MQA Software / Hardware Decode Etc... Questions

[mansr]

Help me out , why couldn't MQA be pressed to provide that?

 

If the labels were dishonest when their files could be checked, what possible reason would they have to suddenly become honest when they can't be verified? MQA only assures that you got what the label wanted you to have. If they wanted to send a fake, that's that.

[bcwang]

If the labels were dishonest when their files could be checked, what possible reason would they have to suddenly become honest when they can't be verified? MQA only assures that you got what the label wanted you to have. If they wanted to send a fake, that's that.

 

Why can't MQA files be checked for cd upsampling like normal flac files? The analog output could be monitored for a 22khz cutout. Or the same technique used to software decode the first stage gives digital output to do the same.

[mansr]

Why can't MQA files be checked for cd upsampling like normal flac files? The analog output could be monitored for a 22khz cutout. Or the same technique used to software decode the first stage gives digital output to do the same.

You're not supposed to have access to the decoded stream. Non-licensed software and hardware is only meant to see the low-quality version which by definition has no high-frequency content. Monitoring the analogue output isn't reliable since the "rendering" process creates artificial high-frequency content not present in the original recording. It is also far more costly to do at scale than a simple spectral analysis of digital files.

[labjr]

And the labels are doing it again with MQA. Case in point, Madonna's Like a Virgin. The difference with MQA is that distributors like HDtracks can no longer verify what they've been given (with officially sanctioned means). All we have is the word of the labels, and as we've seen that isn't worth much.

 

Btw, if someone provides me a short sample (10 seconds will do) of the MQA version of that Madonna album, I can tell what resolution it claims to originate from. Might be interesting.

That Madonna album sounds like crap. It was recorded at 16-44 sample rate. And I think digitally mixed. I highly doubt MQA can help!

 

Sent from my XT1528 using Tapatalk

[PeterSt]

Btw, if someone provides me a short sample (10 seconds will do) of the MQA version of that Madonna album, I can tell what resolution it claims to originate from. Might be interesting.

 

This is 48/96, but originally it is 24/192.

Practically ?

Material Girl.zip

[firedog]

Gotta say I mostly side with Chis in this discussion. Whether MQA sounds better than a non-MQA file from the same source remains to be heard. So far we have speculation based on what both sides think "has to be" true.

 

As far as added/high volume compression, I hate it as much as anyone. Especially when done to legacy material that basically sounded good anyway, and only needs some "tweaking" to sound good in a remaster.

 

That said, the reality is the big labels aren't producing content for audiophiles. At best we are an afterthought to them. So hoping they will stop producing content with added VC is basically a pipe dream. We are a tiny fraction of a fraction of the market and have no market power with the big labels. At the small labels the situation is very different, but they probably won't go in on MQA anyway.

 

In that situation, I take it as a positive that MQA is authenticated - I think it's a good thing that as a buyer I know I'm getting what was produced, and not something altered after being mastered/released. Whether I actually think it has good SQ is a different issue. And yes, even for an album that's been volume compressed, I don't think the "Q" in MQA is deceptive. It means you are getting the master they produced. We don't have any way of knowing that with any other digital format.

 

Again, whether I like that "master" is a different question. Like many here, I often don't buy modern remasters of albums b/c of the loudness wars factor. But that's true whether something is MQA'd or not.

[mansr]

This is 48/96, but originally it is 24/192.

Practically ?

 

The MQA metadata indeed indicates an original rate of 192 kHz. Very authentic.

 

In reality, it has been recorded with a sharp cut-off just above 20 kHz. Let's see what the MQA version looks like in its various forms:

 

 

Decoding (red) extends the frequency range to 48 kHz and adds a sloping noise floor with a couple of tones at about 29 kHz and 42 kHz. This is consistent with playing the 44.1 kHz original through an analogue stage and capturing at a higher rate. Rendering (green) doubles the frequency range again. Note that everything above 48 kHz (where the red trace stops) is merely a mirror image of the lower frequencies. This is the upsampling filter used by the renderer:

 

 

A proper 2x upsampling filter should drop off at the original Nyquist frequency, in this case 48 kHz. What were they thinking?

[Fokus]

What were they thinking?

 

That was explained, perhaps a bit too implicit, in the earliest MQA documentation three years ago: filters are made as lazy as possible within a specific constraint of tolerated aliasing and imaging. This constaint is informed by auditory perception, by the actual music programme and its innate noise, and (presumably, hopefully) by some intend not to blow up amplifiers and tweeters.

[mansr]

That was explained, perhaps a bit too implicit, in the earliest MQA documentation three years ago: filters are made as lazy as possible within a specific constraint of tolerated aliasing and imaging. This constaint is informed by auditory perception, by the actual music programme and its innate noise, and (presumably, hopefully) by some intend not to blow up amplifiers and tweeters.

 

Well, it's not like we can actually hear any of those frequencies. Is that what they mean?

[PeterSt]

This is 48/96, but originally it is 24/192.

