Three versions of Björk’s Vulnicura under the Xivero MusicScope


“Heartbreak hurts but you can dance it off”. Clearly The Hold Steady’s message of cutting a rug to ease the pain has yet to reach Björk. Her eighth studio album loosely documents the dissolution of her long-term relationship with artist Matthew Barney.

The album title gives the game away up front: Vulnicura translates as “cure for wounds”. This ain’t no party album, it’s a brooding, majestic ache for something gone. Lush string arrangements sit atop meticulously crafted beats. On two cuts, programming comes from The Haxan Cloak (“Family”) and Spaces (“Quicksand”). Björk delivers an understated vocal performance throughout, positioning Vulnicura as closest genetically to 1999’s Vespertine. (I’ve never really been one for her shriekier stuff).

An Internet leak kicked Vulnicura’s official release from March to January but CD and vinyl are formats are holding fast to the original schedule. You can’t yet stream it on Tidal, Qobuz or Spotify either. The latter reports the omission like this:


That leaves us with n lossy download (US$9.99), a lossless 16bit/44.1kHz ‘CD quality’ download (US$11.99) and a 24bit/96kHz hi-res download (US$15.99).

I snagged all three from so as to be confident that each are derived from the same master.

Why? To road test Xivero’s MusicScope software is why.

Xivero CEO Lars Inger introduced his Windows / OS X app via email:

“We are a startup in digital audio processing, just releasing our new product MusicScope which is a music microscope to do a deep analysis of audio tracks. It is our goal to offer the hi-fi enthusiast a tool he can use to do a fast analysis on his digital audio tracks, whether it be DSD, FLAC, ALAC or WAV.”

“Our first reference customer, who operates an online store to sell high resolution music, is HighResAudio in Germany. They use the MusicScope to do perform quality checking on their new releases, verifying whether the records are really up to native studio master quality.”


MusicScope sells for the not unreasonable amount of €24.37. That’s just shy of US$30 at present exchange rates.

The all-black user interface is simple to use: load in a file – or files – and hit the microscope button to kickstart the analysis. A live reports is generated as the software works its way through the track. Text and graphical formats are saved to the song’s hard drive location upon completion.

To properly baseline all three editions of Vulnicura, the 16bit/44.1kHz (CD version) of opening track “Stonemilker” was the first to see MusicScope treatment.

Before the results, we need to consider a little digital audio theory. The Nyquist–Shannon sampling theorem states that the digital encoding of an analogue signal at a sample rate of XkHz will be sufficient to capture all frequencies up to a limit of 0.5XkHz. In the case of Redbook CD’s 44.1kHz sample rate that means all frequencies up to 22.05kHz are captured in the encoding process.

In theory, the limit of human hearing is 20kHz so a CD should cover all bases but the majority of ears over 30 years probably won’t reach that high. Hearing degrades with time and abuse, leaving the majority of middle-aged folk with upper limits of 16kHz or 17kHz.

However,we don’t expect to be short-changed when buying a CD or lossless download. The MusicScope analysis of our 16bit/44.1kHz-encoded Björk file should show its frequency range extending up to 22.05kHz…


And the green ‘smear’ at bottom of the graphical report shows precisely that. In the gently downward sloping orange data plot above it you’ll note the upper limit of 22.05kHz measures close to -96db. That’s also good news: 1bit of data depth is equivalent to 6db of dynamic range. 16 x 6db = 96db, just as we’d expect from a 16bit file.

Now let’s see what a 320kbps-encoded MP3 of the same song looks like:


Note the upper frequency limit is low-pass filtered around 19kHz? Lossy compression relies heavily on psyhcoacoustic theory to reduce file size by (among other things) removing the very highest frequencies that we supposedly cannot hear. And yet listening to this MP3 next to the lossless version via Chord Hugo and Beyerdynamic T1 headphones exposes the compressed version as relatively tense and nervy. In view of the additional two bucks required for the lossless download, why wouldn’t you?

Time for the 24bit/96kHz analysis. Drawing once again on the Shannon-Nyquist theorem, we’d expect to see a frequency range extending all the way to to 48kHz:


Do we see that? Yes we do. Also note the file’s higher dynamic range – it’s closing in on 140db. That tallies with the expected theoretical value of 144db (24bits x 6db).

Our MusicScope analysis indicates nothing to suggest that this hi-res release is a direct up-sample of the Redbook version. This shows up in listening tests too with the hi-res version sounding that bit more easeful and relaxed than its Redbook cousin but the delta is far from major. Careful: this difference might not just be a matter of ‘more data’. The digital filter’s ringing being shifted out of the audible band also helps. More info on this here.

Asked to put a number on it I’d probably peg the HRA as a 5% improvement over the CD version. Is it worth the extra US$4? For now, yes, probably. However, as soon as Vulnicura drops into the Tidal or Qobuz lossless streaming catalogues, its value for money quotient will take a serious hit.

Subscription streaming services are slowly killing the download market whilst knifing the back of music piracy along the way. Where’s the incentive to buy the MP3 version of Vulnicura for US$10.99 when the same cash nets a month’s worth of Spotify Premium (and access to millions of songs)? And if you’ve not stumped up to kills the adverts, a lossy version can still be streamed for free. The need to hit up your favourite torrent site is all but dissolved.

The audiophile world likes to make a lot of fuss about the benefits of hi-res audio but the conversation quickly shifts to a mumble once the issue of pricing is raised. At US$15.99 for the 24bit/96kHz, juicing that little bit extra from the latest Björk album doesn’t come cheap. That’s fine for audiophiles who want the very best but the minor audible improvements fail to support an argument for broader adoption by Joe Public and his mates.

