MQA is Vaporware

[Miska]

5 minutes ago, Jud said:

 

That’s the way one would think it ought to work, but in actuality the music companies have much more economic control over contractual arrangements and profit margins with the disc distribution system than they do with a download market where the price and profit margins for the vast majority of downloads are set by contract with Apple.

 

Hmmh, but Apple is not selling DSD downloads? I doubt selling SACD discs through Amazon is much more profitable than selling downloads through nativedsd.com or HDtracks...

 

At least if I would sell my software on CDs in a cardboard box that old way, it would at least double the price or more... Almost nobody sells boxed software anymore. It is hard for me to imagine selling music would be much different from selling software.

 

[Miska]

30 minutes ago, Jud said:

 

Yep, agreed.  But something the size of the DSD download market is not why some labels and streaming services are trying out MQA.  I would guess they must want MQA as a better quality alternative for streaming and download markets that are now built around mp3/AAC.

 

I fail to understand why they would go with MQA instead of standard FLAC. It doesn't save any bandwidth - in fact it consumes more. Unless they wish it as a DRM vehicle...

 

But MP3/AAC don't contain DRM aspects by the spec, anybody can write a decoder. And now MP3 is patent license free too, so quite a bit cheaper than it used to be.

 

[Miska]

9 hours ago, Shadders said:

The extra layer of MQA has to be paid for by someone. So, maybe as you have said the price for MQA will be higher.

 

Yeah, degraded DRM MQA versions of hires album download cost 2€ more than the original non-MQA hires version...

 

[Miska]

7 hours ago, FredericV said:

MQA does not contain real music content (e.g. harmonics) above 48 Khz, everything above 48 Khz nyquist is fake and contains aliasing errors.

 

It is even quite rolled off compared to brickwall 96k sampling rate PCM. Because their filters begin to roll off really early. At 48 kHz you find mostly artifacts if the source was higher than 96k rate.

[Miska]

6 hours ago, PeterV said:

It is very nice to notice that most classic albums from the label Deutsche Grammophon are being released on Tidal in both 16/44 and in MQA 24/96. Perfect comparison material. It will be interesting to compare Qobuz 24/96 flac with Tidal MQA.

 

Why do you compare RedBook to hires and not original hires to MQA hires?

 

In addition, classical music is poor for these comparisons because there's only small benefit of hires there and very few if any transients that could have any blur...

 

So far I've purchased bunch of albums in both original hires and MQA hires, but don't really see much point in the MQA version, since it is 2€ more expensive than the original and so much more restricted use... In addition it reduces quality...

[Miska]

1 hour ago, PeterV said:

As you will be aware, there are various MQA encoded albums which are unfolded into 24/44, 48, 88.2, 96, 176, 192 and even 384 bit/kHz.

 

Well, MQA contains only max 96 kHz sampling rate worth of content, but not even that properly that as the response is notably rolling off already at ~25 - 30 kHz. Further "unfolding" is just upsampling with crappy leaky filter. And they cut a lot of bits out to do that for content that actually contain some high frequencies in order to get space for the stupid "folding".

 

1 hour ago, PeterV said:

With classical music it is important to know the album you are comparing with also very well is my experience.

 

It doesn't matter, because audio analyzer easily tells what there is and there isn't.

 

1 hour ago, PeterV said:

So I would advise you to try 1 Month Tidal MQA for free. There are thousands of MQA albums available and it is a superb bonus

 

I'm a long time Tidal subscriber. Luckily there are only very few MQA albums for the music I listen. I rather listen RedBook properly processed than MQA-crappified version.

[Miska]

11 hours ago, mansr said:

Here's a better example, 2L-053:

 

 

Here it is also visible that MQA render begins to roll off above 20 kHz compared to the original, before the distortions begin to rise higher up.

 

[Miska]

45 minutes ago, PeterV said:

Technical analysis without complete information how the encoding is being done, why the focus of MQA is totally focussed towards the improvement of temporal blur and to deliver a very fine time description is incomplete and inadequate as well.

 

Measurements of DAC output reveal that MQA actually blurs by rolling off high frequencies and adding high frequency distortions...

 

[Miska]

1 hour ago, PeterSt said:

Two days ago I revisited Houses of The Holy and Physical Graffiti (LZ) to try out that strange hollow colored cymbal sound on these two, with my current settings and everything. Well, it seems to be gone. Mind you, this is so strange already on the LP, but MQA manages to remove it. And didn't I tell about a "too white" sound of MQA ?

 

Maybe MQA reduces the resolution enough to remove disturbing extra details? A bit like MP3/AAC can do for really bad recordings by removing some of the distortions by deciding those are not worth encoding.

 

You could probably get similar result with similar noise shaped dithering to 13-bit resolution or so. No need for MQA. Try with iZotope RX6?

