🛠 Voicebox — an open-source tool for smart transcription
The project uses Whisper Large v3 Turbo to convert speech to text and the Qwen 3-30B model for intelligent formatting. The application is built on Wails (Go + React) and runs via Cloudflare Workers.
🌍 Demonstrates an efficient serverless architecture: the use of Cloudflare Durable Objects and Workers AI minimizes latency when processing audio and allows for scaling heavy STT and LLM models without managing your own infrastructure.
👤 Enables the use of powerful voice input that understands application context and transforms a stream of thoughts into a structured message, immediately inserting it into the active window.
Source 1: https://github.com/PrestigePvP/Voicebox