MQA is Vaporware

[soxr]

1 minute ago, mansr said:

Moreover, if those alias artefacts are inaudible, what's the point in high-res, and by extension MQA, in the first place?

The same question can be asked if real ultrasonics (real upper harmonics of the musical content) are audible. This is still a big debate.

Some years ago we did a test with content that had 20 Khz + content: the DVD-A version of this.

https://www.discogs.com/Laurence-Juber-Guitar-Noir/release/8486906

A very good sounding highres album. I had the 24/96 version, and cropped a one minute piece out of it, saved as 24/96 WAV. Made sure the file had ultrasonics to begin with so ran a spectrum through audacity.

I used my sox method as disclosed in my first post here, to downsample to 16/44.1 and upsample back to 24/96, with rate -vsM . Posted both 24/96 files on some forum and only one hifi dealer with super expensive gear could tell the files apart.

All the other forum members reacted like: great resampler you have there, can't hear the diff.

I did this test because I had this test improvised live in foobar: I used a secret rabbit code downsampler + upsampler in the DSP list which I saved as a DSP profile so I could quick switch, and then I could live switch between native 24/96 and the band limited version in realtime. I created this on the fly as some customer asked: does highres matter. So we live tested this and they could not believe it. Foobar was running in WASAPI or kernel streaming bitperfect mode, so it did not interfere with windows resampling.

Something similar is also documented here:

https://secure.aes.org/forum/pubs/journal/?ID=2&pg=2

So yes, the arguments in the MQA AES paper debunk the need for highres.

[soxr]

6 minutes ago, Fitzcaraldo215 said:

Sorry, can you give me the entire context of the quote before I agree with your conclusion?

http://www.aes.org/e-lib/browse.cfm?elib=17501

[soxr]

9 minutes ago, Fitzcaraldo215 said:

Greater timing accuracy is what is conveyed by hi rez

 

If you study Monty's video's from xiph.org, you would know that band limited PCM has already infinite timing resolution.

Those hammering on time-smear should watch from 21:00 as many times until they get it.
 

 

[soxr]

 

10 minutes ago, mansr said:

MQA discards everything above 48 kHz and replaces it with rubbish. There are only two possibilities here:

1. Frequencies above 48 kHz are audible.
2. Frequencies above 48 kHz are not audible.

This leads to two possible conclusions:

1. The mangling by MQA causes audible distortion.
2. MQA is pointless.

There really are no other options.

A twist from their paper:
 

 

Quote

Even though some musical instruments produce sounds above 20 kHz [53] it does not necessarily follow that a transparent system needs to reproduce them; what matters is whether or not the means used to reduce the bandwidth can be detected by the human listener.

 

 

[soxr]

8 hours ago, Fitzcaraldo215 said:

Filtering above audibility also conveys advantages, as does increased bit depth

Sampling at higher rates than 44.1K has some advantages: the AA filter can be outside the treshold of human hearing range, it can be less steep, ... and any DSP artefacts can be shifted to ultrasonic range where you can't hear them.

So basically highres is not about more music content (bit depth mainly determines noise floor), but about avoiding audible artefacts in the baseband as you have more places to hide mistakes into parts of the data that can't be heard.



 

[soxr]

27 minutes ago, Fokus said:

 

I asked for this before.

 

They need two quadrature mirror filter pairs for the split and the join.

 

The following (pretty dire)  filter response always surfaces when an MQA DAC is tested with non-MQA signals:

 

I recently asked Mytek if they always enforce the minimum phase upsampler MQA is using.
Mytek was against it, but MQA tried to force it anyway.

So I believe MQA is making sure redbook will sound worse than MQA because of those crappy filters being active.

But Mytek refused, and you can turn off the MQA decoder and turn off the crappy minimum phase upsampler.

It's crappy because it's limited in #taps and cannot be compared to minimum phase resampling such as in libsox (which auralic is using, just dump their firmware and run ldd on lightningserver and it will reveal libsox) or the soxr library, which is the resampling-ony part of libsox.

