Friday, 6 February 2015

MEASUREMENTS: Bob Dylan's "Shadows In The Night" - when 24-bit HRA isn't! (Qobuz)

Of shadows... and hot air...
A reader gave me a tip about the new Shadows In The Night album from Bob Dylan. The allegation is that the 24-bit high-resolution downloads of this album are in fact NOT true 24-bits as claimed!

To start, here's a video to show what the inversion-null should look like with a true 24-bit audio sample:



Now, consider what happens when we use a 24-bit track from Qobuz versus the same track ripped off the 16-bit CD:


As you can see, there is essentially a complete null using the 16-bit CD version on the 24-bit file when the amplitude is boosted by 0.1dB in 24/32 bits and time aligned! Upsampling like from 44kHz to 96kHz is usually easy to spot but this is one of the first times that a "fake" 24-bit file is so easily spotted! Shame on Sony/Columbia/Qobuz for doing this...

If we look at the DRDatabase, one sees that the DR values for the LP appear very similar to the digital editions. It's quite possible that the LP was pressed with the same mastering as well.

16/44 CD Dynamic Range log.

As you can see from the DR log for the CD above, the results are a bit of a mixed bag. DR11 is good! (Maybe we're getting somewhere with new releases not being so badly dynamically crushed?) But what's the deal with the peak volume of only -6.45dB? We've basically wasted 1-bit's worth of dynamic resolution which could have just been optimized with a simple normalization step! Instead of being unnecessarily loud, this CD is unnecessarily quiet! In effect, the maximum bit-depth resolution has been reduced to 15-bits; which in this case also means the 24-bit version is no better!

I guess if I bought the 24-bit version, I'd be wanting my money back. Note that the 24-bit sample I was sent to analyse was from Qobuz, it would be interesting to know if the HDTracks version is also the same. Maybe others can double check if what I'm reporting here holds for all the tracks on that album.

This kind of thing really cannot be tolerated as it's basically a cash grab for zero benefit. There's a 30% price premium on the 24-bit file on Qobuz! You really have to wonder in this digital download model, who is responsible for quality control - especially for a high profile artist like Dylan?

Bottom line: Buy the Bob Dylan CD if you want the album, but do not bother with the 24-bit high-resolution download until there's clarification that either what I'm seeing is wrong or updated 24-bit files have been uploaded. Also, I wish companies like HDTracks and Qobuz would open up a review/comment system like Amazon for people to share information about the quality of the files since the issue of provenance and concerns such as this (and Beck's Morning Phase last year) do arise.


Sony, so does this Bob Dylan album deserve to be called "Every bit a master."? I think it would be precious if we start seeing this album cover on one of those ads promoting HRA instead of Bennett/Ga Ga! :-)

----------------

BTW: On a positive note, as an avid 80's music guy, I like looking out for compilations of stuff I might have missed... Recently I took a chance on the Blank & Jones' "So80s Presents Alphaville" 2 CD set since it had some mixes I had not heard/seen before... WOW! I was impressed. Sounds great and the dynamics were excellent (DR11 both disks). Considering I had been disappointed by previous Blank & Jones compilations, I was pleased by the significant improvement! Keep up the good work, guys...

9 comments:

  1. This comment has been removed by the author.

    ReplyDelete
  2. Your friend should send an email to Qobuz, they will answer or at least see in to that problem.
    From what I'm told they work with what the sources give to them.

    ReplyDelete
    Replies
    1. Yes. I believe an E-mail has been sent...

      Delete
  3. Of course there is a way that any supplier could ensure that this test result is never repeated again, if you get my drift. It would involve 'modifications' being applied to *all* high resolution files before they are released into the wild.

    ReplyDelete
    Replies
    1. Yeah... It would be easy to hide any track and prevent detection by messing up the last 8 bits in all kinds of ways. It's already hard to detect this kind of thing happening so to find it with the Dylan album is surprising!

      As a customer, I would hope there's some *honesty* in the "system". Otherwise, how can we ever trust anything under "high resolution" banner!?

      Delete
  4. This comes down to one thing no download vendor or stream suplier do any quality control at all .They implement everything on the cheap ,building somekind of autamated upload service for the labels to use or abuse . No human hand involved anywhere .

    HRA or any Hifi labeled service should work the other way around the vendor Qubuz HD-tracks ,who ever should actively cooperate with the labels (and possibly artist ) have meetings to evaluate the avaible masters and do listening sessions . And then make thier pick !

    I would gladly be separated from 25$ in a jiffy if I knew i got the definitive version of anything . Then if it came as 16/44.1 or 24/96 is not an issue realy ,but if marketting demands 24/96 so be it .

    Regards /mnyb

    ReplyDelete
    Replies
    1. Sad... And sickening...

      I was very disappointed recently when I heard that the NativeDSD.com download of the Fonè "30th Anniversary" (Natural Jazz Recordings) was likely just a 44/48kHz upsample being sold as "native" DSD!?

      Again, there's a major problem with quality control when even audiophile labels and download sites can't get their act together.

      Delete
  5. 16 Bit Signal in 24 Bit container

    Hi Archimago

    I enjoy reading your posts. You could have checked this more easily with looking at the bit statistics of the signal and would be able to see, that the lower bits (17 to 24) are static. This can be done with the Audio Precision measurement system, or if you own RME hardware, with RME Digicheck software, or with Wavelab Bit Statistics or with iZotope Bit Statistics. A second method would be to look at the level probability distribution. And here you would see no content, below – 96 dBFS.

    Best Regards
    Juergen

    ReplyDelete
    Replies
    1. Thanks Juergen for the techniques!

      Delete