ScriptDeck Studio
A self-hosted AI explainer-video station that turns a pasted script or rough outline into a finished, narrated, subtitled video with a synthetic presenter — running on a small on-prem box so course creators and small L&D teams avoid per-minute cloud fees and keep proprietary material in-house.
Online course creators and small L&D teams producing explainer or training videos
- One-box self-host installer that runs on a mini-PC or existing office server with GPU acceleration
- Script-to-storyboard inference that breaks a long script into timed scenes with suggested camera angles and visual cues
- Multi-language neural TTS with presenter avatars and auto-timed, typography-aware subtitles
- Batch queue with preview-and-edit checkpoints so teams can scrub, fix a single scene, and rerender without redoing the whole video
Self-hostable AI video is suddenly feasible thanks to open text-to-video and TTS models, and the GitHub trending repo (152 stars) shows real pull from creators fed up with watermark locks and per-minute cloud pricing.
AI video for course creators / L&D is a hot category — comparison guides for education list 11+ tools and HeyGen/Synthesia reviews surface repeatedly, but the narrow 'self-hosted on-prem' subset is smaller and more technical.11 Best AI Video Generators for Education in 2026 - Teachfloor ↗HeyGen vs Synthesia in 2026: Features, Pricing, and When to Go Pro ↗
Crowded both open and closed: 31 open-source video models already catalogued (Wan, HunyuanVideo, Mochi, Open-Sora 2.0, CogVideoX etc.) and Synthesia/HeyGen own the hosted explainer+avatar niche — few genuinely empty angles exist.31 Open-Source AI Video Models: Free Tools to Make Videos in 2026 ↗10 Best Open Source AI Video Generator Tools for 2026 ↗
Course creators and L&D teams already pay $24+/mo for HeyGen and significantly more for Synthesia — clear willingness to pay exists, but a perpetual/self-hosted license for a non-technical buyer is a harder sell than per-minute SaaS.HeyGen vs Synthesia: AI Video Showdown (2026) ↗HeyGen vs Synthesia: Complete Feature and Pricing Comparison ↗
Privacy/compliance need for in-house training content is structural and growing, and open TTS + avatar model quality is improving fast — multi-year relevance looks solid barring a major cloud-price collapse.31 Open-Source AI Video Models: Free Tools to Make Videos in 2026 ↗
Hard to ship as advertised: top open video models need 14–60GB VRAM (HunyuanVideo 14GB+, Mochi 60GB, Open-Sora 2.0 40GB+), so the 'small on-prem box' pitch conflicts with current model requirements; orchestrating TTS + lip-sync + subtitles + presenter into a turnkey box is also non-trivial engineering.31 Open-Source AI Video Models: Free Tools to Make Videos in 2026 ↗