Fast, local, and offline dictation

A developer built a fast, accurate offline dictation tool using a small local speech-to-text model.

The project started after spotting a local speech-to-text backend in a new VS Code release and was built with Sherpa-ONNX and an AI agent.

It runs entirely on-device, avoiding cloud processing and offering better speed, privacy, and convenience for voice input.

The author says the tool will be polished and released later as a standalone app.