[soxr]

7 hours ago, Digital Assassin said:

Well, Darko is at it again...another massive shill:

 

 

"The point here is that, once again, with this one album, heard via PS Audio’s glorious D/A converter, keeping one eye on Godwin’s Law, MQA sounds different. And to my ears, that difference translates to better."

 

http://www.digitalaudioreview.net/2017/07/mqa-donald-fagen-ps-audio/

This article again mentions that the source is corrected, but from what we've seen from archimago's/mansr's work, the source is not corrected, but the encode contains a flag, so that the renderer can post-process using one of 32 anti-ringing filters.

If the source was corrected, the waveform would be drastically different from let's say DXD vs upsampled MQA with minimum phase. I did not find evidence of pre-processing.

[soxr]

19 minutes ago, mansr said:

What I'm getting at is whether the digital releases are sourced from analogue tape.

 

Those old CD's are typically from analog tape, but was it ADD or AAD ?
Reminds me of SPARS code:

https://en.wikipedia.org/wiki/SPARS_code

 

[soxr]

7 hours ago, esldude said:

Would have been more obvious is they named it MQAQM.

 

Master Quality Authenticated Quality Master

 

 

Master Quality Approximated ?

 

[soxr]

6 hours ago, Sal1950 said:

It's all about the "Deblurschitt"  

At least real studio engineers authenticate MQA not to be the original studio sound, but:
 

Quote

MQA is, in all of my tests so far, brighter and thinner with more distortion, and a sense of excitement (remember Aphex?) and even volume from the artifacts. Sure many lay people will be fooled. I don't like to see pros fooled, but oh well, that's why I'm busy I suppose and they're posturing next to a new product so I'm free to tell the truth here.

https://www.gearslutz.com/board/12548751-post460.html

So MQA deblur (killing post-ringing of minimum phase resampling in the renderer phase with weird filters) is the shit sold to audiophiles as the new PCM, but it's not the studio sound.

And let's remember: all DSP applied by MQA is still in the PCM domain.
 

[soxr]

MQA will get competition from DSD streaming.

Qobuz is going to offer DSD streaming somewhere between september and the end of this year.

http://blogsv2.qobuz.com/qobuz-blog-en/2017/04/03/march-29-2017-our-new-announcements-qobuzisback/ (mentions september)
https://www.alpha-audio.nl/2017/08/qobux-komt-richting-einde-jaar-dsd-streams/ (mentions end of the year)

This is good news, as for the DSD format, it's all well documented and implemented:

- all major players can pass DSD via DoP to DSD dac's
- all major players can decode DSD to PCM and these libaries are also available as FOSS / open source
- postprocessing decoded DSD as PCM works pretty good
- DFF and DSF formats are well documented and open
- DoP is a well documented packing format for DSD packed in PCM

a lot of players already support DSD, so to add DSD via HTTP instead of via file playback is not a big change in code, and players such as squeezelite can already to DSD over HTTP, and logitech LMS has extentions for DSD. Other open source players such as MPD also do DSD.

Doing this over HTTP will be a small change for most file based players, as they can use the excellent curl libraries to read the file from HTTP.

[soxr]

9 minutes ago, George Hincapie said:

Why is Quobuz so expensive? £350 a year? WTF? I pay £20 a month for Tidal HIFI.

From the article:

1. these are real lossless 24 bit files, not lossy fake 24 bit files like MQA (as MQA is not 24 bit due to higher noise floor)
2. 60,000 albums in Qobuz 24-Bit Hi-Res  (compared to how many on tidal?)
3. listen either streamed or offline

I pay 20 euros per month for Tidal, which is 240 / year. For 110 euro extra, I have access to real studiomasters, and a much larger catalog of studiomasters. And I get to keep my existing 7.000 EURO R2R dac without having to buy a lesser MQA DAC (I have a metrum adagio, no MQA dac is going to come even close).

A good deal to me.

[soxr]

3 minutes ago, Sal1950 said:

How do you define a "real studiomaster"?

Just curious.

 

The lossless master, not some lossy approximated DRM'ed crypto format derived from the master.