Custom AI Text-to-Video Generator

I need a standalone application that converts written scripts into polished videos—true-crime narratives will be my primary focus, though the engine should handle any subject just as smoothly. The workflow I have in mind is simple: paste text, choose an animation style, press render, and receive a finished video complete with synced voiceover and captions.

Core functions
• Text-to-speech voiceover integration so narration sounds natural and matches the pacing of the script.
• Automatic subtitle generation with the option to edit or export the caption file.
• 2D animation pipeline: motion graphics, character rigs, stock footage compositing, and lower-thirds that fit darker true-crime aesthetics yet remain flexible for other genres.
• Template management—store, edit, and reuse scene layouts, colour palettes, and font sets.
• Media library support for images, short clips, and sound effects.
• One-click export in 1080p (minimum) to MP4 and direct share to YouTube/TikTok.

Tech preferences
I am open to the stack you’re most comfortable with, whether that’s Python + FFmpeg + Stable Diffusion, a Node.js Electron desktop build, or a cloud-hosted solution wrapped in a clean web UI. As long as the system runs on Windows and Mac, produces reliable renders, and future feature additions are straightforward, I’m happy.

Acceptance criteria
1. A compiled, installable app (or web deployment with user accounts).
2. Source code with clear documentation for local build and future maintenance.
3. Demo video showing the conversion of a 500-word true-crime script into a finished 2-minute video that includes voiceover and timed captions.
4. Short user guide covering script import, template selection, render settings, and exporting.

Please outline the architecture you propose, any third-party APIs or libraries you’d leverage, and an estimated timeline from prototype to final build.

Please use Vid.ai as an example. I need to be able to do that that's website does.