# Versegen.AI

> Versegen is a local-first AI music video editor that runs every
> model on your machine — no cloud, no API bills, no content filter.

Drop in a song + a folder of clips, get back ten beat-synced cut
variants, pick the one you like, export to your NLE (Final Cut /
Premiere / DaVinci / EDL).  All AI processing runs locally on
macOS (Apple Silicon), so unreleased material never leaves your
device.

## Quickstart

- [Download](https://versegen.ai/download): free macOS .dmg, signed
  + notarized by Versefactory.AI, Inc.
- [How it works](https://versegen.ai/#how-it-works): the six-stage
  local pipeline.
- [Pricing](https://versegen.ai/pricing): free for full Layer 1 + AI
  features; Pro tier coming when paid features launch.
- [About](https://versegen.ai/about): the company + person behind
  Versegen.AI.
- [Blog RSS feed](https://versegen.ai/blog/feed.xml): new posts in
  RSS 2.0; ingest pipelines can subscribe here.

## Articles

- [How to make an AI music video in 2026 (free, local, no cloud)](https://versegen.ai/blog/ai-music-video-2026): A practical 2026 guide to making music videos with AI: source footage, beat-synced editing, hybrid generative shots, and exporting to your NLE.  Free + local-first workflow.
- [Local AI vs cloud AI for music video editing: an honest 2026 comparison](https://versegen.ai/blog/local-vs-cloud-ai-mv): Cost, privacy, control, and speed compared between local and cloud AI music-video editors.  When each wins, and why most pros use both.
- [MV自動カット編集ツールの仕組み：ローカルAIで音楽動画を自動生成する](https://versegen.ai/blog/mv-auto-editing-ja): 音楽のビートと歌詞に合わせて動画素材を自動配置する仕組みを、ローカルAIワークフローの実例とともに解説。MV制作AI/自動カット編集の選び方。
- [歌詞字幕付きMVを自動生成する：日本語歌唱でも実用なローカルAIワークフロー](https://versegen.ai/blog/lyrics-video-japanese-ja): 歌詞テキストを正として、AI（Whisper・Demucs）をタイミングガイドとして使う実践的な歌詞字幕MV制作ワークフロー。日本語歌唱で精度を上げるコツも。

## Technology

- Tauri v2 (Rust shell) + Svelte 5 frontend
- Python engine (librosa + PySceneDetect + OpenCV + MediaPipe for
  Layer 1; PyTorch + Whisper-class ASR + Demucs + OpenCLIP +
  ONNX Runtime for Layer 2 AI)
- macOS hardware acceleration via VideoToolbox + Metal Performance
  Shaders
- Outputs FCPXML / Premiere XML / DaVinci XML / EDL

## Open source

- [GitHub repository](https://github.com/makotunes/automad)
- License: MIT for source code; bundled ML models retain their
  individual licenses (mostly MIT / Apache, some research-only)

## Company

Built by Versefactory.AI, Inc. (Tokyo, Japan).