MusicBrainz and its software companion, Picard, are absolute blessings when it comes to micromanaging a music library in this day and age. It can't find _everything_ I have due to entire artists appearing and disappearing between the closure of Napster and the creation of YouTube, but it gets me to that 95% CI that puts me at ease and lets me enjoy my collection. The fact it's global instead of regional (like a lot of automated DB lookups that cannot find my JP/ZA/DE/FR/etc albums here in AMER) is also a big notch in its belt.
Which reminds me, it's about time for the yearly re-scan and re-tag.
My fondest memory of Picard is the time it cheekily rewrote all my Russian music (Rachmaninoff and all that) into actual Russian with their alphabet, and how I never could find any of it after that.
That's partially why I started learning Japanese...Spotify used to use English translations of JP song titles, or romanizations at the least, but one day a few years ago I opened one of my playlists and found them all in kana and kanji.
Or we could expand the spec to contain alternate fields for a given tag. That way the default tag is in the origin language of the content, but alternate tags could surface localizations into search or display.
The only downside is dusting off the ID3 spec, amending it, and then somehow coercing everyone who thought music tagging was “done” to update their software to support it.
The software part of the solution is easy-peasy. The political part, infinitely harder.
To be fair, we (myself included) look at you weird because the ZFS folks haven’t done a great job at extolling its virtues in a way that the average consumer can comprehend. It’s important to acknowledge that the vast majority of consumers simply don’t care about the underlying file system or its capabilities, even if they would benefit immensely from the features of something like ZFS.
It’s a problem of communications as much as it is vendor support. If you get the consumers wanting it, vendors will switch to it. Instead, right now we get companies like Apple reinventing the wheel via their own proprietary FS just to adopt some features of ZFS, rather than _just supporting and contributing to ZFS_.
Which means I get to anxiously eye the Synology with btrfs instead of resting easy with ZFS. It’s better than ext4, but I also know I’m the exception rather than the rule.
Hah! Yeah, I whoopsied on some JP soundtracks a few times in the past as well. While world-ending the first time it happened, now I just view it as a glib, "oh _no_, now I have to relisten to the _whole album_ and figure out which track is which!"
It really is an incredible resource, and Picard is a wonderful app. Very satisfying getting a library properly tagged! Takes a while, but totally worth it. Shoutout to ListenBrainz as well, their scrobbling service: https://listenbrainz.org/
If linux people are interested in using listenbrainz and can't find a player with plugin for it (usually media players ignore it in favour of last.fm) I developed a generic scrobbler daemon that works well with it (also with libre.fm and last.fm): https://github.com/mariusor/mpris-scrobbler#authenticate-to-...
MPRIS is an under-appreciated standard and a great example of how different technologies on the open desktop should integrate with one another.
For example, there's <https://github.com/altdesktop/playerctl>, which can control any MPRIS-compatible client: VLC, web browsers, different music players, mpd (with a plugin), etc.
MPRIS can also be used to e.g. allow controlling playback without unlocking the screen - the screen locker itself doesn't need elaborate support, something as simple as slock could be hacked to recognise a key combo and call out to playerctl; more fancy login managers/lockers could talk DBUS/MPRIS directly and even e.g. display album artwork.
Unfortunately, the standard doesn't seem to specify a simple way to indicate whether video is being played back, which could be a saner way to inform the screen locker; the actual protocol to directly inhibit locking is unfortunately a little bit insane.
Funny you mentioned playerctl. Due to the fact that over time its scope expanded from a simple MPRIS based control application, into a full blown library with various unneeded (by me) features. Therefore as a starting point for the scrobbler daemon I wrote the simpler mpris-ctl: https://github.com/mariusor/mpris-ctl, which does just the control and meta-data output parts that playerctl has, and only depends on libdbus.
