Amazon UHD vs Tidal MQA

[Brantome]

I have been doing A/B testing, though not blinded, through my system, which is a Wiim Pro connected over optical (S/PDIF or Toslink) to a Linn Klimax/2 upgraded to the Organik DAC DSM (digital streamer), a D'Agostino Momentum MxV S250 stereo amp, Wilson Audio Sasha W/P floor standing speakers, and a JL Audio Fathom f113v2 powered subwoofer.

The Linn DSM can play streams from Qobuz directly at high resolution, Tidal at CD quality only, and also will take Airplay/2 input directly. These capabilities were used in my testing, though I did not notice any difference when I played Qobuz at hi rez directly on the Linn or when the stream was passed from the Wiim over optical to the Linn.

I compared:
44.1 kHz/16 bit FLAC recordings of my CDs streamed from Minimserver media server on the Linn DSM
44.1 kHz/16 bit Tidal recordings that are not unfolded on Linn DSM
44.1 kHz/16 bit Qobuz recordings over Airplay (maybe downsampled from 96 or 192 kHz) on Linn DSM
96 kHz/24 bit or 192 kHz/24 bit Amazon recordings via Wiim Pro
96 kHz/24 bit or 192 kHz/24 bit Qobuz recordings on Linn DSM
96 kHz/24 bit Tidal one unfolding via Wiim Pro

About 15 years ago, I heard clear differences between an SACD recording of "Come Away with Me" by Nora Jones and the CD. The differences was in the "atmospherics" - high frequency cues of echos in the room, etc. This was also where I heard the biggest differences when comparing audio amplifiers, DACs, and speakers, in A/B testing.

So I sought out tracks that have this characteristic. One album that is good for this is "Paradise Valley" by John Mayer. Both the first track, "Wildfire", and the eighth, also confusingly called "Wildfire", start with atmospherics: people talking in a bar and outdoor sounds of crickets. the tracks on "Come Away with Me", "Don't Know Why" and "Seven Years" also have subtle echos and spacial sounds in the piano, percussion and guitar.

I cranked up the volume to where the loudest tones were about 85 dB, measured with "decibel meter" on my iPhone at my listening location. 85 dB is between food blender and motorcycle in loudness; my wife wasn't home.

I tried to find albums that weren't remixed/remastered between the versions I was comparing, since remixing/mastering can substantially change how an album sounds. For example, "Come Away With Me" was remixed/remastered in 2022, long after I bought my CD. This is noticeable on the third track, where the bass is a little sharper, with higher frequency fret noise more prominent.

My conclusion:

On the John Mayer tracks I get the sense that I am hearing more "air", more of the room, in the higher bit rate recordings, with no substantial difference between Qobuz, Tidal unfolded, and Amazon. However, if I go back and listen again to the 44.1 kHz/16 bit recordings, all the sounds seem to be there as well, so maybe I am just paying more attention when listening to high res.

On the Norah Jones tracks, I can't hear any difference. Even playing the unfolded master 96 kHz/24 bit version vs. the 44.1 kHz/16 bit (CD) version of "Come Away With Me", both are available on Tidal, I hear no difference, no matter the track. The Qobuz and Amazon tracks also sound basically the same. *Maybe* the track "Come Away With Me" the Amazon 96 kHz/24 bit and Qobuz 192 kHz/24 bit has a little more of that sense of the room, but I am not sure I could identify this definitively in a *blind* A/B test.

The new mix of "Come Away with Me" is clearly better, however, and I could spot it blind.

What I do take away from this is I need to play my music louder on a regular basis, if I really want to hear all that is there, and fully engage with the music.

Plus you have the great benefit of the Organik DAC that few will have had the luxury of listening to. Nice to know that a unit like the WiiM Pro at a few percent of its cost can contribute something to it. Maybe you should post this on the Audiophile UK Facebook page too where it might wipe the LOL off the sneering comments from a certain member when the WiiM units are mentioned

 

[Al Lee]

It is not that simple because part of that is felt also by our body not ears. That is why everyone feels/hears music in their own way. And that is why theoreticaly ideal mp3 (even 320kbps) are poor for us.

