v0.1.0 — Initial Release

Real-time Voice Translator for Video Calls

First open-source release. Translates both sides of a conversation in real-time — you speak your language, the other person hears theirs.

Features

• 29 languages with speech-to-text, translation, and text-to-speech
• Local TTS via Piper (no cloud dependency for voice synthesis)
• Live transcript with timing metrics
• Call history with AI-powered summaries
• Download additional voice models from the web UI
• Localized tooltips (16 languages)

Stack

• Elixir OTP — process supervision and orchestration
• Rust — audio capture/playback, STT streaming, TTS synthesis
• Flask — web UI on port 5050
• Deepgram — speech-to-text (Nova-3)
• Groq — translation via llama-3.3-70b
• Piper — local neural TTS

Requirements

• macOS 14+
• BlackHole audio driver
• Free API keys from Deepgram and Groq

See README for setup and USAGE.md for the full guide.