And concerning your last point, are you advising someone that built at least one MPRIS tool to read the spec? :P
> And concerning your last point, are you advising someone that built at least one MPRIS tool to read the spec? :P
The whole comment was for the general HN audience who might not be aware of MPRIS.
The general problem I keep seeing is silos; we've had the XDG standards/specs for decades, yet keep reinventing new, stupid, and broken ways to do things. Under X11 it was normal that every window got a border; under Wayland, GNOME doesn't want to adopt a protocol where a window can ask the compositor to paint a basic border, instead tells SDL to link against libadwaita. XScreenSaver has been telling apps to do "while sleep 60; do xscreensaver-command --deactivate; done" which worked well enough for decades; to "modernise" that, we have a DBus protocol which makes an app grab a cookie, and if it fails to return it (e.g. because it crashed) - the screensaver remains inhibited forever. Trying to use (or improve upon) a FOSS desktop is death by a thousand papercuts.
afaik, funkwhale and navidrome have overlapping usecases, but don't have the same goals. funkwhale with its federation features wants to be more than a web player for your library
Yes, although I don't use the federation features from Funkwhale. I just enjoy its clean and awesome interface. Funkwhale is a joy to use and fits very well my music habits (listening to selected albums instead of all over the place - though this is supported too, of course).
I pair Funkwhale with Snapcast-Mopidy-Iris stack, in case I want to have multi-room audio or listen to other sources, beyond my own library.
I wrote about the history of MusicBrainz for the EFF in 2021, as part of a series looking at how "public interest internet" (ie commons-based work) survives outside of the constant coverage and mergers of bigger, more commercial projects:
For context, Acoustid is a MusicBrainz-adjacent service for figuring out the MusicBrainz ID of a song based on the sonic content alone, even if it has been distorted or compressed. Chromaprint is the logic for computing an Acoustid given a song as input.
Oh, really cool to see my old code implemented in a different language. There is one other reimplementation, I think in C#, but that's much much more verbose. I wanted to experiment with Zig for a long time, this seems like a good opportunity to get at least familiar with it.
Thanks for Acoustid! It's a bastion of the old guard free and open source software movement. I love that you figured out how to run the service for free for a long sustained amount of time by rate limiting it to keep the costs reasonable.
Looking forward to integrating with acoustid.org in my hobby music player side project!
MusicBrainz is great! Every now and then I'll get an email notification that someone has updated an attribute on some obscure local band that I put on there.
ca. 2017 I undertook the considerable task of building a GraphQL interface to MusicBrainz, to support a side project of mine. This was a great experience for learning the breadth of MusicBrainz and how to design things with GraphQL. Sometimes I look at the documentation generated from the resulting schema and wonder when I ever had that much time: https://github.com/exogen/graphbrainz/blob/master/docs/types...
Reposts are fine after a year or so! (see https://news.ycombinator.com/newsfaq.html) and historical material is always welcome here. Links to past threads are just to satisfy extra-curious readers.
(Whether the current submission was a good repost or not is a separate question of course.)
It’s also used by other software. My music player MusicBee (RIP MediaMonkey) and the CD ripper ExactAudioCopy [1] both use MusicBrainz for lookup. I have also occasionally entered albums there myself when the CD was still too new.
Quaint memories of a carefully curated library of music.
iTunes was my pride, every file properly tagged with beautiful music. Spending whole evenings doing nothing but scrolling through my music and listening to gems I had found.
iTunes Match sounded great. After all Apple was dedicated to Music (Liberal Arts & Technology!) and I would finally have the ability to have my *entire* music collection with me on the road.
Hey - iTunes replaced a version of a song (Max Richter On the Nature of Daylight) with a slightly different version (Weird vocalized version from a Leonardo di Caprio Soundtrack) and for some reason even *my* copy on my Mac is now replaced by this one. Lot's of angry e-mails with Support.
They bought Beats by Dr. Dre - a bit odd and that Jimmy Iovine guy starts showing up at WWDC. Odd. He's not so sympa.