Note: the reason a "compressed" MP3 file works for this test tone file is because the tones are pure sine tones, exactly one note at a time with no overtones, the easiest sound to compress without loss of fidelity. If you feed the output of this MP3 after decompression into an oscilloscope, all you will see is pure sine waves at each frequency in turn, plus some noise at the transitions.

I agree the body also is sensitive to sound. However, that is predominately for low frequencies, 100 Hz and down, the bass drum one feels in the chest. Even the crappiest compressed MP3 get these right, because there is so little information per second in a 100 Hz or less bass line.

As for higher frequencies, starting somewhere above 100 Hz, it appears that only the ear is capable of generating electrical pulses sensitive to these frequencies that go to the brain. The deaf cannot hear these frequencies via some other method.

There was some evidence back in the early 1990s that ultrasonic sounds (well above 20 kHz), at loud volumes, may be transmitted by the skull to the brain. However, it looks like that research has gone nowhere, indicating it might have been a spurious result.

Experts Find Deaf Can Hear At High Ultrasonic Ranges (Published 1991)

www.nytimes.com

 

[Al Lee]

Plus you have the great benefit of the Organik DAC that few will have had the luxury of listening to. Nice to know that a unit like the WiiM Pro at a few percent of its cost can contribute something to it. Maybe you should post this on the Audiophile UK Facebook page too where it might wipe the LOL off the sneering comments from a certain member when the WiiM units are mentioned

Agreed. As far as I can tell, the Wiim Pro does its digital job well, with EQ off and fixed volume, in taking in losslessly compressed or uncompressed stereo digital streams, and putting them out over the optical link in bit perfect form.

The Wiim Pro even seems to do the MQA unfolding reasonably correctly. At least Tidal streams show up in my Linn DSM as the expected 96 kHz unfolded sample rate, and do not sound worse than the 44.1 kHz/16 bit sources from Tidal.

I will admit I am very fortunate to be able to afford a Linn Klimax Organik DSM. I inherited some money just as I was planning to upgrade my stereo. This is a huge step up from my previous system. It was A/B tests of CD quality content at my local audiophile dealer (Definitive Audio of Seattle) that convinced me Linns were worth the money for the quality of the DACs, and their approach of "source first" to get the best quality is right...

...not that the D'Agostino amp or Wilson Audio speakers are slouches.

This is why I explained above that what one gains from high sample rate recordings is *not* higher audio frequencies in the output, but *perhaps* better performance by your DAC.

Strictly speaking, if one looks at the math Shannon did, he assumes the DAC will be able to sum a very large number of sinc functions (sin(x)/x) of all the adjacent samples in order to reconstruct the audible values *between* the samples. The sinc functions are constructed to be exactly zero at all the sample points except for the one where it matches the sample value. However, at times between the samples, these sinc functions that are further away both add to and subtract from the values of sinc functions on either side of the place where the output value is being determined.

Adding sinc functions is in principle a very hard thing for a DAC to do. Providing more samples per second, so making the time between samples shorter so less interpolation is needed, *could* lead to more accurate reproduction of the original audio signal by the DAC, particularly a cheap DAC.

This was the idea behind the DSD audio format used by SACDs back in the 1990s. Sampling at megahertz rates, even if the samples are only 1 bit in resolution, produced an information stream that was easier for DACs of the day to turn into accurate audio output.

 

[WiimFan]

For people who are curious what maximum frequency they can hear, I recommend playing the following file through a system capable of reproducing 20 kHz frequency audio. The tones are all equally loud (same amplitude) but you will find it gets hard to hear the last 6 seconds, which are tones between 13 kHz and 20 kHz.

Thanks for posting this fun clip. I could hear up to the last 3 seconds.

But I don't think my hearing is the problem, it's my annoying NAD amp cutting off the last 7k before 20k by about 10db, based on the UMIK-1 measurement using the Dirac Live app.

I will try playing the clip later on my AVR which is capable up to 100khz (!) and see if I can hear the missing last 3 seconds.

 

[Al Lee]

Thanks for posting this fun clip. I could hear up to the last 3 seconds.

But I don't think my hearing is the problem, it's my annoying NAD amp cutting off the last 7k before 20k by about 10db, based on the UMIK-1 measurement using the Dirac Live app.

I will try playing the clip later on my AVR which is capable up to 100khz (!) and see if I can hear the missing last 3 seconds.