 

I now see how confusing this is. I better should have said : this is 24/96 because that's the format I delivered in the zip.

Point is : what I tried to say is that this is a 24/48KHz which has been expanded/unfolded to 24/96KHz. IOW, the core decode has been done. So or you did not take this into account (and something strange might have happened), or you did and can re-do the decoding out of this already decoded one ?

[mansr]

I now see how confusing this is. I better should have said : this is 24/96 because that's the format I delivered in the zip.

Point is : what I tried to say is that this is a 24/48KHz which has been expanded/unfolded to 24/96KHz. IOW, the core decode has been done. So or you did not take this into account (and something strange might have happened), or you did and can re-do the decoding out of this already decoded one ?

 

The zip contains undecoded MQA at 24/48, which was actually what I wanted.

[PeterSt]

I now see you are correct. This wasn't even intentional ... but OK if you wanted it like that.

[Fokus]

Well, it's not like we can actually hear any of those frequencies. Is that what they mean?

 

Yes. Aliasing and imaging are allowed when MQA deems them innocuous.

[mansr]

Yes. Aliasing and imaging are allowed when MQA deems them innocuous.

 

Sure, 100 years of mathematics are suddenly invalid because Bob said so.

[Jud]

Yes. Aliasing and imaging are allowed when MQA deems them innocuous.

 

Sure, 100 years of mathematics are suddenly invalid because Bob said so.

 

Within these "constraints," is there any provision you can see that would be "smart" enough to avoid audible levels of IMD?

[mansr]

Within these "constraints," is there any provision you can see that would be "smart" enough to avoid audible levels of IMD?

There is no reason to allow the levels of aliasing seen here, ever.

[Jud]

There is no reason to allow the levels of aliasing seen here, ever.

 

I understand. What I am asking is more along the lines of whether there is a reason *not* to allow them. I posed my initial question about IMD, but please feel free to include anything else that applies.

 

 

 

Edit: Essentially I'm just asking the completely unsophisticated question, "What will I hear that's wrong?"

[Fokus]

There is no reason to allow the levels of aliasing seen here, ever.

 

Bob thinks he has a reason.

[mansr]

I understand. What I am asking is more along the lines of whether there is a reason *not* to allow them. I posed my initial question about IMD, but please feel free to include anything else that applies.

 

Edit: Essentially I'm just asking the completely unsophisticated question, "What will I hear that's wrong?"

 

Allowing aliasing is wrong, plain and simple. Audibly, expect something similar to a poorly filtered NOS DAC.

[mansr]

Bob thinks he has a reason.

Well, he has to claim something novel, or there'd be no reason to buy his stuff. That doesn't mean his claims have any merit.

[Jud]

Allowing aliasing is wrong, plain and simple. Audibly, expect something similar to a poorly filtered NOS DAC.

 

That is quite helpful, thank you. (Edit: Well, the second sentence, anyway, because you told me *why* it was wrong.)

[abrxx]

Sure, 100 years of mathematics are suddenly invalid because Bob said so.

 

We are talking about ultrasonic aliasing are we not? As in, we can't hear it directly. So who cares if its aliased?

[The Computer Audiophile]

The MQA metadata indeed indicates an original rate of 192 kHz. Very authentic.

 

In reality, it has been recorded with a sharp cut-off just above 20 kHz. Let's see what the MQA version looks like in its various forms:

 

[ATTACH=CONFIG]33133[/ATTACH]

 

Decoding (red) extends the frequency range to 48 kHz and adds a sloping noise floor with a couple of tones at about 29 kHz and 42 kHz. This is consistent with playing the 44.1 kHz original through an analogue stage and capturing at a higher rate. Rendering (green) doubles the frequency range again. Note that everything above 48 kHz (where the red trace stops) is merely a mirror image of the lower frequencies. This is the upsampling filter used by the renderer:

 

[ATTACH=CONFIG]33134[/ATTACH]

 

A proper 2x upsampling filter should drop off at the original Nyquist frequency, in this case 48 kHz. What were they thinking?

 

MQA has nothing to do with the decisions of record labels.

 

You are essentially trying to quarterback a game by watching it on TV. You only have some of the information, but you're talking like you have more than the team on the field.

 

Your points would actually be accepted by more people if you didn't say things like "there's no reason for XYZ," when you really don't know if there is a reason.

[The Computer Audiophile]

Allowing aliasing is wrong, plain and simple. Audibly, expect something similar to a poorly filtered NOS DAC.

 

This is where the second leg of research should come in handy. You should be taking ABX tests to back up your claims of audibility. I can claim all day long that DSD noise in the MHz range is "wrong" because there just shouldn't be noise, but if I can't hear it, my claims are kind of senseless.

[Jud]

We are talking about ultrasonic aliasing are we not? As in, we can't hear it directly. So who cares if its aliased?

 

Intermodulation distortion, for one thing (i.e., ultrasonic tones intermodulating with each other to create audible distortion).