So now iTunes becomes Apple Music and you compete with Spotify eh?! Well if you make it just right and don't mess up MY music and combine it with streaming - that could be great! Plus ONLY Apple could pull this off.
Hey, somehow I can't find my Music anymore and some of my demo bootlegs are gone.
Apple Music has disappeared from WWDC, because we all complained that it isn't relevant to developers maybe?! But why doesn't it show up in the iPhone event anymore?
Okay - I give up. Streaming it is. Lost my old collection of digital music. It joins my abandoned tapes and cds.
Spotify is so much better wow! I wish I could integrate it better with iOS. Why does it always loose the now playing screen.
The problem with those tv and movie databases are that they’re owned by Roku and you can’t download any data dumps…you’re also relying on the good graces of a corporation that can cut you off whenever it feels like it. That was the case with IMDb and I’m still upset about that :/
I think the only way to avoid a CDDB is to make the entire database downloadable so that someone, or a group of someones, is able to re-create it in case of disaster.
I think in the case of "great, open service" selling, the new buyer can just change the licensing, and you're out of luck. If there is a license that compels them to maintain some level of availability, you can trust them to cut it down to the absolute minimum, including rate limiting access down to a level of pointlessness.
Their editor is such a nice thing to use. I add a lot of stuff weekly to MusicBrainz, and especially the media editor where you just paste the tracks somehow parsed, and it will create the media with the track numbers, names, lengths and artists. If you do this often enough, you appreciate the efficiency.
MusicBrainz/beets/flac/plex is my choice for music and it beats Spotify with the selection and quality.
MusicBrainz is great. I stumbled upon it about a year ago while trying to figure out how various open-source desktop music apps populate track info... and was delighted to find that some random Internet stranger had helpfully catalogued all my youtube videos years prior!
Way back in 2007-2009, a bunch of Amazon Scotland developers and I were tasked to make a ripoff of MusicBrainz, because Amazon wanted another IMDb to sell ads on. (The brief was literally "make us IMDb but in another vertical"!) It felt like a doomed project from the start, and I kinda hated the idea of ripping off an open project and trying to steal their users, but hey, it paid the bills.
We imported the MusicBrainz database, spent months hotly debating about the data model (releases versus editions, mostly), more months preparing the site for an influx of traffic (our goal for launch was 200 hits per second), and yet more months going from a team-internal design (Amazon at that time believed that engineers were perfectly good designers, which I think says a lot about the 2007 Amazon site design!) to an execrable bright-yellow-and-red "tequila sunrise" design by an internal Amazon designer from Seattle, to finally something attractive once we hired our own designer.
Then after two years of hard work we launched and hilariously sunk without trace, except for one dedicated user who we reckoned had a big paper music encyclopedia and who just kept on trucking, adding basic info, for months. We sent her a T-shirt.
SoundUnwound closed quietly a couple of years after launch. MusicBrainz is still here, and for that I'm very glad.
I would like to sincerely apologize that the autotagger in beets is so fussy. It asks you a lot of complicated questions, insecurely asking that you verify nearly every assumption it makes. This means importing and correcting the tags for a large library can be an endless, tedious process. I’m sorry for this.
Maybe it will help to think of it as a tradeoff. By carefully examining every album you own, you get to become more familiar with your library, its extent, its variation, and its quirks. People used to spend hours lovingly sorting and resorting their shelves of LPs. In the iTunes age, many of us toss our music into a heap and forget about it. This is great for some people. But there’s value in intimate, complete familiarity with your collection. So instead of a chore, try thinking of correcting tags as quality time with your music collection. That’s what I do.
MusicBrainz (and Picard) are amazing tools. I tag all my music with it, and try and contribute to the database when I can (normally for doujinshi stuff)
Is this the same MusicBrainz from the early 2000s? I'm sure it's gotten better, but at the time it completely trashed my music library. It broke my trust in any automatic tagging software for decades. I'm still not over it and if I used a software helper, I will go one album at a time, and validate against another source, and change some things manually along with it.
