by Jim Lesurf (he writes a technical column on Hi-Fi News magazine):

MQA 192k / 96k There and back again.

 

I don't agree with all his interpretation/details of MQA, but anyway... He also talks about determining required bandwidth and word length for the content and use of that to help save encoding space without introducing anything new, proprietary or quality degradation, just regular standard FLAC. Same stuff I was talking about in my blog post here in the past.

[Miska]

In my opinion, main problem of MQA that it is closed standard.

 

I don't see reasons for label owners to wide support the format.

 

They (Meridian) already managed to lobby MLP codec to the DVD-Audio/Blu-ray spec. From that perspective it is not much different than for example DTS which is also used on Blu-ray etc. On Blu-ray MLP is under label of Dolby True HD competing with DTS-HD MA. Both are surrounded with by heavy DRM.

 

So I see it as a real danger, and I wouldn't like the DRM stuff to again spread from the video side back to audio. DRM always not only restricts copying but also restricts many legitimate use cases how the content can be used (and MQA at the moment heavily focuses on the latter restriction). This is especially tricky for computers and free/open-source operating systems like Linux. Since I use Linux as my main OS I'm very concerned about DRM related things. There are also various audio hardware products out there that use Linux, like microRendu etc.

 

Dolby Digital at least was an open standard (AC-3 aka ATSC A/52).

[Miska]

Does this change any views?

 

Now I'm worried if I don't have a way to avoid MQA streams. I don't want it and I don't need it. Since the only DAC with decoder I have (Meridian) performs especially poorly with MQA content and is not great in other ways either, I don't want the quality degradation compared to RedBook when playing it in "compatibility" way.

 

Maybe it's time to cancel my Tidal subscription. I can get quality degraded streaming for cheaper from Spotify.

[Miska]

3,000 albums so far.

 

I didn't check, but are those the same ones highresaudio.com is selling as downloads with 2€ extra price per album compared to the non-MQA hires? It would be of course easy for Tidal to add those albums that already existed in MQA.

 

I've been buying bunch of albums from there in both MQA and original non-MQA hires for technical comparison purposes.

[Miska]

I suspect it's mostly the result of shaped dither.

 

Could be also some inaccuracy or purposeful tailoring of the post-decoding correction Eq discussed here:

MQA 192k / 96k There and back again.

 

(remember that they've crammed the decoding to run on the same XMOS chip as the USB audio implementation, so there's not much resources available for doing things)

[Miska]

+1. The "lossless" stuff is a bogus red herring. As Jud pointed out elsewhere, HQPlayer is no more lossless in converting PCM to DSD than MQA is in converting PCM to PCM. (If you converted HQPlayer's DSD back to the original PCM sampling rate, it would not be bit identical with the original.)

 

Much closer than MQA... And when selecting suitable filter, modulator and rate you could probably even get back a bit-perfect copy of 24-bit PCM original. In fact any difference is more likely to happen when you convert it back to PCM than when it is being converted to DSD.

 

And I'm not supporting saving the result anywhere and even less I would recommend anybody to store and sell such conversions. In fact it would be plain stupid to do the conversion and store the result in a file. Don't do it! I'm fine if MQA would be placed only in DAC where I can choose to use it or some other DAC, like you can choose a different player than HQPlayer. But if the damn thing is force-fed to me and I have no way to opt-out and get a clean copy of the original source, it becomes unacceptable.

 

Even looking from analog domain of DAC output, it is clear that MQA has losses compared to the original.

 

I usually don't use much emphasis. But the difference is that MQA puts it's fingerprint on the source, the content itself and there's no way to take it out, it's mangled forever. Whereas HQPlayer performs it's operations only at playback time, and you can change the behavior at any time, use some different player or not perform any operations on your will.

[Miska]

Yes. Even to mobile devices.

 

4G mobile data where I live is cheaper and faster than fixed copper connection.

 

In big cities copper is a bit cheaper, but then the speeds also allow gigabit internet connections (now 4G is limited to theoretical max of 300 Mbps). But in 2018 it is promised that 5G networks start and then gigabit speeds should be available over mobile connections too.

 

I haven't had any problems streaming Netflix 4K video over 4G connections so far.

 

In the end, optimizing bog standard FLAC results in smaller bandwidth consumption and better quality than MQA.