[Miska]

24 minutes ago, Charles Hansen said:

We don't know the microphones, preamps, or cables, but many of 2L's recordings are done in multi-channel. Apparently most of the recent ones use the A/D converter built into the Pyramix DAW made by Merging Technologies of Switzerland. I've no idea how that sounds, as if you think about it, all of the same factors that affect the sound quality of a DAC also affect the sound quality of an ADC:

 

They usually list the most important pieces in the recording literature, like for 2L-38:

"This recording was made with DPA microphones, Millennia Media amplifiers and SPHYNX2 converters to a PYRAMIX workstation, all within the DXD-domain."

 

There are also some fairly detailed pictures of the recording events in their literature.

 

The better DPA mics use 130V phantom power, so that also limits the set of available microphone amplifiers.

 

Some related things I found somewhere in the past:

http://www.dpamicrophones.com/microphones/ddicate/4003-omnidirectional-microphone-130-v

https://www.mil-media.com/HV-3C.html

https://www.mil-media.com/HV-3D-8.html

http://www.digitalaudio.dk/AX24-ADDA-Converter-.2111.aspx

 

 

Channel Classics on the other hand, IIRC, uses also B&K / DPA mics and the Grimm AD-1 converter. I believe the mic pre-amps are custom built. I believe they now also use Merging Horus converter.

 

[Miska]

1 minute ago, crenca said:

 

After a little bit of googling, apparently there was a controversy about rather there was enough beryllium in the cone material for Usher to claim that is was a "beryllium" driver.  Apparently (if I am following the controversy correctly) all "beryllium" cones are in fact an "alloy" (i.e. a mix of beryllium and some other metal), and this reality gives manufactures much room to drop a little (some small percentage) of beryllium in their cones and claim to have a beryllium driver.

 

Yet more tales from the crypt of subjective Audiophiledom...

 

There is a nice way to find out material properties of these metal dome tweeters. Just check out the frequency where the about +20 dB dome breakup resonance is. For aluminum domes it is around 25 - 30 kHz (combine that with leaky digital filters for some ear piercing fun). For harder materials it can get up to around 50 kHz.

 

Soft dome tweeters typically have that resonance below the cross-over frequency.

[Miska]

16 minutes ago, mansr said:

In the B&W 600 series the aluminium dome resonance is at roughly 25 kHz. In the more expensive CM series, still aluminium, it is at 35 kHz or so. Can't say I can hear any difference from this.

 

Maybe you don't use as leaky filters to put much energy there, although it doesn't necessarily need input at that frequency to trigger it...

 

In any case, the main point was that it is good indication of material properties of the tweeter...

 

[Miska]

I like (and use) Supra cables, not much B$, and the price isn't too bad. And they conform with my view on how cable properties should be, and publish the relevant specifications...

 

http://www.jenving.com

 

[Miska]

3 hours ago, Em2016 said:

But as you pointed out, their Apple Music subscriber numbers continue to climb even with just 256k. And Spotify's numbers continue to climb with just MP3320k, so it's no wonder neither of those companies is in a rush for Hi-Res streaming.

 

Spotify uses free Vorbis codec, not MP3...

 

[Miska]

2 hours ago, PeterSt said:

Another mistake for comparison (read : I shouldn't have done it) is that the MQA version will have been taken from the "DVD rip". So what happened in the end is that I compared a 2ch Redbook with a from 5.1 downconverted to 2ch DVD version.

 

One thing I don't understand is why people compare RedBook and MQA versions and not original hires and MQA?

 

DVD's usually have uncompressed 2-channel LPCM version too, in addition to a lossy compressed DD/dts 5.1. And quite frequently for this kind of album the original recording is 48 kHz 24-bit, and the RedBook is conversion from that one. So the better starting point would be to rip the DVD 2-channel version as audio-only and then compare against that.

 

I have ripped audio from all the concert DVD's into FLAC (both stereo and 5.1). So I don't need the original discs and don't need to deal with the video part.

[Miska]

On 11/17/2017 at 5:29 PM, mcgillroy said:

On top of it they keep pushing their narrative that DACs with 

out MQA are incomplete: „Bryston BDA-3 D/A processor ($3495)
While it was frustrating not to be able to decode MQA files...“

 

One needs to attempt very very hard to get any MQA files in first place, apart from couple of demo tracks...

 

I'm still curiously waiting where can I download/buy MQA encoder so I could create some files of my own...  No problem getting MP3 or AAC encoder, or something else along those lines.

 

[Miska]

35 minutes ago, Lee Scoggins said:

I think you have to evaluate the sonic merits and business elements separately.  It is possible that MQA sounds better and the business elements are not good in terms of licensing, etc.  It is possible the sonic merits are good and there are benefits to the business aspects.  It is possible that none of it is good.  There is, unfortunately no objective way to score the business pluses and minuses.  Everyone will have an opinion.

