Enterprise-grade AI Sermon Transcription & Bible Tagging Suite
SermonSync is a high-performance Windows desktop application designed for churches. It features a multi-engine speech-to-text pipeline optimized for religious contexts and Nigerian accents, combined with real-time metadata tagging.
The core of SermonSync is its versatile AudioService, which supports multiple industry-leading transcription providers:
- 🇳🇬 N-ATLAS (Internal): A custom-built local service using a fine-tuned Whisper model for Nigerian Accented English. Runs locally on port 5003 for maximum privacy and zero latency.
- ⚡ Deepgram Nova-2: High-speed WebSocket streaming integration with advanced Phonetic Boosting to ensure religious vocabulary (e.g., "Mazzaroth", "Ephesians") is captured accurately.
- 🤖 Cloud Hybrid: Native support for OpenAI Whisper, Groq (Llama-3), and Google Cloud Speech for high-accuracy fallbacks and system-audio transcription.
A sophisticated regex-based parser that identifies Bible references in real-time, supporting:
- Standard formats:
John 3:16,1 Corinthians 13:4-7 - Nigerian abbreviations and phonetic matches (e.g.,
Gen,Exo,Matt). - Automated cross-referencing with a local Bible database.
- Identifies worship songs via fuzzy matching against a curated song database.
- Synchronizes lyrics and metadata for live screen display.
- Sync Panel: Multi-device synchronization via Socket.IO.
- Sidecar Services: Managed Python services for local AI processing (N-ATLAS).
- Contextual Notes: Rich metadata-aware note-taking that anchors transcription to specific sermon segments.
- Frontend: React 18 / TypeScript / TailwindCSS
- Backend: Electron 26 (Secure IPC Handshake)
- AI Sidecar: Python Flask / PyInstaller (N-ATLAS Service)
- Database: Supabase (Cloud) + SQLite/sql.js (Local Cache)
npm install