Jump to content
Computer Audiophile
The Computer Audiophile

Article: MQA (for civilians)

Rate this topic

Recommended Posts

Thanks. I personally wonder whether there is much actual content above 48KHz on hi res recordings;

 

There isn't, but that's no excuse for MQA to lie about what their product does.

Share this post


Link to post
Share on other sites
Consider a 48 kHz 24-bit MQA file. The high 15 bits of this file are a reasonable representation of the 0-24 kHz frequency band from the original while the low 8 bits contain a compressed version of the 24-48 kHz band. The "core" decoder decodes the low 8 bits and combines this information with the base band signal from the high bits to produce a 96 kHz sample rate output covering the 0-48 kHz frequency band. Compared to the original file, this decoded version comes pretty close, but there some losses in the high half (24-48 kHz). If the original file had a sample rate higher than 96 kHz, any frequency content above 48 kHz will have been completely discarded. The "render" part upsamples the output of the "core" decoder to the same rate as the original file. Because of the terrible interpolation filter, it may at first glance look like some high-frequency content has been restored here, but it is all fake. All you actually get is images of the lower frequencies and a rising level of dither noise.

 

Thanks, that's a very impressive explanation. You reference the "original file" which I assume is a hi-res file, available via download or in limited selection albums at about $25 a pop. Yes? To my knowledge, these large hi-res files are not being commercially streamed, correct? You must buy the physical CD or download via online purchase. These make up a certain class of audiophiles hi-res libraries. Yes?

 

You write, "Because of the terrible interpolation filter (I wouldn't know a "terrible" IF, if I saw one), it may at first glance look like some high-frequency content has been restored here, but it is all fake. All you actually get is images of the lower frequencies and a rising level of dither noise." Since I'm a simple guy and have yet to master listening to music with my eyes via an oscilloscope, rather than my ears, I'll take your word for it.

 

What's puzzling to many of us is why MQA'd files, streamed via TIDAL, sound aesthetically more pleasing (more spacious/3D, tonally richer, and easeful) than non-MQA's versions. I've done firsthand A/B tests, making appropriate volume adjustments, and I don't believe my ears are playing tricks on me. I make no claim for my eyes. At the end of the day, I'll spend dollars on MQA streaming vs. any other delivery. I guess that doesn't qualify me for the "audiophile" class of listener. :/

 

I hold no opinion regarding the SQ of hi-res albums vs. the same album MQA'd--purchased or streamed. It's really at mute issue for many of the streaming crowd.

Share this post


Link to post
Share on other sites
What's puzzling to many of us is why MQA'd files, streamed via TIDAL, sound aesthetically more pleasing (more spacious/3D, tonally richer, and easeful) than non-MQA's versions. I've done firsthand A/B tests, making appropriate volume adjustments, and I don't believe my ears are playing tricks on me. I make no claim for my eyes.

 

At least some of the MQA files are made from different masters than the non-MQA ones.

Share this post


Link to post
Share on other sites
At least some of the MQA files are made from different masters than the non-MQA ones.

 

As a TIDAL premium subscriber, I'm sure glad they have the grunts searching the archives for all the better master files. Amazing how many they seem to be finding.

Share this post


Link to post
Share on other sites
I think we would all agree that lossy compression "loses" something and as such something is missing. ...

 

 

The real question is whether you can hear the loss. Cables lose something, tubes, transistors, capacitors, etc.

Share this post


Link to post
Share on other sites

I admit that I am just now getting around to reading your article Chris.

 

I am less concerned about MQA as a market/industry standard not because of anything specific you said, but because of the shear scale of the convoluted mess of it all! ;)

Share this post


Link to post
Share on other sites

I lived through almost two years (oh my, maybe more) of online forum blather and gazillions of alternative tests when the big digital camera companies came up with lossless compression. What's different here is that it's not just "is it lossless" (OMG, you can't believe the insane things people did to test for that... most of which illustrated the proponent's lack of understanding of digital imaging) but MQA is claiming "and even better." So far, I'm OK saying it may be a much better compression algorithm. (For all the breakthroughs in genomics, we're still learning how to get better at assembling gene sequences...) Whether the additional information in MQA encoding adds to the sonic quality... that's an area where I have no experience in other digital domains. Other than in photography, where sharpening algorithms were initially embraced and eventually abhorred.

Share this post


Link to post
Share on other sites

Let me talk about an abstract "lossless" and "lossy" (don't touching any certain format).

