Page 1 of 2 12 LastLast
Results 1 to 25 of 44
  1. #1
    Banned
    Join Date
    Oct 2011
    Location
    Amsterdam
    Posts
    3,515
    Blog Entries
    4

    When is 24 bits not 24 bits?

    After users "nu774" and "skamp" @ Hydrogen Audio had pointed out a method to detect straight 16->24 bit padding by using the "analyze" option of the FLAC encoder to show the "wasted" or "unused" bits in tracks, I decided to run a script on my whole music collection of 20,000+ tracks, and only found 6 albums showing a significant pattern of "unused" bits.

    One album, a pure spoken voice one (language training record) had "effective" bits between 13 and 15 depending on the track. 2 CD rips (Smoke & Strong Whiskey by Christy Moore and In My Memory by Dj TiŽsto) had consistent 15 bits out of 16, indicating digital processing and less-than-perfect use of headroom.

    The most interesting discovery was that 2 supposedly 24 bit downloads (A Retrospective by The Unthanks and the LSO Haitink Beethoven Symphony no. 9, both downloaded from the Bowers & Wilkins Society of Sound site) had consistent 16 effective bits, despite being nominally 24-bit FLACs.

    One additional SoS album, Portico Quartet, had strangely inconsistent bit depths:

    23/24 bits Portico Quartet/Portico Quartet/01 Window Seat.flac
    16/24 bits Portico Quartet/Portico Quartet/02 Ruins.flac
    16/24 bits Portico Quartet/Portico Quartet/03 Spinner.flac
    24/24 bits Portico Quartet/Portico Quartet/04 Rubidium.flac
    19/24 bits Portico Quartet/Portico Quartet/05 Export to Hot Climes.flac
    23/24 bits Portico Quartet/Portico Quartet/06 Laker Boo.flac
    22/24 bits Portico Quartet/Portico Quartet/07 Steepless.flac
    20/24 bits Portico Quartet/Portico Quartet/08 4096 Colours.flac
    16/24 bits Portico Quartet/Portico Quartet/09 City of Glass.flac
    22/24 bits Portico Quartet/Portico Quartet/10 Trace.flac

    After user "pdq" at @ HA pointed out that all tracks on a CD are an even multiple of a CD frame (588 pairs of left and right samples), I ran another script to check that as well.

    Turns out all the tracks on the Portico Quartet album had exact multiples of 588 samples (the odds of that happening by accident is 1/588 or 0.17%, for all 10 tracks it is 1 in 940526814607642247350452224), and the album by The Unthanks had exact multiples in 6 out of 11 tracks.

    Conclusion? The Unthanks and LSO Haitink Beethoven Symphony no. 9 downloads from SoS are 16 bit material in a 24 bit dressing, and the Unthanks material is from a CD.

    It also seems the Portico Quartet album material comes from a CD, but with some light processing done on some of the tracks for the "24-bit" release.

    This method of detecting padding only works on totally obvious "padding 16 bits to 24 bits" stuff. As soon as you do any processing/resampling, or add a bit of (inaudible) low-level noise, you cover your tracks, so who knows how many nominally 24-bit downloads are "dressed-up" from 16 bit material, but in the case of these 2 SoS downloads, the evidence is clear.

  2. #2
    Newbie Objectionist wgscott's Avatar
    Join Date
    Apr 2010
    Location
    Gondwana
    Posts
    8,782
    Blog Entries
    41
    Thanks for this incredibly useful, and extremely disappointing (with respect to the Society of Sound) post.

    Have you contacted them? I would be more than happy to. I need to replicate the results first (just a scientist thing, nothing personal).

    Again, thanks for posting this.

  3. #3
    Banned
    Join Date
    Oct 2011
    Location
    Amsterdam
    Posts
    3,515
    Blog Entries
    4
    Quote Originally Posted by wgscott View Post
    Have you contacted them?
    No, can't be bothered - I just let my subscription expire.

