No patches for digital audio releases | Hacker News Reader

No patches for digital audio releases

mcraiha.github.io

53 pointsmcraiha29 comments

A related issue: in some cases the industry has allowed anyone to upload a copy of a song, and so you run into the issue that you're at the mercy of whoever the uploader was when it comes to the quality of your upload.

I ran into a case where there were reports that a specific release of a CD has a much better mastering than other versions, it was a Super Audio CD (SACD) as I recall. I looked up the release on YouTube, since differences in mastering are clearly audible even in a lossy encode. The one upload was by a random person who had put it up in an HD video, but the volume of the release had inexplicably been boosted to the point that the sound was heavily distorted!

Another case: the most popular version of the Eminem "Lose Yourself" music video (over a billion views) is uploaded by some random person, with one of those ludicrous "no copyright infringement intended" messages in the description. It's seemingly the one that the industry approved as the legitimate version of the music video on YouTube, however. I actually tried tracking down the official version on the artist's page, and (at the time) it had actually been taken down by the studio with one of the standard DMCA takedown messages YouTube shows!

Is there some deep-learning tool to reconstruct songs that got distorted in various ways?

"Content aware fill" works on the basis that you just want "some grass" or "some rocks" and don't know or care whether it matches the exact grass that was in the original picture.

If people have already heard the original song, then the computer isn't going to be able to fill in the lost information in a way that matches their memory.. even with deep learning

I'm not convinced. Counterexample: if the original song is a sine wave, and the distortion turns it into a square wave, then a simple algorithm can turn it back into a sine wave with the right frequency, very easily.

Counter-counterexample: your algorithm was too naive. The original song was much more interesting:

A sine wave at a single frequency, but with rhythmic volume swells from nearly inaudible to 0dBFS. The slow hopeful crescendos in the opening give way to a more playful bounce in the middle, before building to a frantic ending with nearly nauseating swings.

Pitchfork gave it a 10.0, and called it “Better than the remaster of In The Aeroplane Over The Sea. If our scale allowed, it would have gotten a 12.” What a ride.

The distortion that transformed it was fairly pedestrian: turn it up until it exceeds the dynamic range of the recording medium.

The result was that boring square wave at constant frequency and amplitude. Can you expect an AI to reconstruct this original, given both of our proposed original songs map to the same output?

I’d also argue that the original proposed algorithm is not simple, since it would have to remove the aliased frequency components when converting from the square wave back to a sine.

Regarding the Eminem song, I found this on Reddit, don't know if it's true:

> YouTube allows copyright owners to choose to leave videos/channels up when they report them and simply divert all monetization from the videos to the copyright owner. That is likely what is happening here.

Could be some sort of authenticity play from Eminem, him wanting to look less "corporate".

I've been uploading a copyrighted show on a small YouTube channel[1] for fun for a long time now. This is exactly what happens[2] and why I've never had a strike or anything.

[1] https://www.youtube.com/c/uktotheworld/videos [2] https://i.imgur.com/r2Gl9cB.jpg

That's interesting phrasing. "This video was found to contain copyrighted material." So what? Tons of stuff contains copyrighted material, under fair use. But there's no acknowledgement of that and it's not obvious that they would allow you to dispute that your use of the material was fair use. (Clearly, in your case, that's not at issue here. I just think it's strange.)

I think there's something to be said for the inability to issue patches. It's created a culture in the software industry, especially in games, of rushing to get the first version out in basic working form and then have a "day one patch" followed by many subsequent fixes. Quality suffers because there's an attitude that problems can always be fixed later, so it's not a disaster if the initial release is below standard.

30 years ago this wasn't an option. Whatever the development team designated as the golden master would be written to physical cartridges, floppy disks, or CDs. Any bugs that were there would stay there, so you had to get it as close to perfect as possible because there were no second chances. This pressure meant that people were, by necessity, more careful. How many times have you installed a patch for a Super NES or Playstation 2 game? Compare that to today's releases.

For developers and other creators, the idea of being able to repeatedly update things is attractive. It's also necessary in certain spheres, particularly in fixing security vulnerabilities. But I know the sense of finality and relief can be immense once you publish something that can't be changed (as was the case with my thesis) and that something is finally "done" and I know I'll never have to worry about it again. The requirement to get something right the first time has a significant impact on quality.

While "Day One Patch" is basically a curse (and a reminder that pre-ordering games is a sucker bet since even people who buy it when it's released aren't sure yet what they'll actually get) it's worth remembering that even when things were "written to physical cartridges, floppy disks or CDs" there actually were version changes.

If you get into speed running you'll find that on some of these very old games it's specified the exact version you must run, e.g. the Japanese original CD release with no version on the title screen, not the English or German versions, and not a later Japanese version showing 1.0 on the title screen. Because even if a casual player doesn't care that there was a glitch where you can phase through walls by pressing PAUSE once per frame ten times in a row on the other versions, the speed runners absolutely do care.