[Miska]

Uncompressed 96/24 is less than 5 Mbps. That's not even half the bitrate Netflix streams at.

 

4K Netflix is 16+ Mbps H.265 or VP9.

[Miska]

I say this bc as someone pointed out in the past, not all player will be able to play a 18/96 FLAC.

 

You can dither it 18-bit and still keep 24-bit samples that are just zero-padded, that's what I did for the tests because there was intermediate WAV file before FLAC compression. FLAC is clever enough to notice those zero-valued LSBs. Such file will play on everything that can play 24/96 FLAC.

[Miska]

Another bandwidth-saving technique is to filter out high frequencies containing only noise. This has roughly the same effect as reducing the sample rate without causing any compatibility issues.

 

As optimization I run the source file through spectral analysis and picked a sampling rate that would still fit all the detectable meaningful frequency components. And ended up with 120 kHz sampling rate in such case. HQPlayer can play such files just fine, but of course probably not all players though. But another alternative is, as you say, just apply filter cut-off at 60 kHz and leave sampling rate for example at 176.4 or 192k. That would make the file smaller too and still keep it playable on everything that supports the original rate.

 

All these things could be done at mastering stage based on source content analysis and doesn't pose any more effort than dealing with MQA coding.

[Miska]

1 hour ago, PeterSt said:

So we have ringing in the ADC (!) which causes "blur" and which is now fixed in the material. You can't filter this out because you don't know where it is and to what degree it is there (could be normal echo/hall, so to speak).

 

You know where it is and it is not hard to replace, that's what has been done for ages with apodizing upsampling filters... Nothing new here.

 

[Miska]

10 hours ago, Teresa said:

So how does a digital or analog master audio recording become blurred? Perhaps they should figure out how masters become blurred and quit blurring them. Does this blurring only happen with heavily processed major label recordings? Or do even minimust naturally made recordings also suffer blurring?

 

Because of the brickwall anti-alias decimation filters used to produce "lo-res" RedBook... It only happens when the musical content contains frequencies that reach Nyquist frequency of the sampling process. So you don't need to worry about something like piano that doesn't really produce anything with high frequencies (piano's attacks are also soft). But with percussions it happens quite a lot with RedBook. I personally like to use soprano glockenspiel for testing it. Metal claves and hardwood castanets also work well.

 

10 hours ago, Teresa said:

And is this deblurring process available without MQA?

 

Of course it is, it is called "hi-res", even better use DSD and avoid the decimation step altogether.  Problem solved!

 

[Miska]

Actually MQA increases blur, because it's slow roll-off lowpass filter starts cutting so low (30 kHz). When the higher frequency components are attenuated or removed it has effect of slowing the transient...

[Miska]

25 minutes ago, miguelito said:

What is subtractive dither please? Stop wasting my time with new info please.

 

Age old stuff going around with different names, you can safely ignore it in terms of anything "new"

 

[Miska]

15 minutes ago, The Computer Audiophile said:

It sounds like Android for cars. 

 

Toyota will be using "AGL" aka Automotive Grade Linux. It is a Linux Foundation project and not related to Android.

 

P.S. You can see couple of other brands on the web site ticker too.

P.P.S. My Volvo also runs Linux in it's IVI system.

 

[Miska]

Did you also notice Michael Fremer's Analog Corner where he says MQA is almost as good as vinyl and all other digital is horrible and unlistenable.

 

Obviously someone wants to force-feed MQA to audiophiles...

 

[Miska]

58 minutes ago, Bob Stern said:

Regardless of Warner's present intention, Meridian's streaming format is specifically designed to implement DRM copy protection.  The US counterpart of their 2013 patent application is at:

https://www.google.com/patents/US20160005411

 

I personally consider MQA a DRM system very similar to how dual-layer SACDs worked. Without certified and pre-approved decoder you can only listen severely quality degraded version. Just like RedBook layer of SACD. Only with the certified and pre-approved decoder you can fully utilize the content.

 

Also SACD player digital output was limited to allow only the RedBook quality, so the hires version could only be listened through analog outputs, or sent to another decoder over protected, certified and pre-approved digital link.

 

I'm seeing the same approach pattern still. Even though MQA claims that "you can still listen to the content without decoder" is only the quality degraded version. On purpose. That doesn't remove the DRM aspect. MQA encoding doesn't save bandwidth compared to comparable quality standard FLAC either, in fact it increases bandwidth usage because for the FLAC encoder, large portion of the data becomes just plain noise due to the embedded encrypted (DRM protected) stream.

