Lossy file analysis: Conclusive enough for notes?

A couple And What Have You Done With My Body, God?

While I don’t necessarily have an issue with the method used to reach this conclusion, I feel like the standard should be higher in order to make such a definitive statement about it in the Release Notes, like something in the liner notes, an interview quote, a forum posting, something on social media, anything from someone who could actually confirm how those tracks came to sound like they do. I feel like one person’s analysis with one tool doesn’t quite make the grade.

Thoughts?

I think this is a really neat and useful addition, especially for releases with multiple versions.

ZJ_AJ
Release Notes, like something in the liner notes, an interview quote, a forum posting, something on social media, anything from someone who could actually confirm how those tracks came to sound like they do.

I wouldn't count on that; it's such a specific wad of information that is hardly noticed, let alone discussed in-depth often. The source for the information is the release itself, which is more important than anything. However, I do think mjb could substantiate their findings a little more, maybe screenshots of the spectrogram and a little more insight?

Some mastering issues routinely are mentioned in comments/reviews, but some are just as often mentioned in notes. There's no list of what does and does not rise to the level of being note(s)worthy, but generally if it's arguably an error, then it's more likely to be seen in the notes—like when a wrong version was used, an intro has been cut off, or something was sourced from an MP3 instead of a lossless digital master (on a release where that's unexpected).

Plus, on ZTT label releases in particular, we've got a long history of being, well, as annoyingly pedantic and detailed as the releases themselves. I blame Paul Morley.

I can show you a quick example for one of the tracks, comparing lossless to lossy. Open these in separate browser tabs, make sure the images are showing full-size (they are 1920x1080), and then switch between them with Ctrl+Tab or whatever you normally use:
• Lossless version, zoomed out fully
• Lossy version, zoomed out fully
• Lossless version, zoomed in to a 3.7-second section
• Lossy version, zoomed in to a 3.7-second section

With hi-res views like this, you can see and infer so much more than with cruder tools. For example, the lossy version technically does not have a "cutoff" from a low prefilter. Rather, the encoder just selectively omitted the frequency components which are likely to be masked or otherwise inaudible: the quieter tones as often found in the upper ranges, and most especially the vast majority of those above 16 kHz, the boundary of "scalefactor band 21" where preserving anything at all is very expensive, bitrate-wise.

Zoomed in, you can also see what I call the "shotgun blast" effect in the lower ranges, where masked frequencies appear as black "holes". You can also see that missing and included parts are usually in 26-millisecond chunks, exactly the size of MP3's "long blocks" at the 44100 Hz sample rate. Where transients occur, such as the peak right before 3:27.4 in the zoomed-in view, you can see MP3's "short blocks", which are one-third the width: a sign of a smart encoder trying to minimize pre-echo. These are not things you will ever see in lossless. (OK, well, unless the lossless original was in fact constructed partially or fully from sampled MP3s, but that's not a possibility for this material recorded in the mid-1980s.)

The spectrograms of all the tracks I noted as being lossy all have these features. Hope this helps.

I know way to little about lossy/lossless files to really comment on the way that it was established that the sources where lossy. But the way it was is that a commonly accepted way to check for lossy source material or a way mjb have thought of themselfs?

I’m on the fence if it’s notes or review material. If the release don’t claim to be all lossless source material it fells like if a release have a poor mastering and then just should be a review. If the release claims to be all lossless source material and it’s proven not to be then it’s release notes material.

I would however like to see either a better explination of the way it was checked in the submission notes (so anyone with the correct knowlege can dublicate it) or even better some source/link to a site that explains the process.

Sometimes I have to catch myself and question whether my own doubt about someone's edit is rooted in genuine concern for whether the information could be wrong, or rather in bureaucracy and personal distrust. As happened on Wikipedia, the edit notification system encouraged us to be somewhat possessive and weigh in on every little thing, and some of the rules intended to put an end to disputes have encouraged a culture of heightened, sometimes absurd skepticism. We all get caught up in it, myself included, sometimes to the detriment of the database.

Says who? The New York Times hasn't run an article on the culture of our little tiny corner of the Internet, so how dare I present such a claim as fact?

Seriously, though, there's just some info which does come from the release, but which we have to trust someone else's capability to interpret. To elaborate, these are the kinds of "trust me bro" statements of fact which we do routinely accept, sans proof, in release notes:

"Track A1 is a mash-up of Joy Division 'Love Will Tear Us Apart' and Missy Elliott 'Get Ur Freak On'"
"Track 3 is the version with Andy White on drums, not Ringo"
"Although the artwork says Stereo, all tracks are actually mastered in mono"
"Track 12 contains Song 1 (3:53) and unlisted Song 2 (4:02), separated by 2 minutes of silence"
"Track B2 samples '(I Can't Get No) Satisfaction'"
"Track 17 is misidentified as the Extended Mix, but is actually the Radio Edit"

You probably know how to play a record and determine the correct speed. You can figure out track durations using a variety of methods. You know how to listen in headphones and hear whether something is mono, stereo, and maybe you can even recognize what they call "process stereo" (mono made stereo-ish with studio trickery). You can also sometimes figure out whether the tracks are correctly identified, or you can find the titles of unlisted tracks with the help of Shazam or similar. You can sometimes tell if something is the original or a re-record …And yet, if questioned, you probably couldn't point to a rule, website or other external reference which verifies that information, or which would explain to a skeptic or non-technical person the exact procedure you used, or just how far someone should trust any such "-analyzed" info.

If the person adding such things to the notes happens to say what source or method they used, then I try to be satisfied with that. Or if it's just something like they're apparently an expert on this particular subject/artist/label, and I'm clearly out of my depth, then I ultimately accept that maybe some topics just aren't for me to weigh in on—I leave it to the experts/pedants/nerds. What I won't do is demand that some provide a YouTube tutorial which will bring me up to speed on what may well have required years of experience for them to learn. I won't demand that they prove in every edit why they are an authority on an esoteric matter.

(That said, if authority matters, I've been at this a good long while; my profile might give you some idea. Aside from my particular interest in ZTT/AoN, the science of sound has been a decades-long hobby and occasional career for me. I've studied books on these topics, and I've spent way too much time on the Hydrogenaudio forum and wiki, where technical matters relating to digital audio are discussed among experts with an emphasis on objectivity.)

The spectogram comparison is very clear about it being a lossy source. No doubt.