02. How VeeJay sees music¶
The concepts every other chapter relies on, explained without any technology. The precise definitions of each field are in 9. The database, and the words are defined in Appendix A.
2.1 Artists, groups and their relations¶
An artist is a person or a group that performs or is credited. A group is not the same record as its leader: a group's members, a backing band and a side project are each an artist of their own, joined by relations.
| Relation | Meaning |
|---|---|
| Member of | A person belongs to a group |
| Backing band of | A band that plays behind a named artist (the E Street Band behind Bruce Springsteen) |
| Side project of | A group that one artist or band runs beside the main one |
| Alias of | The same artist under another name |
A relation is stored in one direction only, and the opposite direction is worked out from it. An artist is never related to itself.
The things that are true of a person are kept apart from the things that are true of a group. A person is Living or Deceased. A group is Active or Dissolved. Each is refused on the other kind.
Names. An artist's name keeps its leading "The" when the band uses it (The Beach Boys, The Rolling Stones). Names that differ only by "The", capitals or punctuation are one artist. In file names the name is written in plain ASCII, so Beyoncé becomes Beyonce. A name is never changed to suit a source: an artist is spelled the way the database says.
2.2 Songs, albums and collections¶
A song is one recording by one main performer. A cover by another artist is a different song, because the performer counts. A film song covered by two artists is two songs.
A song is filed under its original studio album: the release of type album with the earliest first release. A later compilation, remaster or re-release never moves it. A live or remix recording is filed under the original studio album as well, with a disambiguation to tell it apart. A song that was never on a studio album is filed as a single: the album is the song's own title and its position is 1-01.
A song that comes from a film, a television show or a video game belongs to a collection. The collection takes the place of the album artist, and the performer is shown as the track artist. These songs are filed apart from the performer, in their own top-level folders. A cover or fan edit that is not from the film, show or game stays under its performer.
Every song keeps the facts that belong to it, once: its title, its album, its disc and track, and the identifiers that point to it in outside systems. Two videos belong to one song only when all of those agree.
2.3 Videos and their roles¶
A video is one YouTube video, and every video belongs to one song. Each video has a role:
- Primary. The one video kept as the song's main video. Every song has exactly one Primary.
- Alternate. Every other video of the song. An Alternate has no audio file of its own.
Beyond its role, a video is described by separate choices, each with a short closed list. A video always has an answer for every one of them, and none has an "Other" value.
| Choice | Values |
|---|---|
| Origin | Original, Cover |
| Performance | Studio, Live |
| Version | Original, Remix, Edit, Instrumental, Acoustic, Remastered, Orchestral |
| Video type | Music Video, Lyric, Visualizer, Live Footage, Montage, Documentary |
| Content | Clean, Explicit, None |
| Uploader | Artist Or Label, Broadcaster, Auto-Generated, Fan |
The video type describes the form of the video only. Who uploaded it is a separate question: a fan's re-upload of an official music video is still the type Music Video, and its uploader is Fan. A disambiguation holds the words from the video's own YouTube title that tell it apart from other videos of the same song, such as a mix name or a live venue. It is never invented.
2.4 Audio and files¶
A song has one audio file, and it belongs to the song's Primary video. The audio is the song as recorded, which a video can differ from: a video may be a cut, an extension or a medley, so the two are never matched by length. The audio file has exactly the name of the Primary video, apart from the extension.
A file is the physical copy of a video or of audio. The database records what is true of each file, measured from the file itself: its format, its size, its content hash and its last-modified time. Those facts are what let VeeJay tell whether two copies on two devices are the same.
2.5 Identity and trust¶
Before VeeJay writes a fact about an artist, it has to be sure who the artist is. A name alone is not enough: many artists share a name, and a search result can be a namesake, a fan page or a cover band.
VeeJay therefore treats every fact it finds as a candidate until it is verified. A fact is verified when its page has been examined and tied to the artist, starting from something that is certain (an anchor) and following the links that verified pages publish. Only verified facts are shown to you, only they count toward an artist's identity score, and only they may fill a field.
The way this works is the subject of 7. Research and verification.