One has to have a system that is capable of going to 20 kHz, and down to 20 Hz, if one wants to have a chance of hearing all the samples. My system (described above) has tweeters rated as flat to 22 kHz, and the subwoofer definitely goes down to 20 Hz. The amp is rated as basically flat from 20 Hz to 100 kHz. They are just bragging for audio above 20 kHz, since no one can hear it to tell if they are telling the truth.

The fact I can't hear the last 2 or 3 of the 2-second tones is because of my ears, not the equipment. The last 4 tones (8 seconds) are at approximately 10 kHz, 12.5 kHz, 16 kHz, and 20 kHz.

Note: there is a soft click at each tone change. One can count those to know when one runs out of hearing.

 

[Hiphiman]

Agreed. As far as I can tell, the Wiim Pro does its digital job well, with EQ off and fixed volume, in taking in losslessly compressed or uncompressed stereo digital streams, and putting them out over the optical link in bit perfect form.

The Wiim Pro even seems to do the MQA unfolding reasonably correctly. At least Tidal streams show up in my Linn DSM as the expected 96 kHz unfolded sample rate, and do not sound worse than the 44.1 kHz/16 bit sources from Tidal.

I will admit I am very fortunate to be able to afford a Linn Klimax Organik DSM. I inherited some money just as I was planning to upgrade my stereo. This is a huge step up from my previous system. It was A/B tests of CD quality content at my local audiophile dealer (Definitive Audio of Seattle) that convinced me Linns were worth the money for the quality of the DACs, and their approach of "source first" to get the best quality is right...

...not that the D'Agostino amp or Wilson Audio speakers are slouches.

This is why I explained above that what one gains from high sample rate recordings is *not* higher audio frequencies in the output, but *perhaps* better performance by your DAC.

Strictly speaking, if one looks at the math Shannon did, he assumes the DAC will be able to sum a very large number of sinc functions (sin(x)/x) of all the adjacent samples in order to reconstruct the audible values *between* the samples. The sinc functions are constructed to be exactly zero at all the sample points except for the one where it matches the sample value. However, at times between the samples, these sinc functions that are further away both add to and subtract from the values of sinc functions on either side of the place where the output value is being determined.

Adding sinc functions is in principle a very hard thing for a DAC to do. Providing more samples per second, so making the time between samples shorter so less interpolation is needed, *could* lead to more accurate reproduction of the original audio signal by the DAC, particularly a cheap DAC.

This was the idea behind the DSD audio format used by SACDs back in the 1990s. Sampling at megahertz rates, even if the samples are only 1 bit in resolution, produced an information stream that was easier for DACs of the day to turn into accurate audio output.

I have been doing A/B testing, though not blinded, through my system, which is a Wiim Pro connected over optical (S/PDIF or Toslink) to a Linn Klimax/2 upgraded to the Organik DAC DSM (digital streamer), a D'Agostino Momentum MxV S250 stereo amp, Wilson Audio Sasha W/P floor standing speakers, and a JL Audio Fathom f113v2 powered subwoofer.

The Linn DSM can play streams from Qobuz directly at high resolution, Tidal at CD quality only, and also will take Airplay/2 input directly. These capabilities were used in my testing, though I did not notice any difference when I played Qobuz at hi rez directly on the Linn or when the stream was passed from the Wiim over optical to the Linn.

I compared:
44.1 kHz/16 bit FLAC recordings of my CDs streamed from Minimserver media server on the Linn DSM
44.1 kHz/16 bit Tidal recordings that are not unfolded on Linn DSM
44.1 kHz/16 bit Qobuz recordings over Airplay (maybe downsampled from 96 or 192 kHz) on Linn DSM
96 kHz/24 bit or 192 kHz/24 bit Amazon recordings via Wiim Pro
96 kHz/24 bit or 192 kHz/24 bit Qobuz recordings on Linn DSM
96 kHz/24 bit Tidal one unfolding via Wiim Pro

About 15 years ago, I heard clear differences between an SACD recording of "Come Away with Me" by Nora Jones and the CD. The differences was in the "atmospherics" - high frequency cues of echos in the room, etc. This was also where I heard the biggest differences when comparing audio amplifiers, DACs, and speakers, in A/B testing.

