MQA: A Review of controversies, concerns, and cautions

[Rusty Shackleford]

No need to involve an analogue step. Here's the spectrum of an MQA-CD file, the same file after decoding, and the high-res original:
View attachment 35310

This is great info. Is this graph from a longer writeup? If so, could you share the link?

 

[RichB]

This morning I revisited a review of MQA CDs by Techmoan on youtube, and interestingly he concludes that the MQA "unfolded" sound sounds much better than when just playing it as CD.
He draws the wrong conclusion in my opinion (MQA is better), where it is more likely that while the CD is compatible for normal playback, the sound itself is compromised unless you use MQA. This would also be in line with the illustrations in the patent.

Let's see what's wrong with this video:

1) He's using a faulty MQA/CD.
2) He does not understand that MQA CD to a hobbled CD.
3) He thinks that MQA and HD Audio are equivalent but they are not.
4) He does not understand that engaging MQA also changes the reconstruction filter.

It's imprudent to pronounce MQA to be superior based on a broken MQA CD and a single listening session.

- Rich

 

[JJB70]

The selling of classic old recordings as high res is a scam for the most part. The only ones I consider make sense are those re-issued as quadrophonic or surround mixes, that's not because of any upsampling high resolution improvements but the more immersive experience. I have a few classical blu ray discs and if I use the disc there is no doubt that having for example the full Karajan Bruckner symphony cycle on a single disc is good and having the Steinberg Holst Planets in quadrophonic is splendid but I see no improvement over the CDs for two channel.

 

[speedy]

Well I'd like to correct misinformation I promoted in error. MQA is more like 15 bit CD. I made the error reading some of their patents, but in the eventual implementation MQA is only using the lowest bit for the folding.

Edit 1: Changed MQA decoded to 15 bits.
Edit 2: Added MQA native 13 bits.

I think it should be clarified that you are both talking about just MQA-CD and not MQA streaming, right?

MQA Streaming is 16/44.1 or 16/48 when a 24/44.1 or 24/48 files is played back without decoding.
MQA Streaming is 24/88.2 or 24/96 when a 24/44.1 or 24/48 files is played back with decoding.
MQA claims >88.2kHz when rendered on a supported DAC like the Matrix X-SABRE Pro MQA.

 

[RichB]

I think it should be clarified that you are both talking about just MQA-CD and not MQA streaming, right?

MQA Streaming is 16/44.1 or 16/48 when a 24/44.1 or 24/48 files is played back without decoding.
MQA Streaming is 24/88.2 or 24/96 when a 24/44.1 or 24/48 files is played back with decoding.
MQA claims >88.2kHz when rendered on a supported DAC like the Matrix X-SABRE Pro MQA.

MQA claims many things. I think it is wrong to claim a sampling rate other than the actual sampling rate.
It would be more accurate to claim the frequency extension provided by MQA origami.

It is equivalent to Roon claiming 196k simply because I can up-sample to that rate. After all, the Oppo DAC will display that rate.

- Rich

 

[speedy]

MQA claims many things. I think it is wrong to claim a sampling rate other than the actual sampling rate.
It would be more accurate to claim the frequency extension provided by MQA origami.

It is equivalent to Roon claiming 196k simply because I can up-sample to that rate. After all, the Oppo DAC will display that rate.

- Rich

What about my first question though...

Were you referring to MQA-CD or MQA Streaming/Files though?

...focusing on dynamic range instead of sample rate.

 

[Blumlein 88]

I think it should be clarified that you are both talking about just MQA-CD and not MQA streaming, right?

MQA Streaming is 16/44.1 or 16/48 when a 24/44.1 or 24/48 files is played back without decoding.
MQA Streaming is 24/88.2 or 24/96 when a 24/44.1 or 24/48 files is played back with decoding.
MQA claims >88.2kHz when rendered on a supported DAC like the Matrix X-SABRE Pro MQA.

I'm at a bit of a loss to examine this in detail. Until very recently my internet wouldn't support TIDAL download speeds, so I haven't used Tidal. I've seen with others using Tidal the streaming rate is around 1.6 mbps for 24 bit versions of material.

So I'm perhaps mistaken, but believed MQA streams are generally 24 bit. Any streamed as 16 bit I thought would decode into 88 or 96 16 bit music. Also MQA CD's decode into 88 or 96 khz 16 bit. These would stream at around 1 mbps.

So 16 bit MQA streams use the 16th bit for the first fold. These unfold to higher rate 16 bit.

Any 24 bit MQA streams use the lower 7 bits for folding. And unfold into 24 bit.

 

[RichB]

What about my first anthough...

Were you referring to MQA-CD or MQA Streaming/Files though?

...focusing on dynamic range instead of sample rate.

