[NeoG]

  is this post sarcastic?  MQA discards several bits of the original signal to replace them with sample data. so bit perfect...

and MQA relies on DSP.

it just doesn't enjoy other DSPs coming before the extraction of the stuff it has coded inside the PCM signal.

 
Not at all.. The whole idea of is it lossy or lossless is being put through a washing machine. If you say MQA is lossy compared to a 192/24 stream, then literally everything can be lossy compared to whatever you choose as a reference. 192/24 is lossy compared to DXD for example.
 
The fact that MQA has a non-rectangular capture window doesn't make it lossy in the traditional sense. It just means they've optimised the capture window to what they believe is important to record in the first place. It's the same argument for distributing in 44/16, everything that went in will come out as intended.
 
Now, I don't think anyone has any inside information currently on the fold and unfold transform - this may be done as a data based operation (Like with Dolby technologies) or it may be an audio-space transformation. It's entirely possible the low level high frequency information is a data-difference stream embedded in the 44/16 audio, and the MQA decoder is able to fully restore the original triangular window without doing using a break-point notch.
 
If that is the case, the system is lossless as per it's original intention, and DSP is not involved in the reconstruction. But I digress, it's entirely circumstantial.

 

[gregorio]

  The whole idea of is it lossy or lossless is being put through a washing machine. If you say MQA is lossy compared to a 192/24 stream, then literally everything can be lossy compared to whatever you choose as a reference.

 
That would be true, if we were to choose whatever we want as a reference but of course that's not what we do! The reference of what we get out of a codec, is what we put in. If we put in a 192/24 stream but get out less than a 192/24 stream after encoding/decoding, then the encoding/decoding process is lossy. If we put in 16/44.1 and get out less that 16/44.1 then the codec is lossy. If we get out exactly what we put in, then it's lossless.
 
How audible any loss maybe, is entirely another question of course.
 
G

 

[sonitus mirus]

   
That would be true, if we were to choose whatever we want as a reference but of course that's not what we do! The reference of what we get out of a codec, is what we put in. If we put in a 192/24 stream but get out less than a 192/24 stream after encoding/decoding, then the encoding/decoding process is lossy. If we put in 16/44.1 and get out less that 16/44.1 then the codec is lossy. If we get out exactly what we put in, then it's lossless.
 
How audible any loss maybe, is entirely another question of course.
 
G

 
Yep, MQA is like MP3 with Replay Gain, except one format is open source and can play over a wide range of existing products without the requirement of proprietary software or hardware.  Ok, there are other differences, I'm mostly joking. 

 

 

[NeoG]

   
That would be true, if we were to choose whatever we want as a reference but of course that's not what we do!

 
Well, arguably you are still choosing an arbitrary reference point - I suppose I see the 192/24 as an intermediary in what is likely a 2-3 step data reduction process from whatever the source is. So it seems weird to me to single out that particular piece of the process as the point of comparison to the unfolded output...

 

[gregorio]

   
Well, arguably you are still choosing an arbitrary reference point - I suppose I see the 192/24 as an intermediary in what is likely a 2-3 step data reduction process from whatever the source is. So it seems weird to me to single out that particular piece of the process as the point of comparison to the unfolded output...

 
It's not an arbitrary reference point. All digital audio formats (raw, compressed lossy or lossless) can only know, store and restore what they are given. There is no way for a digital audio format to know, store or restore data which was contained in any other version or format prior to the version it's being fed. It would be nonsensical to try and reference an impossibility, the only logical reference point for an encoder/decoder is what you give it to encode in the first place!
 
BTW, it's extremely unlikely that a 192/24 file is the result of even one intermediary data reduction step, let alone two or three. The only time a 192/24 file is the result of a data reduction is when mixing/mastering a SACD and even then, it's difficult to argue logically that the conversion from DSD to 192/24 constitutes a data reduction. The other time could be if you're referring to the mix environment but although this is a master source in one sense, it only ever exists 'virtually", it's never recorded as a file and it's never played or heard by anyone (including the artists/engineers who created it), because it can't be played!
 
G

 

[NeoG]

   
It's not an arbitrary reference point. All digital audio formats (raw, compressed lossy or lossless) can only know, store and restore what they are given. There is no way for a digital audio format to know, store or restore data which was contained in any other version or format prior to the version it's being fed. It would be nonsensical to try and reference an impossibility, the only logical reference point for an encoder/decoder is what you give it to encode in the first place!

