MQA Software / Hardware Decode Etc... Questions

[jmudrick]

This is valid, and certainly has some quarter in audiophila, but it ultimately boils down to subjectivity, the place where all the nefarious things in audiophila live.

Where actual musical enjoyment lies. Perhaps unfamiliar to some "because numbers".

 

Sent from my Nexus 5X using Tapatalk

[Samuel T Cogley]

Where actual musical enjoyment lies. Perhaps unfamiliar to some "because numbers".

 

Yes, the tired, predicable, "only subjectivists enjoy listening to music" trope. Yawn.

[Jud]

I sincerely wish you would overtly advocate for this "middle path" you sometimes hint at...

 

I sincerely wish I would get off my ass and get around to doing this also.

[jmudrick]

Yes, the tired, predicable, "only subjectivists enjoy listening to music" trope. Yawn.

Tired predictable non response to an obvious truth.

 

Sent from my Nexus 5X using Tapatalk

[Fokus]

How sure are you that correct frequency response, arguably at the cost of some time domain distortion, causes your ear/brain to form an "accurate representation" of the original?

 

Because it cannot be denied that decades of experience taught us that a correct frequency response is indeed bloody important, while there is no shred of evidence that these alleged time domain distortions matter at all, once above 20kHz.

[scintilla]

Hearing is a stochastic process that isn't easily defined by simple linear models. All attempts to pigeonhole to mathematical models that don't consider random processes and not just non-linearity but also chaotic systems fail to take a full picture. That's why there is a middleground between subjective and objective evaluation of audio implementations.

[scintilla]

Because it cannot be denied that decades of experience taught us that a correct frequency response is indeed bloody important, while there is no shred of evidence that these alleged time domain distortions matter at all, once above 20kHz.

Really? Is this sarcasm? Because a flat FR is clearly no good in a real room and just about all current DSP's aim for a modeled FR curve that follows a downward slope to HF.

[mansr]

How sure are you that correct frequency response, arguably at the cost of some time domain distortion, causes your ear/brain to form an "accurate representation" of the original?

 

That's not what I said. When a piece of recorded music is produced, there exists at some point a final studio master from which various distribution formats are derived. I prefer to have a copy of that master, and some labels are quite happy to sell it. I do not want a processed version of the master that Bob insists is somehow "better." That doesn't mean I have to listen to it without any additional processing. On the contrary, I routinely use DSP room correction to improve the sound in my room. Some people like to apply a surround upmixing process (I don't). If my processor offered a button to activate Bob's special algorithm, I'd hardly mind at all (every extra feature has a cost).

[mansr]

Really? Is this sarcasm? Because a flat FR is clearly no good in a real room and just about all current DSP's aim for a modeled FR curve that follows a downward slope to HF.

 

He said "correct." That doesn't necessarily mean flat.

[mansr]

Hearing is a stochastic process that isn't easily defined by simple linear models.

 

So what? The electronic reproduction systems are mostly linear. Our aim is to recreate the sound waves captured by the microphones. How the ear reacts to those is beside the point.

[Jud]

So what? The electronic reproduction systems are mostly linear. Our aim is to recreate the sound waves captured by the microphones. How the ear reacts to those is beside the point.

 

I believe I disagree with your last sentence. Not in (what is, for me) a relatively trivial sense like "Oh, of course how the ear reacts is the entire point," but more along the lines of telling us what we need to pay particular attention to getting right. In that sense, Fokus's response is on point.

[The Computer Audiophile]

So what? The electronic reproduction systems are mostly linear. Our aim is to recreate the sound waves captured by the microphones. How the ear reacts to those is beside the point.

 

I'm guessing you don't believe it's possible to correct for issues in the A to D converter, thus producing a version better than the master?

[mansr]

I believe I disagree with your last sentence. Not in (what is, for me) a relatively trivial sense like "Oh, of course how the ear reacts is the entire point," but more along the lines of telling us what we need to pay particular attention to getting right. In that sense, Fokus's response is on point.

Sure, and mp3 is psychoacoustically transparent.

 

Regardless, the MQA process amounts to an irreversible distortion of the signal. Such effects should be optional.

[mansr]

I'm guessing you don't believe it's possible to correct for issues in the A to D converter, thus producing a version better than the master?

 

That would have to be done at the ADC, not after mixing. And it wouldn't involve downsampling with a leaky filter.

[The Computer Audiophile]

That would have to be done at the ADC, not after mixing. And it wouldn't involve downsampling with a leaky filter.

 

So MQA is lying by saying it's correcting for A to D issues?