There are musicians who clearly don't see the recording as the primary outlet, and so aren't bothered if the CD goes out with a mediocre attempt at a song because for them the live performance is what matters. That's not something that really exists for video games.

Believe it or not, games regularly did get updated even back in the NES era. There's many games out there that had a long enough shelf life to get a PRG1/CHR1 revision and reprint. Furthermore, almost no games were sim-ship[0] back then, so even if games didn't get a re-release in the same territory you still had the opportunity to patch a game for the next country's release. e.g. US v1.0 might be equivalent to JP v1.2 even if Japan never got that version.

This is also why a lot of Europeans get really confused at old AVGN videos: some of the games he trashes were dramatically improved for Europe and only sucked in the US. (e.g. Fester's Quest)

Furthermore, developers back then were very much not "more careful"; in fact, there's plenty of examples of games that were shipped back then with little or no playtesting or QA work done. And likewise, being able to patch a game later doesn't really fix the reputational harm from shipping broken software. If the game is not playable day-one[1], nobody will care when it's playable on day 255.

[0] Industry jargon for SIMultaneous SHIP - as in, release in all territories

[1] Day-one patch inclusive. I know people hate them, but they let developers eke out an extra month of QA and bug-fixing time because of how cert works.

I've noticed this too. Combine the constant patching with auto-updates and my whole system starts to feel like it's made of sand, constantly shifting under my feet. It almost feels disrespectful. Like, I use my computers to work. The most important thing is that it work for me, reliably and predictably, not that it serve as a showcase for whatever the product manager du jour needs in their promotion packet. There's just a total failure to prioritize the needs of the user.

Needless to say I keep auto-updates off. What a world we've made where updating your software has become something to be avoided.

The first time I came across patched video games was the late 80s. You'd regularly get updates to PC games on magazine cover disks.

GAS's (Wolfgang Voigt) latest album [1] was criticised by many (including Pitchfork [2]) for having a very high pitched (~12khz) tone throughout the album. Pitchfork was speculating in their review that the artist suffers from hearing loss, after decades spent in the Cologne techno scene. In response they removed the 12khz tone, and uploaded the new versions to all digital platforms, and even sent repressed CDs to customers.

[1] https://kompakt-gas.bandcamp.com/album/der-lange-marsch

[2] https://pitchfork.com/reviews/albums/gas-der-lange-marsch/

I used to report broken tracks to Google when I had Google Music. I gave up. I’d get a form letter that said to reset my cookies, uninstall and reinstall the app, etc. every time.

I’d write and explain the problem was with the same song, on different machines (Windows, Linux, and Max) and both Android and iOS. Still got the same form letter back.

I’ll almost guarantee that if I search my email for the broken tracks they’d still be broken. One was Saddam-a-Go-Go by GWAR. A few seconds in it just skips ahead a few seconds from one lyric to another.

There’s one example I know of in which a ‘patch’ was released for an album: in 2008, Trent Reznor released a corrected version of the 24/96 version of The Slip (the album he had released for free through nin.com, and at the time, their own torrent tracker) due to a mastering error. The re-release replaced the existing release on nin.com.

It obviously wasn’t technically a patch, and was easy enough to do since it was self-distributed, but it’s not entirely unheard of for a major digital music release to have seen corrections.

I also remember that Kahvi Collective re-released one of their netlabel albums, "Tangents", to fix a mastering error, although they decided to actually charge for the new version due to mastering effort - their albums are usually pay-what-you-want on Bandcamp (including 0 dollar if you just want to download the FLAC version).

Kanye[0] and Drake[1] have done the same:

> The changes feature a few new alternate vocal takes and lyrics, but mainly comprise mix and production tweaks.

[0] https://www.rollingstone.com/music/music-news/kanye-west-mak...

> These weren’t overhauls that significantly altered the length or content of songs—just tweaks to the mixes of select songs

[1] https://kotaku.com/albums-are-getting-patches-now-1827376874

> I assume someone joined the invidual tracks manually in their digital audio workstation (DAW) software and botched the thing at that point. And I can safely assume that person didn’t even to bother to listen it (it is over 50 minutes long so…). Nor did anyone else working there with that release.

> I get that people make mistakes. That is natural. But WaterTower Music should have some sort of QA team in the building to prevent something like that from happening.

Former audio mastering engineer here. Audio QA is hard. You need very high attention to detail to work in a mastering role.

Many customers do not have serious QA skills and even if you get everything "right", if they miscommunicated their instructions and you faithfully fulfilled them, unfortunate things can happen.

I once had a guy screaming at me over his assistant's phone (because he'd thrown his own through the window) "I JUST THREW AWAY A HUNDRED... THOUSAND... DOLLARS!" It was their fault, though. 300,000 CDs had to be destroyed. We got paid to make a new master, with rush fees. Not how I like to make my money, but the people who mattered were thrilled with the service we provided and our ability to turn around perfectly executed masters repeatedly under intense time pressure.

