johannes77/amharic-cartoon-ai

Amharic Cartoon AI - Auto Video Generator # Amharic Cartoon AI - Auto Video Generator (defaults) This repository is a ready-to-run GitHub Actions project that: - Generates a ~5 minute kid-friendly story in **Amharic** (OpenAI) - Creates cartoon-style scene images (Stability.ai) - Synthesizes Amharic voiceover (Azure Speech) - Assembles a 5-minute MP4 with soft children's background music (FFmpeg) - Uploads to YouTube automatically (YouTube Data API) as **public** videos ## What I set as defaults - Story: **OpenAI** (GPT-4o-mini) - Images: **Stability.ai** (text-to-image) - TTS: **Azure Speech** (Amharic voice) - Video assembly: **FFmpeg** (Ken Burns-ish + crossfades) - Upload: **YouTube Data API** (OAuth refresh token) - Upload privacy: **public** ## Required GitHub Secrets Add these to **Settings → Secrets → Actions** in your GitHub repository: - `OPENAI_API_KEY` - OpenAI API key for story - `STABILITY_API_KEY` - Stability.ai API key for images - `AZURE_SPEECH_KEY` - Azure Speech key - `AZURE_SPEECH_REGION` - Azure region (e.g., eastus) - `YOUTUBE_CLIENT_ID` - Google OAuth client id - `YOUTUBE_CLIENT_SECRET` - Google OAuth client secret - `YOUTUBE_REFRESH_TOKEN` - OAuth refresh token for uploading - `VIDEO_UPLOAD_PRIVACY` - set to `public` (default) - `BACKGROUND_MUSIC_URL` - optional URL to an MP3 background music (public or S3). If not set, a soft default will be used if available. ## How to use 1. Clone the repo 2. Add the GitHub Secrets above 3. Optionally tweak prompts in `scripts/generate_story.py` 4. Run the workflow manually via Actions or let it run automatically every 3 days. ## Local testing You can run scripts locally (after installing dependencies) to generate a test short video: ```bash python -m venv venv && source venv/bin/activate pip install -r requirements.txt python scripts/generate_story.py --lang am --out output/story.json python scripts/render_images.py --story output/story.json --out_dir output/images python scripts/tts.py --story output/story.json --out_dir output/audio pytho …