    I would be more than happy to. I need to replicate the results first (just a scientist thing, nothing personal).
    More than happy to send you the linux scripts I used. The "wasted bits" one is adapted from this one.

  4. #4
    Newbie Objectionist wgscott's Avatar
    Join Date
    Apr 2010
    Location
    Gondwana
    Posts
    8,782
    Blog Entries
    41
    Yes, please.

  5. #5
    Banned
    Join Date
    Oct 2011
    Location
    Amsterdam
    Posts
    3,515
    Blog Entries
    4
    Quote Originally Posted by wgscott View Post
    Yes, please.
    Here is the shell script for checking for multiples of CD frame size. Takes a directory name as argument, and recursively reports the FLAC files that have a content length that is an even multiple of 588 samples. Relies on the output format of the "file" command, so might not work on all linux versions. Will of course report all CD rips, so only use on downloaded albums.

    Code:
    #!/bin/sh
    
    if [ $# -ne 1 ]
    then
        echo "usage: $0 directory"
        exit 1
    fi
    
    dir=$1
    
    find "$dir" -type f -name '*.flac' | (
        while read f
        do
            length=`file "$f" | sed -e 's/ samples.*//' -e 's/.* //'`
            if [ `echo "$length % 588"  | bc` -eq 0 ]
            then
                echo $f
            fi
        done
    )

  6. #6
    Banned
    Join Date
    Oct 2011
    Location
    Amsterdam
    Posts
    3,515
    Blog Entries
    4
    For the "wasted bits" script, you can use this script (from HA) - all my changes have to do with my local environment.

  7. #7
    Newbie Objectionist wgscott's Avatar
    Join Date
    Apr 2010
    Location
    Gondwana
    Posts
    8,782
    Blog Entries
    41
    Both shell scripts work fine on OS X, without any changes being needed. (Also, I think all linux distros use the same gnu fileutils).

    I had to compile flac for OS X.

    Here is the output I got for several of the B&W LSO flac downloads. Briefly, only Beethoven Symphony #9 was 16 bit, and Brahms Symphony #2 and #3. I have a lot more to check, but the main thing is it appears to be sporadic, not uniformly 16 bit. I'll ask them to fix it.

