Sync metadata supervisors actually search for
If your track tags are vague, you don't exist when supervisors filter the catalog. Here's how to write metadata that lands you in pitches.
Supervisors don't sit at midnight scrolling 60-minute playlists. They have a brief, a deadline, and a search bar. Most of them search by use-case, emotion, tempo range, and scene fit — in that order.
The use-case-first rule
A supervisor cutting a coming-of-age indie drama will type "emotional climax" or "montage" before they type a genre. If your track is built for a love-scene crescendo, say so. Underscore's auto-extraction handles most of this — but if you're filling sync_metadata.narrative.scene_fits manually, lean on the supervisor's vocabulary, not the artist's.
Primary emotion ≠ vibe word
"Vibey" doesn't appear in any catalog filter. "Longing," "Defiance," "Joy," "Empowerment" do. Pick the one feeling that drives the track and tag it as primary_emotion. Save the modifiers for the description field.
BPM and key matter more than you think
If a supervisor needs a 96-100 BPM track in a minor key for a slow-burn montage, your track at 134 BPM in D major will never surface — no matter how good it is. Underscore auto-detects BPM and key from the audio, but verify it after upload. The Whisper + GPT pipeline is good, not perfect.
Lyrics tagged, themes extracted
Every Underscore upload runs through Whisper for transcription and GPT-4o for theme extraction. Heartbreak, identity, defiance, redemption — these are the words supervisors search. Your job is to write songs about real things. Our job is to make sure those real things show up in the search index.