 

And DSD really only took off after it was freed from SACD's DRM. I'm quite sure record companies make more by selling plain DSD downloads instead of SACDs.

 

 

P.S. Meridian has long history on audio content DRM, starting with MLP for DVD-Audio and continuing to Blu-ray audio... There's probably still not much on the market to rip MLP encoded multichannel audio from old DVD-A discs.

[Miska]

6 hours ago, lucretius said:

But you can copy MQA files, assuming you can purchase them. The DSD file on an SACD was never intended to be copied.

Also, the "quality degraded version" you speak of does not really seem that "degraded" as far as I can tell listening to tracks from Tidal.

 

In DRM sense, copying doesn't matter as long as the use of the copy is still restricted. This also applies to movies and audio from iTunes store for example (when they still sold DRM protected music content). Or all kinds of DRM protected video streams over the internet. You can also encrypt a file, for example with GnuPG and share it on the internet for everybody to copy as much as they wish. However, only the ones who have the decryption key can do anything useful with it.

 

I don't evaluate quality degradation by listening, but through objective analysis of the data. And in fact both the undecoded and decoded versions of the MQA are quality degraded compared to the original.

 

This is probably why some record companies like it - the use is restricted, and the quality is like an analog C-cassette copy of the original.

 

[Miska]

8 hours ago, Jud said:

I would not be certain record companies make more by selling downloads, although it is not unlikely.  There has been a long time to attain cost efficiencies with the distribution system for discs, not so long for downloads.  (I recall reading an analysis of the music business’s 2015 or 2016 figures that showed how much more the industry made per unit from discs vs. the equivalent amount - a dozen or so - downloads.  That of course is dealing with companies like Apple, which isn’t a factor in DSD downloads.)

 

CD pressings are much cheaper too than SACD. Because SACDs can be manufactured only at very few locations controlled by Sony.

 

Downloads allow record companies to sell much more directly to the end customers, without long chains of companies each adding 30% profit margin on top of the previous step...

 

[Miska]

8 hours ago, darkmass said:

Given the DVD-A specification of 9.6 megabits per second total data access rate from the disc, some form of compression is needed to be able to provide 96k/24 on all six of the "5.1" channels DVD-A can contain.  The compression scheme certainly didn't have to be MLP, it could have been anything, but I'll let you enumerate how many lossless audio compression schemes were readily available at the time of DVD-A initial release.  (A great compression scheme, FLAC, came along later of course.)

 

 

 

DVD forum could have required ISO, ATSC or MPEG standardization of the codec instead of allowing single vendor proprietary lock-in. AC-3 aka DD aka ATSC A/52 being an example.

 

Ripping the uncompressed PCM streams, or the MPEG encoded streams from DVD-A are much simpler affair.

 

DVD-A entered market in 2000. Shorten was released 1993, WavPack was released 1998, Monkey's Audio lossless was released 2000, FLAC was released 2001.

 

If someone really wants to have two-layer audio with reduced quality and lossless hires combined, for some strange reason, my proposal would be to use MPEG-4 SLS. At least that's a standard. (WavPack has also hybrid two-quality mode)

 

Quote

As far as ripping MLP encoded multichannel audio from old DVD-A discs, I suspect no one will see much development of new DVD-A ripping strategies from what could be considered the "extinct" DVD-A format.  However, the fairly well known DVD Audio Extractor (http://www.dvdae.com/), is on the market and does indeed decode MLP.

 

Even if a format is "extinct" the content is not. So if I ever purchase content I want to have the freedom to losslessly transcode it to a new container without having to purchase the same content over and over again. Standard FLAC, ALAC and such allow this. MQA is breaking this possibility, on purpose. Since I use primarily Linux, I won't be able to use that software.

 

Luckily I never got many DVD-A discs. I've got bunch of movies as HD-DVD discs though and now I don't anymore have capability to watch those and have no means to buy newly manufactured hardware to do that. But I'm not going to buy those movies again in some other format...

 

If the single vendor companies, like MQA, go extinct, possibility to purchase new devices with MQA decoding capability will likely also cease. And it will be very hard to keep the MQA encoded content usable.