    My output:
    Code:
    zsh-% fbits **/*.flac
    16/24 bits	Beethoven - Symphony No 9 Choral-FLAC24/01. Movement 1 - Allegro ma non troppo un poco maestoso.flac
    16/24 bits	Beethoven - Symphony No 9 Choral-FLAC24/02. Movement 2 - Scherzo Molto vivace.flac
    16/24 bits	Beethoven - Symphony No 9 Choral-FLAC24/03. Movement 3 - Adagio molto e cantabile.flac
    16/24 bits	Beethoven - Symphony No 9 Choral-FLAC24/04. Movement 4 - Presto - Allegro ma non troppo - Vivace Adagio cantabile.flac
    24/24 bits	Beethoven Symphonies 1 5-FLAC24/01. Symphony No 5 - Allegro con brio.flac
    24/24 bits	Beethoven Symphonies 1 5-FLAC24/02. Symphony No 5 - Andante con moto.flac
    24/24 bits	Beethoven Symphonies 1 5-FLAC24/03. Symphony No 5 - Allegro.flac
    24/24 bits	Beethoven Symphonies 1 5-FLAC24/04. Symphony No 5 - Allegro.flac
    24/24 bits	Beethoven Symphonies 1 5-FLAC24/05. Symphony No 1 - Adagio molto - Allegro con moto.flac
    24/24 bits	Beethoven Symphonies 1 5-FLAC24/06. Symphony No 1 - Andante cantabile con moto.flac
    24/24 bits	Beethoven Symphonies 1 5-FLAC24/07. Symphony No 1 - Menuetto Trio Allegro molto e vivace.flac
    24/24 bits	Beethoven Symphonies 1 5-FLAC24/08. Symphony No 1 - Adagio - Allegro molto e vivace.flac
    24/24 bits	Beethoven Symphonies 6 2-FLAC24/01. Erwachen heiterer Gef hle bei der Ankunft.flac
    24/24 bits	Beethoven Symphonies 6 2-FLAC24/02. Szene am Bach.flac
    24/24 bits	Beethoven Symphonies 6 2-FLAC24/03. Lustiges Zusammensein der Landleute.flac
    24/24 bits	Beethoven Symphonies 6 2-FLAC24/04. Gewitter Sturm.flac
    24/24 bits	Beethoven Symphonies 6 2-FLAC24/05. Hirtengesang Frohe dankbare Gef nach dem Sturm.flac
    24/24 bits	Beethoven Symphonies 6 2-FLAC24/06. Adagio molto Allegro con brio.flac
    24/24 bits	Beethoven Symphonies 6 2-FLAC24/07. Larghetto.flac
    24/24 bits	Beethoven Symphonies 6 2-FLAC24/08. Scherzo Trio Allegro.flac
    24/24 bits	Beethoven Symphonies 6 2-FLAC24/09. Allegro molto.flac
    24/24 bits	Beethoven Symphony 7-FLAC24/01. Poco sostenuto Vivace.flac
    24/24 bits	Beethoven Symphony 7-FLAC24/02. Allegretto.flac
    24/24 bits	Beethoven Symphony 7-FLAC24/03. Presto.flac
    24/24 bits	Beethoven Symphony 7-FLAC24/04. Allegro con brio.flac
    24/24 bits	Beethoven Symphony 7-FLAC24/05. Allegro con brio.flac
    24/24 bits	Beethoven Symphony 7-FLAC24/06. Largo.flac
    24/24 bits	Beethoven Symphony 7-FLAC24/07. Rondo alla Polacca.flac
    16/24 bits	Brahms - Serenade No 2 Symphony No 3-FLAC24/01. Serenade 2 Movement I.flac
    16/24 bits	Brahms - Serenade No 2 Symphony No 3-FLAC24/02. Serenade 2 Movement II.flac
    16/24 bits	Brahms - Serenade No 2 Symphony No 3-FLAC24/03. Serenade 2 Movement III.flac
    16/24 bits	Brahms - Serenade No 2 Symphony No 3-FLAC24/04. Serenade 2 Movement IV.flac
    16/24 bits	Brahms - Serenade No 2 Symphony No 3-FLAC24/05. Serenade 2 Movement V.flac
    16/24 bits	Brahms - Serenade No 2 Symphony No 3-FLAC24/06. Symphony 3 Movement I.flac
    16/24 bits	Brahms - Serenade No 2 Symphony No 3-FLAC24/07. Symphony 3 Movement II.flac
    16/24 bits	Brahms - Serenade No 2 Symphony No 3-FLAC24/08. Symphony 3 Movement III.flac
    16/24 bits	Brahms - Serenade No 2 Symphony No 3-FLAC24/09. Symphony 3 Movement IV.flac
    24/24 bits	Bruckner Symphony No 4-FLAC24/01. Mvt 01 Bewegt nicht zu schnell.flac
    24/24 bits	Bruckner Symphony No 4-FLAC24/02. Mvt 02 Andante quasi Allegretto.flac
    24/24 bits	Bruckner Symphony No 4-FLAC24/03. Mvt 03 Scherzo Bewegt - Trio Nicht zu schnell Keinesfalls schleppend.flac
    24/24 bits	Bruckner Symphony No 4-FLAC24/04. Mvt 04 Finale Bewegt doch nicht zu schnell.flac
    zsh-% pwd
    /Volumes/WarfRat/Media/non-iTunes Music/Other/Society of Sound/Bernard Haitink London Symphony Orchestra

  8. #8
    Newbie Objectionist wgscott's Avatar
    Join Date
    Apr 2010
    Location
    Gondwana
    Posts
    8,782
    Blog Entries
    41
    I've now checked through 39 of the B&W putatively 24-bit downloads. Of these, three are 16 bit (Unthanks, Beethoven Sym #9, and the Brahms Sym 2/3).