I quoted this post:

This morning I revisited a review of MQA CDs by Techmoan on youtube, and interestingly he concludes that the MQA "unfolded" sound sounds much better than when just playing it as CD.
He draws the wrong conclusion in my opinion (MQA is better), where it is more likely that while the CD is compatible for normal playback, the sound itself is compromised unless you use MQA. This would also be in line with the illustrations in the patent.

This is an MQA CD. If streaming files are MQA, it seems likely that they are also 44.1 to reduce bandwidth.

Dynamic range because that is one aspect of HD Audio.
Many on this forum would like to see products capable of producing the full dynamic range of a CD, 96dB.
This seems like a reasonable baseline.

From what I have read, an MQA encoded download will be 44.1/13 bits when not decoded.
I don't wish to be constrained by MQA's proprietary reconstruction filter which is marketed as Deblurring.
I believe Archimago has some measurements and they show poor reconstruction performance and nothing that would indicate reduction of "blur" whatever that is.

- Rich

 

[Blumlein 88]

These links have been in this thread a couple times. By Jim Lesurf, I think it is the best least technical explanation of what MQA is doing. Do note the author was using the patents and based his examples for 16 bit on MQA using the bottom three bits for folding and not just one bit. Still a good explanation. I'd suggest reading the three articles in this order.

http://www.audiomisc.co.uk/MQA/origami/ThereAndBack.html

http://www.audiomisc.co.uk/MQA/bits/Stacking.html

http://www.audiomisc.co.uk/MQA/cool/bitfreezing.html

 

[speedy]

@Blumlein 88 & @RichB - Based on what you're both saying, please correct me if any of this is wrong (setting aside the fact that I know both of you would prefer lossless/native PCM files):

1. The MQA-CD format's patent states the it would only be 13/44.1 when not decoded (only a pre-release spec though)
2. The MQA-CD format is actually 15/44.1 when not decoded in spite of what the original patent states (no unfold)
3. The MQA-CD format is up to 16/96 when decoded (1st unfold)
4. The MQA-CD format is up to 16/384 when fully rendered by a licensed DAC (2nd unfold)
5. The MQA File (streaming) format is up to 16/48 when not decoded (no unfold)
6. The MQA File (streaming) format is up to 24/96 when decoded (1st unfold)
7. The MQA File (streaming) format is up to 24/384 when fully rendered by a licensed DAC (2nd unfold)

Also, it seems to me like the MQA-CD format is dead and barely deserves discussion... Aren't most involved in this thread here to discuss MQA files/streaming?

Basing #1 & #2 on this:

Well I'd like to correct misinformation I promoted in error. MQA is more like 15 bit CD. I made the error reading some of their patents, but in the eventual implementation MQA is only using the lowest bit for the folding.

 

[Blumlein 88]

@Blumlein 88 & @RichB - Based on what you're both saying, please correct me if any of this is wrong (setting aside the fact that I know both of you would prefer lossless/native PCM files):

1. The MQA-CD format's patent states the it would only be 13/44.1 when not decoded (only a pre-release spec though)
2. The MQA-CD format is actually 15/44.1 when not decoded in spite of what the original patent states (no unfold)
3. The MQA-CD format is up to 16/96 when decoded (1st unfold)
4. The MQA-CD format is up to 16/384 when fully rendered by a licensed DAC (2nd unfold)
5. The MQA File (streaming) format is up to 16/48 when not decoded (no unfold)
6. The MQA File (streaming) format is up to 24/96 when decoded (1st unfold)
7. The MQA File (streaming) format is up to 24/384 when fully rendered by a licensed DAC (2nd unfold)

Also, it seems to me like the MQA-CD format is dead and barely deserves discussion... Aren't most involved in this thread here to discuss MQA files/streaming?

Basing #1 & #2 on this:

I think that is correct other than possibly #4. I'm not sure the MQA CD unfolds past 96 khz though I am not certain.

I would also note it appears MQA doesn't really have any info beyond 96 khz. Unfolds higher are simply noisy upsamples.

Supposedly MQA CD are more popular in Japan.

What are most here to discuss? Probably Tidal streaming I would guess.

 

[tmtomh]

No MQA source - CD, digital file, or streaming - has an actual sample rate above either 88.2kHz or 96kHz. Anything above that is fake - it's MQA upsampling the source to generate a "192" or "384" or whatever readout on the playback equipment. But it's simply a doubling (or quadrupling) of the highest frequency/sample in the MQA file.

The reason for this is that regardless of the sample rate of the original PCM file, MQA encoding downsamples to either 88.2 or 96k. So if you have, for example, a 24/192 PCM original, MQA will throw out half the samples to make it 24/96k. Then it will encode that 24/96k source, using lossy compression to encode half the samples in order to create a 24/48 file, with the higher samples "folded in" to the lower bits of the 24/48 file.

When the first unfold happens, the file is decoded from 24/48 to 24/96 (albeit it is not restored to the original 24/96 PCM, because lossy encoding was employed during MQA encoding). Then with the "final render," the file is upsampled to 192k. But make no mistake, that 192k does not restore any samples lost when the original 24/192k PCM file was downsampled by the MQA encoder to 96k.