 

If mean here "and even better" as "and even sound better", "lossless" may "sound worse" than "lossy".

 

Example: Sound engineers know tricks how to edit file for "sound better" subjectivelly. We can edit WAV1, save as WAV2. These files will not binary identical. I.e. WAV2 is "lossy". But WAV2 may sound better.

Share this post


Link to post
Share on other sites
When I think about lossy, I ask myself what is lost. With MP3 and AAC real music is lost. With MQA I don't believe real music is lost. MQA changes digital, making existing terminology require more discussion than in the past.

 

Chris, that was what I was looking for. But when I compared a 2L 24/192 file with MQA of the same track I felt a significant difference (already commented here) of real music left, specifically an hardening of a soprano voice on MQA, in a way that also occurs on the CD track on Tidal...It was clear to me that in my situation (no MQA DAC) I gained nothing on MQA and I lost a very important feature of the "original" file, which is a natural presentation of a humans voice.

 

I wonder when can I stream the MQA to be able to narrow down the variables...

Share this post


Link to post
Share on other sites

When I was a subscriber to Tidal before MQA, it was relatively easy to compare a ripped Redbook album to a streamed version of the same. From this comparison it was very easy to discard the Tidal offerings and cancel the subscription.

 

The only difference an MQA file would make, as mansr pointed out, the source files are different (better to begin with) than the trash that's being streamed at 44.1kHz. No thanks, adios Tidal.

 

For music discovery, the lossy format sites will be fine, Spotify, Apple, et al or even Amazon to sample music. For serious listening... spinning discs or file playback from a local machine. If the Internet shuts down, you still pay for the streaming access and no music. Poor idea that one and the consumer misses out again.

Share this post


Link to post
Share on other sites
That is exactly what I'm saying.

 

That's not my understanding of how undersampling a signal works. Surely its not the case that you lose all the information, but the opposite:

 

- higher frequencies are aliased down into lower frequencies (the folding)

- on reconstruction the higher frequencies are indeed restored (unfolding, you claim this cannot or does not happen), but the problem is now all the aliasing left behind. MQA take advantage of the "known shape" of the audio data to apply some EQ to hopefully remove some of the aliasing, or at least reduce its level.

 

- BTW My understanding is based on this article - MQA 192k / 96k There and back again.- if its wrong please help us understand why.

Share this post


Link to post
Share on other sites
That's not my understanding of how undersampling a signal works. Surely its not the case that you lose all the information, but the opposite:

 

- higher frequencies are aliased down into lower frequencies (the folding)

- on reconstruction the higher frequencies are indeed restored (unfolding, you claim this cannot or does not happen), but the problem is now all the aliasing left behind. MQA take advantage of the "known shape" of the audio data to apply some EQ to hopefully remove some of the aliasing, or at least reduce its level.

 

- BTW My understanding is based on this article - MQA 192k / 96k There and back again.- if its wrong please help us understand why.

If a signal has been undersampled, there is no way of separating aliases from actual baseband components. When such a signal is upsampled using a leaky filter, the result is an unchanged baseband including the aliases from undersampling along with a high band composed of images of the undersampled signal, i.e. the original baseband plus aliases. Sure, the alias components will recreate the original high frequencies, but they'll still remain in the baseband, and they'll be accompanied by unwanted images of the low frequencies. In other words, your "reconstructed" signal is doubly corrupted, and there's nothing you can do about it. Selectively removing the aliases and images is as impossible as unscrambling an egg.

Share this post


Link to post
Share on other sites
It's great when so called "objective" analysis correlates with listening impressions. That just doesn't seem to be the case here though. I've read very few, if any, negative impressions of the actual sound of MQA.

 

Me seems, subjective (and objective) sound qualuty here depend more on DAC implementation, than MQA vs. uncompressed high resolution.

Share this post


Link to post
Share on other sites
Me seems, subjective (and objective) sound qualuty here depend more on DAC implementation, than MQA vs. uncompressed high resolution.

 

Yes, but for those people who don't wish to buy a new DAC (or who aren't using their DAC's filtering, but use software and have a particular filter or filters they like), it is worth considering which of the two might potentially provide better sound for one given implementation.

Share this post


Link to post
Share on other sites
Yes, but for those people who don't wish to buy a new DAC (or who aren't using their DAC's filtering, but use software and have a particular filter or filters they like), it is worth considering which of the two might potentially provide better sound for one given implementation.