    So the fail rate is about 1/13. I'm willing to give them the benefit of the doubt. The only one I really care about is the 9th. I will find out what they have to say.

  9. #9
    Newbie Objectionist wgscott's Avatar
    Join Date
    Apr 2010
    Location
    Gondwana
    Posts
    8,782
    Blog Entries
    41
    From my (limited) collection of HD Tracks downloads that I haven't deleted in disgust, the only problematic one I found was this:

    16/24 bits John Coltrane/A Love Supreme/1-A Love Supreme Part I Acknow.flac

    (I'm only checking the first track now for expediency.) What is interesting is that Lush Life, which has been withdrawn by HD Tracks because it is problematic, reports as 24/24 bits.

  10. #10
    Newbie Objectionist wgscott's Avatar
    Join Date
    Apr 2010
    Location
    Gondwana
    Posts
    8,782
    Blog Entries
    41
    (duplicated your result -- deleted post )

  11. #11
    Banned
    Join Date
    Oct 2011
    Location
    Amsterdam
    Posts
    3,515
    Blog Entries
    4
    Quote Originally Posted by wgscott View Post
    (duplicated your result -- deleted post )
    Well, still good to get my results independently verified.

  12. #12
    Banned
    Join Date
    Oct 2011
    Location
    Amsterdam
    Posts
    3,515
    Blog Entries
    4
    Quote Originally Posted by wgscott View Post
    Both shell scripts work fine on OS X, without any changes being needed.
    Great! Good to hear!

    I had to compile flac for OS X.
    Maybe you could make the binary available to Mac users?

  13. #13
    Quote Originally Posted by wgscott View Post
    I've now checked through 39 of the B&W putatively 24-bit downloads. Of these, three are 16 bit (Unthanks, Beethoven Sym #9, and the Brahms Sym 2/3).

    So the fail rate is about 1/13. I'm willing to give them the benefit of the doubt. The only one I really care about is the 9th. I will find out what they have to say.
    I always wondered about the rather funky format of 24/48 they had on many recordings (I got a free subscription to the society of sound when I got my B&Ws, but haven't renewed it).

    Indeed, given this failure rate, you'd hope that there is no concious large scale effort of cheating the customer ongoing, but it is still scary that this thing keep happening.

    Please keep us posted.
    iMac late 2009 8GB > Audirvana+/Amarra > Audioquest Forest USB > BelCanto mlink > Atlas Ascent BNC > Exposure 2010S2 Dac > Audioquest King Cobra > Exposure 3010S2 Integrated > Chord Carnival Classic > B&W CM8

  14. #14
    Site Founder The Computer Audiophile's Avatar
    Join Date
    Nov 2007
    Location
    Minneapolis, MN, USA
    Posts
    12,013
    I've found a txt file with some B&W downloads that says the music's original bit depth and sample rate and what was used to create the downloadable version. The txt file wasn't obvious but at least it was full disclosure (after the fact).

  15. #15
    Banned
    Join Date
    Oct 2011
    Location
    Amsterdam
    Posts
    3,515
    Blog Entries
    4
    Quote Originally Posted by The Computer Audiophile View Post
    I've found a txt file with some B&W downloads that says the music's original bit depth and sample rate and what was used to create the downloadable version. The txt file wasn't obvious but at least it was full disclosure (after the fact).
    Of all the B&W downloads I have, the only one that had a "Read Me" text file was the album "Square Peg In A Round Hole" by Apparatjik. It said "These files have been prepared in 24-bit 44.1kHz FLAC format. 24-bit FLAC files are similar in quality to the studio master. They work like zip files so the size is compressed but none of the sound quality is lost. Unfortunately iTunes and Windows Media Player don't accept these files yet but many others do."

