This is not true. Given two songs A, B, an informed human can tell whether A and B are the same song or not. Now you’ve created a UNIQUE constraint on your data, so you have a unique identifier.
This is not true. The article gives the great examples of a song through its lifecycle: covered, remixed, remastered, compiled on a greatest hits record, included on a soundtrack, etc. Same song, sometimes. Same recording, sometimes. Different database entries, always.
To the extent that "an informed human" and "the same song" have meaningful definitions of "informed" and "same", you've just recreated the problem.
All this song data has been standardized and revamped multiple times and these standards have been implemented in mp3 players to decode and display info about that to which you are listening. Waitaminute -- mp3 players can play lots of sounds, including non-song sounds, e.g. speeches, air-checks, phone messages, commentaries, advertisements, radio dramas or comedies, bird-calls of birds, bird-calls of hunters, podcasts, phone-taps, recipes, soundtracks for walking or driving tours, audiobooks of diverse genres, etc, etc, which are quite likely to be acceptably identified and described by various attributes other than those applicable to songs. However, humans find it easier to put square pegs in round holes than to rebilk Romeo in a day.
quesera|2 years ago
To the extent that "an informed human" and "the same song" have meaningful definitions of "informed" and "same", you've just recreated the problem.
lucas_membrane|2 years ago