To be fair - you trashed your music library. I used Picard many times. I just never pushed a button labelled "go ahead and do whatever you want to all my files".
The only issue I have with MusicBrainz is that they generally use correct grammar/spelling/style over what the artist uses themself. For example, "accapella" -> "a cappella" or "Remix" -> "remix".
If you ask yourself in what style something should be entered into MusicBrainz, the following principles apply:
Follow Artist intent.
If no definite proof can be found for the correct spelling/punctuation, the most common version should be used.
Follow the style guidelines.
"remix" would be an example of extra title information as outlined here https://musicbrainz.org/doc/Style/Titles. "Titles and subtitles of mixes/versions are formatted according to the appropriate language's guidelines; the other parts of this extra information should be in lower case except for words that would normally be capitalized in the language."
"acapella" would technically be considered a spelling error, which is something that we correct.
The artist "intent" is often misguided. CD-Text usually is assumed to be CP1252 99% of the time where I frequently see punctuation misused for "style" purposes. My biggest peeve is right single quote being used as a fancy apostrophe when that is an inappropriate substitution.
Are you claiming that ’ (Unicode code point 002019) is not suitable as an apostrophe? In that case, what is? The classic ASCII 000027? Isn’t that one straight? (As opposed to ’, which is slanted correctly.)
Or are you perhaps thinking of ´ (Unicode code point 0000B4, ACUTE ACCENT)? Yes, that one is ugly and should not be used as either an apostrophe or a right quote.
This does not work, because the “right-skewed” apostrophe is not appropriate for every case of “straight” apostrophe placement.
However, making every “skewed” apostrophe “straight” is very easy for a computer to do (e.g. MusicBrainz Picard ships with a setting for this). This is why the MB guidelines say to prefer correct typographical apostrophes.
Note that they are preferred and not required, so you can still enter straight ones, but another editor may update them. I honestly don’t see the problem, but I am very pedantic about other things so I guess I can imagine how seeing the curlies could rub you the wrong way!
The reference glyph for ' (code point 000027) <https://www.unicode.org/charts/PDF/U0000.pdf> shows a straight rendering. Any Unicode font rendering it differently can reasonably be said to be rendering it incorrectly.
Interestingly, this is not the case with Japanese artists.[1]
I've viewed/edited/added many Japanese entries on MB, and the names and spellings for them are usually always spelled exactly as they appear on the official release.
I noticed that with the artist names. However, I think that comes partially from the albums themselves if the info is in the inserts. It is one thing I wish picard would do better with artists and aliases. Then give you an option/plugin to pick the one you like. Maybe it does. Havent, touched it in a few years.
This is just incredible. I listen to tonnes of old Vaporwave playlists and I've been desperately looking for somewhere to record who-wrote-what who-sampled-whom and so on. Mostly so I could find the original work and check it out.
Used it on and off since the early days, and it is ok. Last time I tried, I hoped to automate the tagging of all my music collection, but I had to give up and revert back to manually edit the tags using mp3tag (https://www.mp3tag.de/en/index.html), as there were too many mismatches (either the album were wrong, or the title of the songs for the most part).
I'm cobbling together something that includes links to musicbrainz and other encyclopaedia https://ont.fyi/entity/Q15920 just search for your artist (or anything really).
It's very WIP and rough around the edges, but feedback welcome (it is mostly a wikidata wrapper at the moment, but that isn't the intended goal).
Minor thingy: it labels an organ as a wind instrument, which is unusual. A piano is labelled string instrument, equally unusual. And it says piano quartet consists of a piano, a violin, a viola and a cello, which is not unusual, but definitely not always the case.
And P.I. Tchaikovsky has different spellings: often it's in Cyrrilic, but it also occurs in Latin. His "end area", a weird term for the place of death, however is consequently called "Sankt-Peterburg".
