Introducing Onset Engine
We’re excited to announce the public launch of Onset Engine — a local, AI-powered video editing and mixing engine that automatically creates beat-synced music video compilations.
What is Onset Engine?
Onset Engine analyzes your video footage using OpenCLIP ViT-L/14 (a state-of-the-art vision-language model), classifies clips by visual content, and then assembles them into coherent, music-driven edits. Everything runs locally on your existing hardware — no cloud uploads, no subscriptions, no data leaving your machine. Works on CPU, accelerated with NVIDIA GPU.
Drop a folder of clips, choose a music track, and Onset Engine handles the rest: beat detection via librosa, visual embeddings via OpenCLIP, intelligent clip-to-beat mapping, and timeline assembly. The output is a fully rendered video — or an OpenTimelineIO file you can open in Premiere Pro or DaVinci Resolve for further refinement.
Three Ways to Create
- Autopilot — Describe your vision, and the AI builds a complete timeline in under a minute. Best for batch workflows where you need volume: social content, ad variations, event recaps.
- DJ Mode — Live-mix your video library with zero-latency, hardware-decoded MPV playback. Keyboard-driven clip switching synced to audio analysis in real time. No MIDI controller required.
- Studio Mode — Full NLE-style control with timeline, inspector, library browser, and cinematic VFX presets. For when you want AI assistance but manual creative control over every cut.
Why Local AI?
We believe your creative tools shouldn’t require uploading gigabytes of footage to someone else’s server. Cloud-based AI editors introduce three problems that local processing eliminates:
Privacy. Your footage never leaves your machine. No cloud scanning, no training on your content, no terms-of-service surprises. For creators working with NDAs, unreleased content, or private footage, this isn’t optional — it’s a requirement.
Speed. No upload queues, no shared GPU contention, no waiting for cloud rendering slots. Your local GPU processes footage as fast as your hardware allows. A typical 3-minute edit renders in under 60 seconds on an RTX 3060.
Cost. Cloud editors charge per minute of processed video or per month of access. Onset Engine is a one-time purchase. Process unlimited footage, render unlimited exports, forever.
Onset Engine runs OpenCLIP inference directly on your NVIDIA GPU, performs audio analysis with librosa on your CPU, and stores everything in a local SQLite database with vector search.
Zero telemetry. Zero tracking. Zero cloud dependency.
What’s Included
- 9 cinematic style presets (Standard, Aggressive, Sensual, Raw, Prestige, Trailer, AMV, Hypnosis, VHS)
- Smart Crop face tracking for automatic 16:9 to 9:16 reframing
- Memory-safe chunked rendering pipeline with multithreaded encoding
- Triptych mode for TikTok/Reels 3-panel exports
- OpenTimelineIO export for Premiere Pro and DaVinci Resolve hand-off
- Pointer-only ingest for massive libraries (zero extra disk space)
- Community-driven JSON driver system for content-agnostic AI direction
- NVIDIA NVENC hardware-accelerated encoding
Pricing
Onset Engine is a one-time purchase with lifetime v1.x updates. We’re running a 50% launch discount:
- Free Demo — Prove it runs on your hardware before you buy
- Core ($29.50) — Unlimited DJ Mode, 1080p rendering, full Autopilot
- Studio ($59.50) — 4K rendering, Smart Crop, Triptych, OTIO export, premium presets
No subscriptions. No monthly fees. No watermarks on exports. View pricing →
Get Started
Download the free Demo and see it work on your footage. No GPU required for the demo — just your existing Windows machine.