Skip to main content

Multi-Search

When one request carries several music needs (on purpose or not), CMI lists them and lets you pick before it searches. It is a clarification step, not a separate engine.

Several needs in one message

Hand CMI a brief or a few lines with more than one cue ("a tense opening chase, a tender mid-film love scene, and triumphant end credits"), and instead of guessing which one you mean, it numbers the cues it found, suggests one to start with, and waits for you to choose.

Upload a whole script or briefing

You can also hand over the entire document. CMI reads Word (.docx), PDF, and Markdown, breaks the text apart, and returns a list of the decisive musical moments: for a long script that is the key moments rather than every last one; for shorter material you can rely on it being the full set.

A screenplay, '2012.docx', uploaded to CMI. The system replies 'This script contains many distinct musical moments. Which one are you looking for?' and lists six, each with a short description: the opening and closing bookends (suggested), the Yellowstone eruption, the Las Vegas escape, the cruise ship disaster, the ark boarding chaos, and the final reunion. The user picks the third; CMI then runs a Narrative Search on the Las Vegas escape and returns three directions (Digital Decay, Ghost in the Machine, Frantic Recovery) with tracks.
Drop in a whole screenplay and CMI pulls out the decisive musical moments; pick one and it routes that moment into the right mode. Here, the Las Vegas escape becomes a Narrative Search.

One pick, the right mode

Reply with a number and CMI takes that moment straight into the mode that fits it (Precision if it is a nameable sound, Narrative if it is a scene) with no second round of routing. Each moment keeps its own context, and several searches can run in the same chat session, so you can work through a whole script one moment at a time without starting over.

How CMI decides between routing and asking is in Conversational Dialog.