Don't treat that info as consistent, complete or correct.
First time browsing this website, and it seems like an amazing project!
I particularly like that they link to places where you can purchase a song or album. As someone who's considering streaming my music from a local server, those links could come in real handy!
I have struggled with how to get lodged a correction to a CD match which I have in my hands, and which I found digitised track lists for in (of all places) a chinese university music library collection, but which doesn't match the ISBN and other tagging in the musicbrainz DB. It's like they take the "best" match from their US/EU pressings and if you live in asia and have an alternate market and a slightly different combination in the pressing, you're SOL. I am in the asia-pacific. I do have this recording. it does not match what they and discogs have for this CD.
(its a recording of the bach double violin concerto over 2 CDs by Zukerman/Stern/Perlman with other materials, which seem to vary by pressing)
The tooling is great for the mainline. If you have anything on the margins, it can make your collection look very strange. Maybe I missed how the document how to lodge corrections.
Really? I blame the music industry for re-using the barcodes and identity info behind different pressings.
Assuming it's the first link though, here you can already see two versions of this release, a US pressing, and a UK pressing.
To add another, simply look at the right sidebar - there's a link "Add Release" in the editing section (if you're logged in). The wizard should (hopefully) be fairly straightforward after that.
It's neither of these, the listings for these don't include Neville Mariner. It is the same CBS master works branding from Columbia, but dated 1985. It has 15 tracks. On 2 discs.
I will try to add this one.
BTW this ebay item is the disc. I gave tried to match it using musicbrainz search and nothing comes up so I really do think it's another previously uncatalogued compilation. The outer box art says CBS. The CDs themselves say Sony, with a copyright date of 1985 on the cds which I believe is before Sony bought Colombia, which is how CBS masterworks were being branded. The notebook says Columbia (c) Sony 1991. This is usual. This is confusing. It's an Australian manufactured cd burned by disctronics. So it's US and British recordings, owned by american-canadian recording companies, sold on to Japan, and minted in Australia with branding from all 3.
Really like them in general and did my best to correctly submit information about media in my collection.
Then everything went bad in an unexpected way. I was using Firefox on a Mac to interact with the site. Apparently they added a bunch of javascipt to the site that would automatically fill some fields that I didn't even understand, but the scripts did not execute correctly or even at all on my system. Later I got an angry message from someone at the site telling me that my submissions were unacceptable, though it isn't clear if some or all data was backed out or what. Given the timing it seems they were targeting IE6 on Windows and could not imagine any other kind of user. They could have somehow checked script function on the incoming data, and apparently eventually did that, but they got angry at me instead even though I put quite a bit of effort into correctness.
Have stayed away since, but still take this as a warning about the kinds of things that can go wrong even with the most well meaning contributed data type site. And unfortunately this kind of music data continues to have relatively low value to most audiences.
I can't speak to your experience, and I'm sorry you had a bad one, but just to clarify for anyone who is not familiar with the site, the review and voting on the data submissions is done by other users.
Perhaps you ran into someone who was particularly passionate about that artist, or having a particularly bad day, but I'd be surprised if it was any of the site's developers.
(PS: If a submission does not get any negative feedback, it gets auto-accepted after some time in the queue. If it gets negative feedback, it will be rejected, and the author can modify and resubmit it.)
Wow, a bunch of neg votes because I tried to summarize my experience. I should just hide instead?
Sure, I may have misunderstood that part. So not a dev. Which means we have what?
Get excited about a web site that implements a cool idea. Read the instructions carefully. Submit what appears to be correct and relevant information. Receive a stream of psycho rage emails from various people over a period of months. Sorry I didn't understand something about the input process or how to correctly respond to the rage emails. My central point of being careful interacting with contributed data sites stands. Most people do not go into such things expecting the result might be a great deal of rage mail with no clear way to respond.
Which reminds me, it's about time for the yearly re-scan and re-tag.