 

I think it is not too hard to score business effect on consumers, at least some of the minuses:

- Severely restricts legal use of the purchased content

- No assurance about long-term usability of the technology

- Licensing costs will be paid by the consumer in the end

- Gives a single entity broad (end-to-end) point of control

 

Standard FLAC doesn't have any of these minuses.

[Miska]

9 minutes ago, Samuel T Cogley said:

I know that MQA and HDCD are vastly different from a technical and consumer standpoint

 

They certainly have a lot of same elements. Need of both ends to pay licenses for the closed technology, degraded performance without decoder, encoded content hiding in LSBs of "legacy/compatibility stream". If the single company goes extinct, the technology becomes inaccessible without reverse-engineering effort.

 

[Miska]

1 hour ago, Lee Scoggins said:

Fair points but I think FLAC is not a great analogy as it does not include a process for improving such as the apodizing filters.

 

It shouldn't! And there's no need. There is totally no need for content delivery vehicle to deal with such things. Same goes for dynamic range compression too. We can have those apodizing filters and dynamic range compression per-need basis at the player device side. Over time we can improve the apodizing process, just like has been going on with CD players. Content stays the same, tomorrow you just play it with improved processing and get improved playback result. Same goes for dynamic range compression, I may need it when I listen in my car, but I don't need it when I listen with headphones at home. My car system can apply the dynamic range compression.

 

I've had apodizing filters in my playback system for nearly 20 years now.

 

There's no need to irrevocably bake something like this into the content itself. Keep the content intact and leave playback gear take of those details.

 

Same goes for digital room correction, you are not going to be able to apply it generally at mastering stage, it needs to be processed separately for everybody's own room. Same goes for other DSP stuff too.

 

[Miska]

4 hours ago, mansr said:

The older ones (Micro, Nano, iDAC2) use the built-in filters of the TI DSD1973 DAC chip. I don't know if the newer (black) models do things differently.

 

iDSD Micro Black Label is same as the non-BL in that respect. It has the same firmware and DAC, but improved analog section. Now especially DSD512 gives better performance than before.

 

I think the new Nano Black is quite a bit different in this respect from the other models. Probably uses same XMOS code from MQA as the Pro-Ject DAC does.

 

[Miska]

1 minute ago, beetlemania said:

Meanwhile, I found that John Atkinson DID send his raw files plus impulse response to MQA 

Why on Earth would he do that?

 

Because MQA doesn't let anybody have the encoder (without $20k encoder-box they sell). They sell encoding service where you pay to get your tracks encoded. And probably requires signing very restrictive NDA to get any kind access to any of this.

 

[Miska]

1 hour ago, The Computer Audiophile said:

Perhaps I'm wrong, but I always thought HDCDs offered redbook quality on standard players and "enhanced" quality on HDCD players. I don't believe there is a version of degraded quality, but I could be wrong.

 

Just like MQA claims? Sounds familiar?

 

In practice the HDCD makes you lose one or two of the precious 16 bits of RedBook, standard players will clip the peaks harder due to the way it is encoded and instead of proper dither the LSB is not working the way it should (very important for CD too) causing more loss of precision. As result HDCD is not properly dithered.

 

Sony's SBM which was earlier developments of noise shaped dither for CD use was better approach and didn't need any special playback hardware or "decoding".

 

[Miska]

1 hour ago, beetlemania said:

So, MQA can only be created from the raw tracks, not the final master?

 

No, from the final master... Don't know why he would send anything else. I'm pretty certain nothing else than final master has been sent for all those Tidal masters...

 

My point was just that one cannot encode MQA from the final master on their own.

 

[Miska]

Just now, davide256 said:

And conversely I bought a Linn Genki HDCD player in 2001 because it was the first time I heard a CD format that sounded  like music instead of torture 

 

Only the few HDCD discs, or in general?

 

AFAIK, Linn never had any special DSP, just COTS stuff on the digital side. IIRC Wolfson's DAC chips, which, BTW, some have apodizing digital filters.

 

[Miska]

39 minutes ago, mansr said:

Even large companies often tolerate competition. I don't see Ford pushing "tech" that would make every Mercedes incompatible with new roads.

 

Reminds me of clueless press reporting based on Fraunhofer & Sisvel & Technicolor's press release about end-of-licensing this spring that now MP3 is now dead and everybody should move to AAC and other codecs which of course was highly recommended in the same press release.

 

Because the patent licensing ended. Which happened because all relevant patents have expired across the globe.  So now everybody can use MP3 without paying royalties (that lovely minimum $20k/year that all small companies must love). But since it is open standard, it also means that people can now freely develop codecs and it by no means ends use of MP3 in products.

 

If it would have been like MQA, it could actually be dead by now without any new products coming out. And everybody could be required to upgrade new shiny hardware with MQA2.