Now, there is nothing necessarily wrong with upsampling a file during playback - many software playback apps offer that as an option, and of course almost all DACs perform upsampling of incoming material before decoding to analogue.

But MQA is unique in that they don't call it upsampling or advertise it as upsampling, or clearly explain that it's just upsampling. Instead, they create the misleading impression that their "final render" recovers or decodes an extra level of sample rate beyond the first unfold - which is just flat-out false. If you listen closely to Bob Stuart's video interviews from 2-3 years ago, he almost always says something like, "then we put the sample rate back to what it was." What he's saying there is, "we upsample the file so the sample rate appears to be what the original studio master PCM's sample rate was."

IMHO it's pure charlatanry.

 

[RayDunzl]

No need to involve an analogue step. Here's the spectrum of an MQA-CD file, the same file after decoding, and the high-res original:
View attachment 35310

A HiRes and MQA-CD file of what?

 

[dc655321]

A HiRes and MQA-CD file of what?

Whatever track 2L-053 corresponds to on 2L's list of freely available test tracks.

I assume...

 

[RayDunzl]

Whatever track 2L-053 corresponds to on 2L's list of freely available test tracks.

I'm curious because there appears to be no content above -100dBfs in the graphic.

 

[dc655321]

I'm curious because there appears to be no content above -100dBfs in the graphic.

Chief Scrutinizer indeed!
Maybe the curves are the residuals (null test)? Dunno. Have to ask @mansr.

 

[Sal1950]

The selling of classic old recordings as high res is a scam for the most part. The only ones I consider make sense are those re-issued as quadrophonic or surround mixes, that's not because of any upsampling high resolution improvements but the more immersive experience. I have a few classical blu ray discs and if I use the disc there is no doubt that having for example the full Karajan Bruckner symphony cycle on a single disc is good and having the Steinberg Holst Planets in quadrophonic is splendid but I see no improvement over the CDs for two channel.

Agreed, but we can't discount the stereo hi-rez releases that have been remastered by some wizards like Steven Wilson. The increase in SQ could have been just as easily done on a Redbook CD, but the tag of HiRez sells product.

 

[firedog]

If a HiRes track is taken from a tape recording made 40 years ago, it can't be a HiRes track. Any and all analogue tape recorders 40 years ago managed 20kHz at a push, 18 kHz flat at 15ips and provided the HF was kept reasonably low. So, when digitised, that tape recorded track will have nothing at all above 20kHz that can't be fully capture with a 44.1kHz sample rate. Everything above that is noise, tape noise, amplifier noise, nothing whatsoever of the original recording. The very few digital tape machines of the era sampled at 50kHz so again, nothing above 20kHz or thereabouts. (Bop Till You Drop I think was the first major label pop recording done digitally and released in 1979)

Those selling classic rock/pop recordings as HiRes are merely scamming the public and so perpetrating a fraud on the record buying public.

S.

Technically true, but also not relevant in some ways. I have lots of old jazz and rock remastered to either SACD/DSD or hi res pcm from tape. Many that are just remastered, and not even remixed. In many cases, they are the best sounding version of the music available. I don't really care whether they are "true hi res" or not. I care that they sound good. If they sound better than a vinyl or CD version I already own, why should I care whether it's what you call hires or not? I care that the hires remaster (and possibly BECAUSE it was a hi-res remaster) was done skillfully and captured as much information as possible from the tape master.

Personally, I think tape to DSD conversions often sound better and "more like the original" than conversions to PCM. That's just a personal preference, but I'm glad to have them. The Bob Ludwig remasters of the 60's Stones records to SACD sound great, IMO. A bit better than the CDs made from the same master by him.

Digital has also improved over the years, and sometimes more recent remasters to digital sound better than previous attempts. In addition there are lots of tape masters that have been archived in DSD or hi-res PCM. The tape isn't being used anymore. If these are the best versions available, and very close to the sound of the master tape, I'm glad to have them.

 

[sergeauckland]

Technically true, but also not relevant in some ways. I have lots of old jazz and rock remastered to either SACD/DSD or hi res pcm from tape. Many that are just remastered, and not even remixed. In many cases, they are the best sounding version of the music available. I don't really care whether they are "true hi res" or not. I care that they sound good. If they sound better than a vinyl or CD version I already own, why should I care whether it's what you call hires or not? I care that the hires remaster (and possibly BECAUSE it was a hi-res remaster) was done skillfully and captured as much information as possible from the tape master.

You may not care what they're called, I do. If something is described as HiRes, but can't possibly be, then that's fraud, and I don't want to be taken in by fraud. Words and their precise meaning matters to me a lot.

S.

 

[mansr]

I'm curious because there appears to be no content above -100dBfs in the graphic.

That plot is from a quiet part of the file to better show the added noise.