Three Modes of Search

People come to a music corpus knowing what they want in very different ways. A music supervisor with a brief might have a precise sound in mind. A composer might have a reference track and want something like it but darker. A director might have a scene and no idea what music it needs. A single search field treats all three the same; CORPUS treats them as different problems.

One semantic layer, three ways in. You never pick between them; the CMI chat reads what you bring and routes it, then holds the session in that mode so follow-ups refine rather than restart.

Precision: name the sound

When you can describe the sound at a supervisor's level ("dark jazz, no vocals, only piano, slow") the brief becomes an explicit query, matched semantically against the track descriptions rather than filtered on tags. → Precision Search

Audio similarity: match a reference

When you already have the right music and just can't use it, an uploaded or linked track is compared audio to audio for the closest matches, which you can then steer by tempo, key, or instrumentation. → Similarity Search

Narrative: describe a scene

When you have a moment, not a sound, the scene is read into three dramaturgically distinct directions: different readings of the same moment rather than variations of one, because the right music for a scene is an artistic choice, not a single answer. → Narrative Search

Behind these sits a fourth path: name a single track and CMI understands it first, then grounds a search on it.

What this resolves against

All three modes query the same annotation layer; the dimensions are in What CORPUS Knows About Each Track. The step-by-step how-to for each, with examples and screenshots, is the CMI manual.