🌱 Sprout — Kids Content Engine

Sprout is an AI-powered content generation platform that creates safe, age-appropriate educational videos for children. It uses a multi-stage pipeline with safety checks, human approval gates, and automatic voice/image generation.

Features

• Age-Adaptive Content : Three tiers: Sprout (3-6), Explorer (6-10), Tween (10-13)
• Safety-First Pipeline : Multiple safety checks before content is generated
• Human Approval Gate : Review and edit scripts before final generation
• Smart Voice Selection : LLM-powered voice matching based on content
• Auto-Generated Visuals : Scene images created via Pollinations.ai
• Content Nutrition Label : Parent-friendly rating breakdown
• Provenance Manifest : Full audit trail of the generation process

Tech Stack

Backend:

• Node.js + Express
• Anthropic Claude (script generation, safety, voice selection)
• ElevenLabs (text-to-speech)
• Pollinations.ai (image generation)

Frontend:

• React 18 + Vite
• Tailwind CSS

Project Structure

sprout/
├── client/                 # React frontend
│   ├── src/
│   │   ├── components/     # UI components
│   │   │   ├── AgeTierSelector.jsx
│   │   │   ├── ApprovalGate.jsx
│   │   │   ├── BundleOutput.jsx
│   │   │   ├── ManifestViewer.jsx
│   │   │   ├── NutritionLabel.jsx
│   │   │   ├── PipelineProgress.jsx
│   │   │   ├── SafetyBlock.jsx
│   │   │   ├── SeedPicker.jsx
│   │   │   ├── SlideshowPlayer.jsx
│   │   │   └── StyleConfig.jsx
│   │   ├── App.jsx
│   │   └── index.css
│   └── tailwind.config.js
├── server/                 # Express backend
│   ├── pipeline/           # Content generation pipeline
│   │   ├── approvalGate.js
│   │   ├── bundleAssembler.js
│   │   ├── contentRatingEngine.js
│   │   ├── costTracker.js
│   │   ├── elevenLabs.js
│   │   ├── manifestWriter.js
│   │   ├── nutritionLabelWriter.js
│   │   ├── router.js
│   │   ├── safetyChecker.js
│   │   ├── scenePlanner.js
│   │   ├── scriptGenerator.js
│   │   ├── slideGenerator.js
│   │   ├── topicExtractor.js
│   │   ├── vocabularyChecker.js
│   │   └── voiceSelector.js
│   ├── bundles/            # Generated content storage
│   ├── manifests/          # Provenance manifests
│   └── index.js
├── shared/
│   └── types.js            # Shared type definitions
└── .env                    # Environment variables

Setup

Prerequisites

• Node.js 18+
• npm or yarn
• Anthropic API key
• ElevenLabs API key

Installation

1. Clone the repository

   git clone https://github.com/yourusername/sprout.git
   cd sprout

2. Install backend dependencies

   cd server
   npm install

3. Install frontend dependencies

   cd ../client
   npm install

4. Create environment file

   # In the root directory
   cp .env.example .env

5. Configure environment variables

   # Required
   ANTHROPIC_API_KEY=your_anthropic_api_key
   ELEVENLABS_API_KEY=your_elevenlabs_api_key
   
   # Optional
   PORT=3001

Running the App

1. Start the backend server

   cd server
   npm start

2. Start the frontend dev server (in a new terminal)

   cd client
   npm run dev

3. Open the app

   http://localhost:5173
   

Pipeline Stages

1. Seed Discovery — Parse input topic
2. Safety Check — Validate content is age-appropriate
3. Voice Selection — LLM selects optimal voice for content
4. Script Generation — Create scene-by-scene script
5. Content Rating — Score on violence, fear, complexity, inclusivity
6. Human Approval — Review, edit, approve, or reject
7. Media Generation — Generate audio (ElevenLabs) + images (Pollinations)
8. Assembly — Bundle all assets with manifest

API Endpoints

Content Generation

• POST /api/generate — Start a new content generation
• GET /api/status/:id — Get pipeline status
• GET /api/bundle/:id — Get completed bundle

Approval

• POST /api/approve — Approve content for generation
• POST /api/edit — Submit edited scenes
• POST /api/reject — Reject content

Voice

• POST /api/voice-preview — Preview voice selection for seed
• GET /api/voice-sample/:voice_id — Get voice audio sample

Downloads

• GET /api/audio/:id — Stream audio
• GET /api/download/audio/:id — Download MP3
• GET /api/download/captions/:id — Download SRT
• GET /api/slides/:id/:sceneId — Get slide image

Voice Library

Sprout includes 8 pre-configured ElevenLabs voices:

Voice	Tiers	Character	Best For
Rachel	Explorer, Tween	Calm, friendly	Educational explainers
Domi	Explorer, Tween	Energetic, playful	Science, how-to
Bella	Sprout, Explorer	Warm, gentle	Storytime, nature
Antoni	Tween	Friendly, calm	History, social studies
Arnold	Explorer, Tween	Energetic, curious	Adventure, space
Adam	Tween	Calm, authoritative	Technology
Sam	Sprout, Explorer	Playful, warm	Animals, art
Gigi	Sprout	Warm, gentle	Preschool content

Content Safety

Sprout implements multiple layers of safety:

• Input Safety Check — Blocks inappropriate topics before processing
• Vocabulary Checker — Flags complex words for age tier
• Content Rating — Scores violence, fear, complexity (0-5 scale)
• Human Approval — Required before final generation
• COPPA Compliant — All content adheres to children's privacy standards

License

MIT