Further information: Bleep | Xivero MusicScope

Written by John H. Darko

John lives in the NOW + HERE = NOWHERE. He derives an income from the ad revenues of DAR. John is also an occasional staff writer for Stereophile, 6moons and TONEAudio.

Twitter: DarkoAudio
Instagram: DarkoAudio
Facebook: DAR


Leave a Reply
  1. John,
    Great article, thanks for the effort in doing this work, and kudos to Bjork and her team for being so straightforward with the stuff they sell.
    Me, I’ll gladly pay a few more dollars for genuinely higher quality files, but there’s too much mutton out there being sold as lamb. If you pay for 24 bits @ 96khz that’s what the content should be, not just the wrapper it comes in.
    Thanks a bunch

  2. Probably the best analysis of A,B,C comparison I’ve read so far. Credit where credit is due on this one.
    Looks like a good piece of software to back up the audible differences experienced as most of us like to see things in numbers if possible. I’m wondering how this will pan out, will we expect the sites issuing high res audio files to validate them or expect the public to invest in this software and catch them out and try and get a refund? I suspect a bit of both.
    I believe only HIFI News check digital file quality on behalf of the public currently

  3. It is on Qobuz for quite some time. And if you are a Qobuz Sublime subscriber it will cost you 7.99 EUR , same as the CD version. Just un case.

  4. John, a few questions. First, did you level match before doing your comparisons? I’m not sure which MP3 version you have, but the DR database shows that the 24/96 version as sold by is louder than the iTunes/AAC version by at least 0.5dB, and on some tracks much more than that. When comparing different versions like this, you should measure them all using the TT dynamic range meter or a similar tool that can give you RMS and peak levels per track. If the various versions were all derived from the same master, RMS levels should be identical down to a hundredth of a dB. The process of MP3 encoding may push peak levels to 0dBFS if the peaks are very close to that on the lossless version, but RMS levels should always be identical, and any A/B results with varying degrees of loudness are automatically invalid for obvious reasons.

    Second, did you use something like Foobar’s ABX comparitor tool so that you couldn’t tell which track was which when you were listening?

    The mastering on this album is a bit better than average for a typical modern pop release, but only “Stonemilker,” “Lionsong,” and “History of Touches” can be considered reasonably dynamic. The album’s last two tracks on the other hand are horribly crushed, and likely sound like garbage regardless of whether the format is MP3 or 24/96 lossless.

    If they bothered to do a dedicated vinyl master for this album, it will likely hammer all of the digital versions, but just looking at the DR scores for Bjork’s previous vinyl releases, it at least appears like they just cut from the same master created for the CD.

    • Hey Dave. I level matched as best as any could with the comparative listening taking place across both portable head-fi and two-channel rigs. I don’t use Foobar as I’m on a Mac and double blind testing has never been part of DAR’s MO.

      I have the vinyl of Homogenic and it sounds pretty crappy.

      • John, your explanation on how MP3 works is not correct. It is not just the high frequencies that are removed, there is a lot to that psychoacoustic model than just a low-pass at 19kHz.

        As Dave pointed out, comparisons like these are not really valid since looking at spectral content to figure out what it sounds like is dubious at best. For example, imagine source material A was high-res version, mastered at DR3 of oh, say Metallica’s Death Magnetic. Then I give you a source material B which is a remastered, uncompressed version of Death Magnetic at DR10 in 192kHz MP3 format. You pull up the spectral and say to the audience, “See guys, MP3 cut you off at around 17kHz. Ouch, that psychoacoustic model sucks. Give me high-res.”

        You just made a TERRIBLE mistake. That MP3 will sound an order of magnitude better than the high-res version and will easily be able to even DBT it. I realize that is not what is going on here with Bjork’s record, but you are treading on dangerous territory just looking at spectral output and go, “See, MP3’s suck.” They don’t.

        So now we are on to a listening test part. On Mac, Decibel supports ReplayGain natively, I suggest you give it a try. At least then you can level match appropriately and compare. At 320k CBR, I doubt other than bats, dogs, and Pono owners can hear any difference.

        Before anyone calls me out on my playback chain, I will gladly share it:

        MBP->GeekOut IEM->JHA Roxanne’s
        HFM-901 with balanced card -> LCD-3 (Vegan!)

        I think these are “good” enough.


        • Hey Alex – nowhere to do I say MP3s suck, nor do imply it. I’m quite fond of the format actually. I’ve added “(among other things)” to clarify that the low-pass filter isn’t the only thing applied to a file when compressed with the MP3 codec. My point with this article was to show how purchasers of the hi-res version do indeed get what they might expect, both from a spectral POV and from listening. IOW, the numbers and spectrals SUPPORT what I hear.

          I get that DR matters but in the case of this Bjork album, the LRA values – 12.7, 12.8 and 12.7 – are near-identical suggesting the SAME master was used. More info here too:

          • Fair enough John. I actually agree with you 100% about your overarching points that high-res’ biggest issue isn’t proving its fidelity, but its worth. And you even talk about it in this article as well.

            My only though looking at the spectral content between the CD and MP3, I don’t see how anyone could hear a difference in a controlled environment level matching the two. I think if you did a real ABX test, you’d realize you are statically guessing on which is which.

            But even with that said, I still agree that the extra $4 bucks is definitely worth it to have prestine copy and transcode as needed. That’s always been my MO.

            Btw, am I the only that likes her earlier stuff better? I feel I’m all alone in this. The trippy stuff is starting to get on my nerves.

  5. Just to be clear : Qobuz Sublime subscription gives you lossless CD quality streaming and reductions for buying 24bits albums. Most of the times, you can buy 24 bits quality for the mp3 price.
    There is only one thing : Qobuz Sublime costs… 219 Euros a year.