Video Content Factory
Organize video content ideas and draft AI scripts with browser-native TTS
Episodes & demos
- S01E01 Playhttps://jpeckenpaugh.github.io/vid-factory/ ► Watch (opens in a new tab)
About
Video Content Factory is a web application for organizing early-stage content ideas and drafts. Its primary local runtime is a browser-native standalone application (browser-edition/) that runs entirely client-side with no backend server: a dedicated Web Worker database layer uses vendored sql.js (SQLite compiled to WebAssembly) with persistence in the Origin Private File System (OPFS), communicating through an RPC message-passing protocol between the UI (app.js) and db-worker.js. It manages catalogs for applications (YouTube, TikTok, Instagram) and companies (Acme Studio, Northstar Media, Pine & Peak), content projects with optional application/company links, and one persisted script/prompt draft per project. Workspace settings store AI provider credentials (Gemini, OpenAI, Ollama) in a local SQLite ai_settings table with custom endpoint URLs and connectivity tests; API keys are stripped during export. A client-side AI script generator formulates formatted drafts from user prompts and confirms before overwriting manual drafts. A dedicated TTS worker (tts-worker.js) runs kokoro-js speech synthesis in WebAssembly/WebGPU, exposing voice selection (af_heart, af_bella, am_adam, am_michael), a 0.75x-1.25x speed slider, an inline HTML5 audio preview player, and local audio track persistence with metadata (voice_id, speed, script_snapshot, duration). Workspaces are portable via JSON export/import with structural validation and seed reset. The original FastAPI JSON API and Bootstrap browser UI (backend/ and frontend/, pinned fastapi==0.115.12 and uvicorn[standard]==0.34.0) remain available unchanged as a parity reference and fallback runtime, serving a seeded local SQLite database (backend/data/vid_factory.db). Multi-speaker timeline synchronization, cloud rendering pipelines, server-side audio processing, and cloud sync are explicitly deferred. Status: Sprint 02 complete (client-side AI & dialogue TTS), verified pass with sandbox environment limitations.
Summary
A browser-centric web application for organizing early-stage video content ideas, managing catalogs of applications and companies, drafting scripts with client-side AI providers, and synthesizing dialogue with in-browser text-to-speech.
Tech
- JavaScript
- Python
- FastAPI
- SQLite
- sql.js
- WebAssembly
- kokoro-js
- OPFS
- WebGPU
- Bootstrap 5
- HTML/CSS
Tags
- ai-script-generation
- content-creation
- sqlite
- text-to-speech
- video
- web-app