 
I guess the point I'm trying to make is, why is MQA all of a sudden being thrown in the lossy bin when 44/16 distribution has been doing the same thing in a different form for years from 96/24 or 192/24 masters?
 
What is it about decoded MQA that is worse than mastering down to 44/16 for it to be labelled as such?

 

[castleofargh]

  I guess the point I'm trying to make is, why is MQA all of a sudden being thrown in the lossy bin when 44/16 distribution has been doing the same thing in a different form for years from 96/24 or 192/24 masters?
 
What is it about decoded MQA that is worse than mastering down to 44/16 for it to be labelled as such?

https://en.wikipedia.org/wiki/Lossless_compression
we're not talking about final resolution, downsampling, filtering, bit reduction...  we're talking about lossless format.
take the 16/44 PCM, make a flac out of it, decompress the flac and you can go back to the same identical 16/44 PCM.
try doing that with MQA. if you think 16/44 is a bad starting point for MQA  you can pick 24/192, or even let's be super nice and say 15/192 or 17/192 PCM. if the decompression uses dither, it's lossy decoding. the end.
 
how some people assume that lossy is a pejorative word,  now that's a very different matter but I don't feel the need to be dragged down to their level of interpretation. yes a lossy format can have very high resolution and yes a lossy format can sound transparent.

 

[gregorio]

   
[1] I guess the point I'm trying to make is, why is MQA all of a sudden being thrown in the lossy bin when 44/16 distribution has been doing the same thing in a different form for years from 96/24 or 192/24 masters?
 
[2] What is it about decoded MQA that is worse than mastering down to 44/16 for it to be labelled as such?

 
Huh, you do realise that makes no sense? It's also bizarre that someone has given you a "reputation" for not making any sense!
 
1. 44/16 is not a format, it's a resolution. So you're trying to compare a resolution to a format, which makes no sense! If you want to logically compare MQA to something you have to compare it to another format, say FLAC, WAV or ALAC. If you don't reference a format/codec to what you feed that codec then any comparison is nonsense, one could for example easily demonstrate that a 128kbps MP3 is better than DSD Wide.
 
2. Let's compare apples to apples shall we? What's the difference between 44/16 in say FLAC format and 44/16 in MQA format? Firstly, the FLAC format provides a bit perfect copy, MQA does not AND Secondly, the FLAC format is free, MQA has to be licensed. If you don't like 44/16 for some reason and prefer 192/24, no problem, let's ask the same question again: What's the difference between a 192/24 in say FLAC format and 192/24 in MQA format? Answer is the same!
 
G

 

[wormcycle]

Lots of  posts here saying "well Spotify is lossy, Tidal hifi is FLAC therefore it must be a difference and I hear the difference". And it is like "day and night". Yeah.....
 
If you listen mostly using good amp, good DAC and resolving  headphones then listen for yourself and decide if you hear the difference. On some tracks you will on some you will not. I went for days A/Bing between the Spotify and Tidal classical tracks and decided on Tidal on the a very slim margin. But if I used mostly my phone to stream music I would not even think about Tidal. 

 

[nrvenice]

I volume-matched Spotify Extreme (their app has separate volume control) and Tidal HiFi on my macbook through LCD-2f. I played my go to test tracks (track 1 of https://www.amazon.com/Montserrat-Figueras-Emotion-Jordi-Savall/dp/B006MZS932 and track 2 of https://www.amazon.com/Tous-Matins-Monde-Bande-Originale/dp/B00005S0MD). I'm very familiar with the 2nd track and have heard it live in 3 different concerts over the years. I had no difficulty picking out what was playing and thought Tidal was far superior.

Tidal have since added Masters for some albums for their desktop app but I have not tested this.

 

[Niouke]

I volume-matched Spotify Extreme (their app has separate volume control) and Tidal HiFi on my macbook through LCD-2f. I played my go to test tracks (track 1 of https://www.amazon.com/Montserrat-Figueras-Emotion-Jordi-Savall/dp/B006MZS932 and track 2 of https://www.amazon.com/Tous-Matins-Monde-Bande-Originale/dp/B00005S0MD). I'm very familiar with the 2nd track and have heard it live in 3 different concerts over the years. I had no difficulty picking out what was playing and thought Tidal was far superior.

Tidal have since added Masters for some albums for their desktop app but I have not tested this.

