MIR metadata is largely assigned by expert listeners, motivating automation as music creation accelerates

Evidence

The driver of MIR systems is metadata. Music tracks are analysed by different systems, and metadata that describes the tracks from different aspects is collected. The simplest type of metadata is factual metadata, which describes the objective truths about a track, such as the performer, the name of the piece, and so on. The factual metadata can also be accompanied by cultural metadata, which describes properties like mood, emotion, genre, style, and so forth. Most often, these attributes are specified to the database by expert listeners, implying that each track added to a database should first be listened to. The fast pace of music creation motivates the use of automatic systems for these tasks. — Ville Pulkki, Matti Karjalainen, Communication Acoustics- An…, 2580

Categorization

• SEED: Angle: the hidden human labor — armies of expert listeners — behind commercial song-recognition services, and why automation is eroding that labor force.
• QUESTION: Who employs these expert listeners, and what do they actually do across a typical work week?