 

Decoded MQA is usual PCM. Thus I see there 2 ways:

 

1. MQA Decoder > PCM > Analog

2. PCM > Analog

 

So, theoretically, there should not be difference in frame one given implementation.

Except [PCM > Analog] area that may be released different ways.

As example, different oversampling filters, intermediate sample rate, low frequency filter.

Share this post


Link to post
Share on other sites
Decoded MQA is usual PCM. Thus I see there 2 ways:

 

1. MQA Decoder > PCM > Analog

2. PCM > Analog

 

So, theoretically, there should not be difference in frame one given implementation.

Except [PCM > Analog] area that may be released different ways.

As example, different oversampling filters, intermediate sample rate, low frequency filter.

 

But doesn't this (not having a difference between 1 and 2), *if using the same DAC for both*, depend on the MQA decoder creating PCM in 1 that does not differ audibly from the PCM in 2?

 

Because in my listening an MQA file created from a hi res file results in a different sound than the hi res file, when everything else is the same.

 

 

Sent from my iPhone using Computer Audiophile

Share this post


Link to post
Share on other sites
But doesn't this (not having a difference between 1 and 2), *if using the same DAC for both*, depend on the MQA decoder creating PCM in 1 that does not differ audibly from the PCM in 2?

 

Because in my listening an MQA file created from a hi res file results in a different sound than the hi res file, when everything else is the same.

 

Difference may be in filter into decoder. It is my hypothesis by all previous MQA discussions.

But more probably, difference may be in small (may be inaudible or almost inaudible) difference of levels - with decoder/without.

 

For proper comparing need technically minimize the level difference before [PCM > Analog] stage.

 

For it need calculate standard deviation between identical by time samples decoded MQA and original PCM. And correct level of original PCM or decoded MQA on calculated value (I suppose with precision 0.1 dB or better).

 

After it need perform careful double blind test:

 

- tens and more musical samples in different genres,

- hundreds or more measurements,

- tens or more participants with different ear skills,

- proper hidden switch samples/DAC modes,

- single sitting place (not rows of several listeners simultaneously in different places of speaker's apperture, etc.),

- momental switch played back samples,

- participants have only 3 buttons "better than previous track" and "worse than previous track" and "no difference",

- participants should be in comfortable condition (physical and emotional),

- etc.

Share this post


Link to post
Share on other sites

I see things in simple terms. When we discuss the subjective sound of MQA files that is simply a persons opinion on what he heard and is in no way a factual representation of any reality. There are many different factors that could be influencing what the listener hears.

MQA is a very smart compression scheme to reduce file sizes, no more.

 

The dang light people are making a big deal over is IMO nothing better than the Stereo light on a FM tuner. It tells the user that the file is detected as a MQA processed file, and is locked in and being decoded. That's nice to know.

 

Undecoded MQA is a lossy representation of the original file. We can debate whether the lossy part is audible till the cows come home, but lossy is lossy, period. It can in no way be better or as good as a lossless stream of the original. (unless it's been remastered, then all bets are off)

 

No matter your opinion on MQA, the sad part of all this is that over time the new distribution of lossless original data will disappear. All that will be left is what can be gleamed from the old CDs, SACD's, old downloaded HDA flac files, etc. For those of us that want "the real thing" the only way to obtain them will be the used market and no monies will be distributed to the artists from those acquisitions.

Share this post


Link to post
Share on other sites
I see things in simple terms. When we discuss the subjective sound of MQA files that is simply a persons opinion on what he heard and is in no way a factual representation of any reality. There are many different factors that could be influencing what the listener hears.

MQA is a very smart compression scheme to reduce file sizes, no more.

 

MQA is a compression scheme. I disagree about it being smart.

Share this post


Link to post
Share on other sites
No matter your opinion on MQA, the sad part of all this is that over time the new distribution of lossless original data will disappear. All that will be left is what can be gleamed from the old CDs, SACD's, old downloaded HDA flac files, etc. For those of us that want "the real thing" the only way to obtain them will be the used market and no monies will be distributed to the artists from those acquisitions.

 

I used to love going into the city to browse in the used record stores. The artists didn't get paid then either. If anyone complained about royalties back then, I didn't hear about it. Used records were cheap in the 70's too. I don't think a used record store could afford to pay rent these days. But I miss buying physical media in stores. Shopping was part of the experience for me.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

  • Superphonica
×