Cue sign-in screen: "Music perfectly synced to picture."

FIT · Vibeset

Cue

Music perfectly synced to picture

Upload a cut. Cue reads its pacing, mood and moments, then finds licensed music that fits, sync points included.

status
live · free trial
analysis
video + audio + LLM

Open Cue ↳ The beat-map case note ↳

There's an API behind it

Cue is a service too. Here is the contract. It's key-gated, so this is a reference rather than something you can run from here.

  • POST /v1/match Music for a video you have already sent.
  • POST /v1/videos/ingest Send a video URL. Returns a job to poll.
  • POST /v1/underscore Analyse and match in one call.
  • GET /v1/videos/{id} Is the analysis ready yet?
  • GET /v1/match/jobs/{id} Is the match ready yet?
auth
per-account API keys, scoped and revocable
retries
Idempotency-Key — a repeat is not a second charge
cost control
monthly spend cap, enforced server-side
metering
every call attributed to a key

What's underneath

  • Image to setlist

    Point at a room, get the music for it

    A photograph carries the brief faster than a paragraph does: light, crowd, time of night. This reads an image for those cues and turns them into the same structured query the catalog already understands.

    • Vision-to-audio translation
    • Mood extraction from imagery
    • Context-aware track selection

    Multimodal AI · Cross-Modal Ranking · LLM Orchestration

  • Video-music matcher

    Music that lands on the cut

    Two analyses, then one decision. The video side finds shot boundaries and how hard each cut lands. The audio side finds the beat grid, downbeats and section changes. Alignment then scores every candidate offset for how many cuts fall on strong beats, and picks the offset a human editor would have found by ear.

    • Shot-boundary detection
    • Beat and downbeat tracking
    • Section-boundary detection
    • Cut-to-beat alignment scoring
    • Sync-point selection

    Applied Computer Vision · Audio Signal Processing · Cloud Workflows · Human-Centered AI