INTERNET BLIND TEST: MQA Core Decoding vs. Standard Hi-Res (24/88 or 24/96)

Good question liv3evil.

IMO, it should not. At most I added ~2 seconds of fade-in and fade-out so it doesn't end and start too abruptly where appropriate. I believe in one of the tracks, I didn't fade in at all.

Considering I want people to blind test ~2 minutes of each tune, I trust that whatever small change this added is miniscule in comparison. Misalignment in start/stop is likewise minimal such that even without exact matching down to the sample as Testikoff did for delta testing, you should be able to test the A/B variants in ABX Comparator and flip between them instantaneously without much of a noticeable discontinuity.

In any event, the flexibity of one testing these files in the comfort of home is much better than any dealer showroom!

 

Either way Missan, I hope you and others fill out the survey!

Remember folks, I want as many people trying this as possible for the survey database. As I have posted before, like Testikoff above, the objective data already tells us something about how difficult this could/should be.

It's the subjective data around preference that I'm after. That ineffable human intuition of which do you like? Even if consciously we might not clearly know...

 

@Archimago, you and I need to get together with two so-called same masterings and play them here on my system. You won't believe it until you hear it. Others have taken the plunge and walked away going PC me if interested in a real live comparison as opposed to some computer program telling you they are the same.

 

Sure Dave, hey, maybe we can try out the blind test on your system and hear the difference it makes!

 

Sure why not? It's how another Forum member and I discovered the sonic difference/same mastering scenario in the first place.

 

I'm in the process of conducting my blind listening test, but just wanted to relate a strange/humorous occurrence. I downloaded the files, and then transferred them into a subfolder in the folder where I've directed Roon (my preferred listening program) to look for music files. Roon generally treats everything in its own subfolder as an album but was apparently confused by the lack of metadata in these files, so it took a guess. The files showed up in Roon as a 6-song EP by Bebe Rexha (yes, Roon shows the cover), and displayed the 6 tracks with the names of the songs in the EP.

This does not affect my blind test.

 

the same here, with Roon what a mess...
usually I find HD FLAC better than MQA, so I suspect that version B of these songs are FLAC, because I prefer them in all 3 test. My favorite ones are the version B

 

Archimago, thanks for doing this! I have a background in digital signal processing (mainly for video applications) andhad been wondering if something like this could be done also when Tidal came out a few months ago. I agree with your comment on the blog that MQA could have released high resolution versions of music with and without the deblurring process for consumers to listen to and this would have answered questions anyone had. The fact that they did not is telling. The fact that previously they did not run A-B tests as reported in audio shows also says something about their confidence level.

I'll give the test a try this weekend.

 

I preferred A on the first set and B on the last two. The differences were minuscule. I downloaded the files the first day you posted. Did several listens and my choices didn't change.

 

Thanks for the note Ski. Bebe Rexha eh? That's interesting that Roon will assert its "best guess" like this. Considering that I've included tags in the files, it seems presumptuous for Roon to not maintain some semblance of the tags!

 

That could depend on Ski Bum's Roon settings. Roon can use its own tags, your tags, or a combination of the two. It is the same for genres. I learned this when seeing albums listed under genres that were not in my music folder. Clearly, to get the most of out of Roon you need to combine it with something like Tidal, and then this is less of an issue, as you can listen to the files (provided they are part of the music service).

 

I didn't even know who she is (I've since looked it up). I changed the names of the "album" and the tracks in Roon.

I've completed the survey but do not want to affect the blind voting by stating my votes. I hope I preferred the MQA versions or else the coming years could be kind of miserable.

Thanks for doing this.

 

This comment kind of ruins the test. Now expectation bias creeps in.

 

you are correct, I apologize. i hope everyone goes into the test with and open mind and open ears.

 

I really enjoyed this blind test & I'm looking forward to getting the results.

I've also read that MQA outperforms standard CD & equals standard CD if not decoded. Depending on what the results of this blind test show I'd be very interested to partake in further tests (assuming they've not already been done & I've missed them).

I've never heard MQA but one of my friends heard an MQA'd Doors versus a standard Doors CD. He preferred the standard CD but he often prefers newer, more compressed CDs to originals so our sonic tastes aren't the same.

 

Thanks for responses so far everyone...

Just a reminder - slightly less than 2 weeks left in the test!

 

Should I be happier if I can hear the difference, to if I can't hear the difference?

 

Remember that for the survey, although I want people to pick one - either A or B as a preference, you can respond with your level of audibility... From "No real difference" to "Clear difference".

There is not necessarily a "happier" outcome when it comes to audibility . The question really is - does MQA have at least a bias towards sounding "better" in any way! The onus is on the blinded MQA track to reveal with hopefully a good number of testers some kind of preference.

Remember folks, the MQA people have been reluctant to even run A/B demos in audio shows over the years! I bet this is the only time you're ever going to see a "study" of sorts. Stand up and be counted!

 

I'd only be happy with a system where I can hear a difference. I wouldn't want a system where both samples sound indistinguishable from each other.

I did a quick listen with both my main headphone system and my transportable headphone system. I hear a difference with both systems, and I'm very happy with the sound of both systems. I haven't taken the survey yet. I will before the test period expires.

 

I have a silly question: I have 3 separate setups in the house. Can I submit 3 different feedback or do I have to decide with which setup I'm going to use for the blind test and submit only one?

 

I don't know about the OP, but I'd appreciate your results for each setup.

 

I don't think you have any idea how ABX works.

 

I've done a quick ABX of the first pair. The differences, while distinguishable, are not significant and I would be happy with either sample.

While subjective preference between A and B is irrelevant, I prefer A for sounding more "open" and with "better" ambience acoustics. I have't read the details on the source files but if one is the original and the other is the coded version, the MQA codec is clearly not lossless.

As for the MQA sounding "better" than the original - than the original BS - if a codec sounds different than the source it goes directly against the definition of what "HiFi" is all about.

foo_abx 2.0.3 report
foobar2000 v1.3.16
2017-09-02 11:24:39

File A: 01 - Arnesen A.flac
SHA1: 7c05aaac872d4233ad8b231c7f3788e06def36d8
File B: 02 - Arnesen B.flac
SHA1: 4b9282507e9525bc6fca678da0b410db84c3b00c

Used DSPs:
Resampler (SoX) mod2, Resampler (SoX) mod2, Resampler (SoX) mod2

Output:
ASIO : M-Audio USB ASIO
Crossfading: NO

11:24:39 : Test started.
11:26:28 : 01/01
11:27:24 : 01/02
11:28:43 : 02/03
11:30:18 : 03/04
11:31:16 : 04/05
11:31:38 : 05/06
11:32:21 : 06/07
11:32:53 : 07/08
11:32:53 : Test finished.

----------
Total: 7/8
Probability that you were guessing: 3.5%

-- signature --
381bf229668ebe705846a213e7515aa573c2a798

 

I originally planed on testing it using my headphones, the speakers in the study and the speakers in the living room.

Sadly, after spending 20 minutes listening multiple time to each sample through the headphones I got tired and decided that I'm not interested enough in the issue to continue wasting my time. Let's call it "A, B, B almost no difference or too tired to notice" and revoke my audiophile membership card right away to save time to everyone.

I also suspect that no matter whatever the outcome of this test, there will immediately be a counter reaction saying that the MQA samples suffered from processing when flattened to standard FLAC files.

 

no news? c'mon it's 11 september!