  16. #16
    Newbie Objectionist wgscott's Avatar
    Join Date
    Apr 2010
    Location
    Gondwana
    Posts
    8,782
    Blog Entries
    41
    The only way I could figure out how to contact them was via this blog page, which, remarkably, is about two of the defective recordings:

    Exclusive Unthanks compilation plus Beethoven's 9th Symphony | Bowers & Wilkins | B&W speakers

  17. #17
    Newbie Objectionist wgscott's Avatar
    Join Date
    Apr 2010
    Location
    Gondwana
    Posts
    8,782
    Blog Entries
    41
    Quote Originally Posted by Musicophile View Post
    I always wondered about the rather funky format of 24/48 they had on many recordings (I got a free subscription to the society of sound when I got my B&Ws, but haven't renewed it).
    When I got my CM7s, I didn't get anything from B&W (perhaps in more ways than one).

    Indeed, given this failure rate, you'd hope that there is no concious large scale effort of cheating the customer ongoing, but it is still scary that this thing keep happening.
    Of about 6 Beethoven symphonies, this was the only one that was problematic, so I don't see an incentive for them to do it deliberately.

    It is odd you can purchase those same recordings as 24/96.

  18. #18
    Quote Originally Posted by wgscott View Post
    When I got my CM7s, I didn't get anything from B&W (perhaps in more ways than one).
    Sounds like you're disappointed from your CM7? Have never heard them.

    On my side, I consider the CM8 to be among the best in their price range (and great WAF on top).

    Back to topic: I'm really very interested in what SOS has to say.
    iMac late 2009 8GB > Audirvana+/Amarra > Audioquest Forest USB > BelCanto mlink > Atlas Ascent BNC > Exposure 2010S2 Dac > Audioquest King Cobra > Exposure 3010S2 Integrated > Chord Carnival Classic > B&W CM8

  19. #19
    Newbie Objectionist wgscott's Avatar
    Join Date
    Apr 2010
    Location
    Gondwana
    Posts
    8,782
    Blog Entries
    41
    I think I am kind of at a low point in the hobby, where all I am seeing are the various deficiencies.

    It is a post to an old blog topic. They may never see it. If anyone can figure out a better way to get through to them, please let me know.

  20. #20
    Banned
    Join Date
    Oct 2011
    Location
    Amsterdam
    Posts
    3,515
    Blog Entries
    4
    Quote Originally Posted by wgscott View Post
    The only way I could figure out how to contact them was via this blog page, which, remarkably, is about two of the defective recordings:
    I did try to leave a comment as soon as I discovered the defects - my comments never appeared.

  21. #21
    Newbie Objectionist wgscott's Avatar
    Join Date
    Apr 2010
    Location
    Gondwana
    Posts
    8,782
    Blog Entries
    41
    Mine appeared instantly. It is fairly low tech. I don't think they screen the posts. Perhaps it is wroth trying again?

  22. #22
    Newbie Objectionist wgscott's Avatar
    Join Date
    Apr 2010
    Location
    Gondwana
    Posts
    8,782
    Blog Entries
    41
    By the way, I really enjoy your choice of title for this thread. Heh.

  23. #23
    Banned
    Join Date
    Oct 2011
    Location
    Amsterdam
    Posts
    3,515
    Blog Entries
    4
    Quote Originally Posted by wgscott View Post
    Mine appeared instantly.
    Strangely I can't see your comment either.

  24. #24
    Newbie Objectionist wgscott's Avatar
    Join Date
    Apr 2010
    Location
    Gondwana
    Posts
    8,782
    Blog Entries
    41
    Dang windoz users. Refresh your browza

  25. #25
    Newbie Objectionist wgscott's Avatar
    Join Date
    Apr 2010
    Location
    Gondwana
    Posts
    8,782
    Blog Entries
    41
    Quote Originally Posted by Julf View Post
    Strangely I can't see your comment either.
    Yup. It is gone now.

    I guess they deleted it.

    A fine way to deal with the problem. But not original.

Page 1 of 2 12 LastLast