With regards to the Batman Begins error, if I was the mastering engineer AND the one who did the edit, I would definitely have caught that. I listened all the way through to at least one copy of every master I ever prepared — you couldn't decline that service. That's not uncommon for mastering houses, and so I speculate that the error happened earlier in the chain, with a DAW operator who could have been an engineer or perhaps an artist with DAW skills. Somebody was tired and overbudget and couldn't muster the energy to listen to the damn thing for the ten thousandth time.

Regrettably, although it seems clear that the overlap is not what the artist intended, it would get by most listeners — it's not like it's some nasty digital glitch, the overall mood and flow isn't killed dead. At the mastering stage, there's not a guarantee that somebody would point it out, even though we'd really like to have done so and earned the gratitude of the customer. Between the mastering preparation stage and the publishing stage, probably only someone highly familiar with the piece would have picked that error out, and sometimes those people don't bring their A game for QA.

Unrelated question: it seems to me that the majority of music is recorded/mastered in a bad way. This is especially painful when I'm trying to judge a new pair of headphones. What do you think are some music albums that really hit the nail in terms of mastering?

The Batman thing is just the CD tracks bashed together into a single file. My guess is that task was dropped on some minimum wage intern, and that no-one with the title "engineer" or any serious experience with audio was involved.

Source: used to work for Peter Gabriel

My hypothesis about no-patch is that there is no incentive. You can't report an audio problem like you report a software bug. I can't imagine a situation where the kind of audio problem described by the author is show-stopping.

Audio problems are also subject to subjective judgement. What one considers a problem might go completely unnoticed by other people. Heck, the content creator may even argue that it's part of their creative output.

In some cases, there is less ambiguity. e.g. you really shouldn't hear any voice when the track says "instrumental", but does any consumer protection law care about that? Like, in Hong Kong, we have "Trade Description" laws that protect consumers against merchandises or services that don't match the advertised descriptions, although I have no idea if those laws would help here. What judge would not roll their eyes and tell you to just settle out of court? But if you settle out of court, I'd bet their laywer would try to add gag terms to the settlement. Are you really gonna allow yourself to be silenced in order to get a 10.99 USD refund?

Maybe we could create incentives if 1) audio hosting platforms actually allow listeners to report audio problems in a _systematic_, _publicly visible_ way, like you would file bug issues on GitHub, 2) leaving the problems unfixed will lead to demonetization of the content.

Also note that the performance of art seems to become perversed when (big) money becomes the incentive. If you know where to look on the web, there are groups who fansub anime and who will go out of their way to re-encode original bluray releases because they think the audio is out-of-sync, or because they think the picture of the original looks like shit in terms of colour temperature/lightness/saturation/gamma. And they seem to do it out of the passion of sharing the joy of watching anime.

While fansubbers have total freedom to decide when to do a patch or re-release, your professional audio engineer working for Warner Bros (again, hypothetical) doesn't have that freedom. If they report a problem about something already released, the first question they get asked by their boss would be "does this hurt our bottomline?" And probably overwhelmingly often, the answer is "no" because we don't have a system that builds the right incentives.

> You can't report an audio problem like you report a software bug.

Well, that depends on who you’re reporting to. How do you report a bug to MAMAA? Sure, you can click “Send feedback” all you want, but how likely is it to reach a person (as opposed to /dev/null), and then fixed? At the same time, if you tell a small artist or label about an audio problem, they might be able to do something about it.

Exactly. Part of the problem is that the media hosting platforms don't make these reports public. And there are also multiple platforms where Warner Bros (just as an example) can get their soundtracks hosted. And then such reports would get fragmented across the platforms.

It would be nice to patch metadata somehow formally.

I have the version of "Furious Angels" on which a track on the second disc is incorrectly labelled "Clubbed To Death (Instrumental Version)" but er, no, it's just an instrumental unrelated to "Clubbed To Death". I know that, but it does say right on the label that it's Clubbed To Death and so I'm torn about how to record it in the metadata for the album.

The CD that came with the original World of Warcraft collectors edition mislabeled the ironforge theme as stormwind. Is annoying.

And what about if a popular song with some controversial lyrics gets targeted? The artist or the uploader is compelled to censor something - or worse - change the message of the song to something not agreed upon by the community who made it popular. Allowing user edits to published content should always be carefully considered.

Since YT is serving data as DASH, they're probably keeping all the tracks separate on their servers. Patching audio should be a matter of replacing all the audio tracks. It wouldn't even have to be remuxed. But yeah... That's not gonna happen.

Kanye West 'patched' multiple songs on his album The Life of Pablo.

https://en.wikipedia.org/wiki/The_Life_of_Pablo#Post-release...

Reply on news.ycombinator.com