So I sought out tracks that have this characteristic. One album that is good for this is "Paradise Valley" by John Mayer. Both the first track, "Wildfire", and the eighth, also confusingly called "Wildfire", start with atmospherics: people talking in a bar and outdoor sounds of crickets. the tracks on "Come Away with Me", "Don't Know Why" and "Seven Years" also have subtle echos and spacial sounds in the piano, percussion and guitar.

I cranked up the volume to where the loudest tones were about 85 dB, measured with "decibel meter" on my iPhone at my listening location. 85 dB is between food blender and motorcycle in loudness; my wife wasn't home.

I tried to find albums that weren't remixed/remastered between the versions I was comparing, since remixing/mastering can substantially change how an album sounds. For example, "Come Away With Me" was remixed/remastered in 2022, long after I bought my CD. This is noticeable on the third track, where the bass is a little sharper, with higher frequency fret noise more prominent.

My conclusion:

On the John Mayer tracks I get the sense that I am hearing more "air", more of the room, in the higher bit rate recordings, with no substantial difference between Qobuz, Tidal unfolded, and Amazon. However, if I go back and listen again to the 44.1 kHz/16 bit recordings, all the sounds seem to be there as well, so maybe I am just paying more attention when listening to high res.

On the Norah Jones tracks, I can't hear any difference. Even playing the unfolded master 96 kHz/24 bit version vs. the 44.1 kHz/16 bit (CD) version of "Come Away With Me", both are available on Tidal, I hear no difference, no matter the track. The Qobuz and Amazon tracks also sound basically the same. *Maybe* the track "Come Away With Me" the Amazon 96 kHz/24 bit and Qobuz 192 kHz/24 bit has a little more of that sense of the room, but I am not sure I could identify this definitively in a *blind* A/B test.

The new mix of "Come Away with Me" is clearly better, however, and I could spot it blind.

What I do take away from this is I need to play my music louder on a regular basis, if I really want to hear all that is there, and fully engage with the music.

This is a really interesting posting, not least because it come across as a very honest account using equipment which should, if any kit can, be able to highlight differences between the different variables tested. As mentioned elsewhere, I too think the most notable observation is that - in this system anyhow - using the WiiM as the streaming source had no discernible influence on the SQ compared to the Linn equivalent, which is what logic would suggest but is almost never highlighted, doubtless for obvious reasons, by media of all sorts.
I feel this topic could easily spawn several, hopefully independent, PhDs where a rigorous approach to blind testing could enlighten us all in what really matters when it comes to sound reproduction. In this way we could spend our money on the components which really matter, aided by savings from those (like streamers it seems) that don’t.
Not so long ago I remember reading articles from apparently authoritative sources waxing lyrical about the differences between costly HDMI cables. Until overwhelming evidence made them stop.

 

[Al Lee]

So I did another test of bit rate/bit depth differences: "Revolver" by the Beatles, in particular "Taxman," the first track.

Samples:
FLAC 44.1 kHz/16 bit of CD from 1990s (1987 US Parlophone release/mastering)
Tidal 44.1 kHz/16 bit; 2009 remaster
Tidal 44.1 kHz/16 bit; 2022 remaster
Tidal 96/24 bit one unfolding of Master (48 kHz/24 bit) in Wiim; 2022 remaster
Qobuz 44.1 kHz/16 bit; 2009 remaster
Qobuz 44.1 kHz/16 bit; 2022 remaster
Qobuz 96 kHz/24 bit; 2022 remaster
Amazon 96 kHz/24 bit; 2009 remaster
Amazon 96 kHz/24 bit; 2022 remaster

Biggest difference I hear? The cymbals that come in at ~10 seconds are clearly muted on the original CD. The cymbals are also a little less clear (and less centered) on the 2009 remastering compared to 2022.

As for the high bit rate Qobuz, Amazon and Tidal's MQA/Master (first unfolding) vs. "only" CD quality recordings, I cannot hear a difference between all the versions of the 2022 remastering.

I also cannot hear a difference for the recordings I either stream directly on my Linn DSM vs. via optical input from the Wiim Pro. This is good, since Wiim Pro output over SPDIF/optical should be bit identical with what the Linn receives over ethernet.

Having a subscription to a streaming service in order to listen to the 2022 remastered album is worthwhile. Whether it is CD quality or better doesn't matter, at least for my equipment and ears.