Trillion

Promo reel renderer (word-synced karaoke captions)

Assemble a finished vertical promo reel from a cut plan, a voiceover, and word timings. Cuts each segment from its source clip with accurate seeking, strips all source-clip audio, scales and crops to 1080x1920 at 30fps, concatenates the segments, denoises the voiceover, burns flat word-by-word karaoke captions (bold grotesque, all lowercase, all white, no outline or box, soft dark shadow only, spoken words bright white and upcoming words dimmed, no animation, fixed layout, mid-frame), then muxes H.264 video with AAC audio trimmed to exactly the voiceover duration. The calling AI does creative direction (footage selection, beat mapping, transcription); this tool does the deterministic render in one call.

What it asks for

  • voiceover (Optional, file): Voiceover audio file identity (1f: then 64 hex), as find_file or list_files returns it. Its duration sets the final video length.
  • clips (Optional, list): Source clip file identities, referenced by index from segments.
  • segments (Optional, list): Cut plan, laid end to end. Each clip window should show one complete action.
  • selftest (Optional, yes or no): Internal check mode: generate synthetic test media inside the container instead of reading file inputs. Used by the sample-run check.
  • words (Optional, list): Word timings for the captions, e.g. from a speech-to-text transcription tool. Rendered lowercase.

Use it from your AI

Connect the AI you already use to Trillion once. Then ask in your own words; your AI finds this tool by what it does and runs it.

Say to your AI: “Use Trillion to assemble a finished vertical promo reel from a cut plan, a voiceover, and word timings. Cuts each segment from its source clip with accurate seeking, strips all source-clip audio, scales and crops to 1080x1920 at 30fps, concatenates the segments, denoises the voiceover, burns flat word-by-word karaoke captions (bold grotesque, all lowercase, all white, no outline or box, soft dark shadow only, spoken words bright white and upcoming words dimmed, no animation, fixed layout, mid-frame), then muxes H.264 video with AAC audio trimmed to exactly the voiceover duration. The calling AI does creative direction (footage selection, beat mapping, transcription); this tool does the deterministic render in one call”

What your AI calls: run_tool {"tool_id":"tool:promo-reel-render"}