Was it blind testing, or did you switch between apps yourself? Removing the psychological aspect is very important to blind testing. Also Spotify is a bit annoying because the app doesn't say if you are listening to 160kbps or 320kbps, and 160kbps is indeed detectable. maybe you should add other tracks to your comparison?

 

[frodeni]

I hope MQA dies. There is supposed to be clean 16/44.1 stream, if you do not use it, but that is simply not the case. The MQA reproduction is far worse than the 16/44.1, for all the tracks I have tried. That something sounds different, does not mean that it is an improvement. Both the supposedly clean 16/44.1 and the MQA stream is lossy, and if you read the papers closely, that is actually stated in there. Sure, the sound is different, but it is no improvement.

 

[nrvenice]

Was it blind testing, or did you switch between apps yourself? Removing the psychological aspect is very important to blind testing. Also Spotify is a bit annoying because the app doesn't say if you are listening to 160kbps or 320kbps, and 160kbps is indeed detectable. maybe you should add other tracks to your comparison?

I agree re blind testing but I couldn't do it easily. I did try to let the Macbook "choose" which app to play by using the keyboard play button but that was dependent on which app was on "top" of the other most recently. I also tried listening to popular music tracks and even here could discern a difference though not as marked.

I'll probably end up trying to ABX this in the future to confirm but I've participated in many double-blind tests over 20 years (wine, vodka, tequila, scotch, amplifiers, speakers) and have observed that I'm pretty objective when ABing.

The cost difference was nominal and I want to support their support of higher bitrates and artists (I always feel like I should send my favorites a check since they make such little money from my listening).

 

[bigshot]

The problem with comparing streaming services is it's hard to be sure you're listening to the same mastering. Add to that the likelihood that the levels aren't matched and it can be tough. It takes a lot of work to do a solid A/B test even if it isn't double blind. To do this properly, I'd say that you would need to...

1) Capture the stream in high quality so you can line them up for direct comparison
2) Level match the two samples
3) Try to null the two samples to determine if the mix is the same
4) Set up a true double blind direct switchable comparison

Obviously, this would take the better part of a weekend to do. It's probably not worth the effort. But when you are presented with two nearly identical sounding samples, this is how you would go about figuring out if there is a difference between them or if you are imagining a preference. If the difference was clear, you wouldn't need to be so stringent in setting up your test.

Personally, if it requires that much effort to even determine if there is a difference, I would just decide that it doesn't matter at all. If I can't perceive a clear difference, I'll take either.

 

[nrvenice]

The problem with comparing streaming services is it's hard to be sure you're listening to the same mastering. Add to that the likelihood that the levels aren't matched and it can be tough. It takes a lot of work to do a solid A/B test even if it isn't double blind. To do this properly, I'd say that you would need to...

1) Capture the stream in high quality so you can line them up for direct comparison
2) Level match the two samples
3) Try to null the two samples to determine if the mix is the same
4) Set up a true double blind direct switchable comparison

Obviously, this would take the better part of a weekend to do. It's probably not worth the effort. But when you are presented with two nearly identical sounding samples, this is how you would go about figuring out if there is a difference between them or if you are imagining a preference. If the difference was clear, you wouldn't need to be so stringent in setting up your test.

Personally, if it requires that much effort to even determine if there is a difference, I would just decide that it doesn't matter at all. If I can't perceive a clear difference, I'll take either.

Thanks for outlining the methodology. I looked into Foobar but it doesn't run on mac and didn't immediately find other software to use. I would put some faith in someone else's double blind if the participants were classically trained musicians listening to works they were very familiar with using highly resolving headphones or speakers that they'd used for more than 3 months but search (including sound science) hasn't yielded any.

I trust Linkwitz's methodology and he uses lossless for his test cd (http://www.linkwitzlab.com/sound_picture_cd.htm) and personal recordings and makes some interesting comments on sound quality (http://www.linkwitzlab.com/ALMA'14/Sound_quality.htm). Aczel observes (http://www.biline.ca/audio_critic/audio_critic_web1.htm#red) that higher resolution recordings signal a greater care in recording process and I have made similar observations, hence my continued support of Tidal hoping to signal a consumer demand for better product.

Tl;dr Aczel: "Now, Meyer and Moran are careful to point out that the new hi-rez formats generally sound better than standard CDs, but not because the processing technology is superior. The hi-rez discs are aimed at a more sophisticated market, and therefore the recording sessions and production techniques tend to be more sophisticated, more puristic, in terms of microphoning, compression, editing, etc."