Open-source video cloning skill turns reference videos into editable, locally rendered projects
oran_ge · x · 2026-09-24
The author open-sourced the skill behind the earlier cloning demo, "listenhub-hypit" (MIT licensed), which turns reference videos, briefs, or your own footage into a still-editable video project.
How it works:
- ListenHub generates images, video, voiceover, voice cloning, and music; hypit handles local timeline editing, word-level subtitle alignment, and rendering, delivering MP4 plus project files
- Use cases: adapting a reference video (swap product/character/voiceover while keeping BGM), character dialog, pet videos, podcasts
- Reusing existing footage and local audio work is free; cloud calls are quoted before execution
- Requires ListenHub API key, Node.js 22.15+, ffmpeg, uv
The author lists honest limits: no guaranteed exact replication, voice-driven audio isn't the original, stem separation may leave artifacts, and you need rights to third-party audio.
More from coding & agent
- Claude Code 5.5 refuses to use Mac terminal, claims it's technically impossible — burkov · 2026-09-24
- Liquid AI to demo first reliable on-device agent model in live talk at AI Engineer Paris — helloiamleonie · 2026-09-24
- Interactive 3D island built in 8 hours with Opus 5.5 using $1,874 of tokens — Outside-Iron-8242 · 2026-09-24
- A real-time voice agent with zero US servers: Gladia STT, Gemma 4 on Scaleway, KugelAudio TTS — tobowers · 2026-09-24
- Devs add 10x more test harnesses just to slow down AI coding agents — cjimti · 2026-09-24
- Screenshot-annotated web edits tested with Ling-3.0: only 2 of 3 runs passed — alifcoder · 2026-09-24