[mansr]

So MQA is lying by saying it's correcting for A to D issues?

I have no way of knowing. However, they also do a lot of other things. It's those other things I'd rather not have.

[Jud]

So MQA is lying by saying it's correcting for A to D issues?

 

You're talking past each other, because mansr thought you were talking about correcting *in* the A/D filter, not correcting *for issues with* the A/D filter.

[Jud]

Sure, and mp3 is psychoacoustically transparent.

 

Some of the same sorts of "masking" arguments have been used to say that various other differences must be inaudible. I tend to be skeptical of many of those as well.

 

Regardless, the MQA process amounts to an irreversible distortion of the signal. Such effects should be optional.

 

Look, I'd like it if the labels provided us downloads in the highest possible resolution too (Beatles in 24/192, nice), so there are lots of "effects" involving filtering and loss of resolution I'd prefer to be strictly optional. The best I've got now is that my dollars are optional.

[mansr]

You're talking past each other, because mansr thought you were talking about correcting *in* the A/D filter, not correcting *for issues with* the A/D filter.

It's absolutely possible to correct for (some) errors in the ADC. Most trivially, a non-flat frequency response is readily compensated for. However, any such correction, whether trivial or complex, must occur before any mixing takes place. Once multiple sources have been mixed, it is impossible isolate and correct errors introduced by each one.

 

If MQA have a method of improve the accuracy of ADCs, that's great. The thing is, this is necessarily distinct from the encoding of the final master into the MQA distribution format. It is in this latter process that the problematic filtering occurs. Sadly, MQA seem unwilling to offer one part without the other.

[Jud]

It's absolutely possible to correct for (some) errors in the ADC. Most trivially, a non-flat frequency response is readily compensated for. However, any such correction, whether trivial or complex, must occur before any mixing takes place. Once multiple sources have been mixed, it is impossible isolate and correct errors introduced by each one.

 

Interesting. Miska seems to feel his apodizing filters can correct for A/D filter ringing. Or am I misunderstanding?

[mansr]

Look, I'd like it if the labels provided us downloads in the highest possible resolution too

 

The smaller labels (Blue Coast, Sound Liaison, etc) are happily doing this. Guess where most of my music money gets spent.

[scintilla]

It's absolutely possible to correct for (some) errors in the ADC. Most trivially, a non-flat frequency response is readily compensated for. However, any such correction, whether trivial or complex, must occur before any mixing takes place. Once multiple sources have been mixed, it is impossible isolate and correct errors introduced by each one.

 

If MQA have a method of improve the accuracy of ADCs, that's great. The thing is, this is necessarily distinct from the encoding of the final master into the MQA distribution format. It is in this latter process that the problematic filtering occurs. Sadly, MQA seem unwilling to offer one part without the other.

 

As I understand it, the only adc correction is for the studio mastering adc upon mastering a new a new MQA master. That correction would be based on a subtractive input-output transfer function of the adc used for the master tape if it is known and/or the the MQA adc if taken from an analog master. The transfer function based on some objective critera they decided upon, but probably impulse response figuring prominently. I don't see that as a negative. But clearly people disagree here.

 

 

Sent from my iPhone using Computer Audiophile

[Jud]

The smaller labels (Blue Coast, Sound Liaison, etc) are happily doing this. Guess where most of my music gets spent.

 

Yes, I'm happy about this too. I will still occasionally grab an old favorite from HDTracks if it is on sale, after I've done enough checking to satisfy myself the mastering isn't awful.

 

 

For classical there are fortunately lots of choices. And for new "alternative" and world music, it is surprising how often sites like Bandcamp or the artists' own sites offer 24/96 versions at the same price as 16/44.1, often less than the price of physical CDs. Of course, with music downloaded from these sites you frequently get to hear garage band distortion in all its hi res glory. (But I wouldn't give up for anything Mitski's track "Your Best American Girl" from her "Puberty 2" album, complete with what sounds to me very much like mics or the board being overdriven by the louder passages of her vocals.)

[mansr]

Interesting. Miska seems to feel his apodizing filters can correct for A/D filter ringing. Or am I misunderstanding?

That's a special case. An apodising filter is merely a minimum phase low-pass filter with the cutoff somewhat below the Nyquist frequency which is where ringing occurs.

[abrxx]

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.[snip].

 

Bob Stuart has previously mentioned that their software to analyze an existing archive recording will do some checks to look for simple upsampling etc and report that. However, as we all know, if the label insists that is the best master, then that is what will be encoded. So MQA Ltd may be performing similar automated checks to what HDTracks are doing with their verification.