SwiftSpeak
Voice-to-text, done well
Dictation app with AI formatting, context-aware output, power modes with document RAG, and cost analytics. 100+ languages, 10+ AI providers, works offline.
Every project here started from a personal problem. Translation apps because my family speaks three languages. Dictation apps because I got tired of typing. A singing app because I wanted to practice without embarrassing myself. All free, all open source.
Voice-to-text, done well
Dictation app with AI formatting, context-aware output, power modes with document RAG, and cost analytics. 100+ languages, 10+ AI providers, works offline.
Speech to text, instantly
Push-to-talk dictation for macOS. 25+ languages, real-time translation, custom vocabulary. Uses OpenAI Whisper, GPT-4o Mini, or ElevenLabs Scribe.
Real-time speech translation
Self-hosted PWA for multi-language rooms (up to 5 languages). Built so my Polish parents and Egyptian in-laws could actually talk to each other.
Automatic group chat translation
A bot that sits in a WhatsApp group and translates every message. Polish parents write in Polish, Egyptian wife reads in English. No app switching.
MCP server for agentic development
Runs multiple AI agents in sequence to build software features. Breaks work into phases (PM, Architect, Designer, Implementer) with approval gates and telemetry.
Singing practice with any YouTube song
AI stem separation, real-time pitch detection, synced lyrics, phrase-by-phrase scoring. Paste a YouTube link, isolate the vocals, practice along.
These all started as conversations. If you have a problem that sounds like it could use a small, focused app, let me know.
Say hello