Jev in production › Model and agent routing

Light-Whisper

Windows dictation app where, in Auto mode, Jev decides whether AI polish, screen context or web search is needed for a transcript before the LLM runs; optional Jev checks also flag meaning changes after polishing.

Open on GitHub ↗

The source publishes no measured number.
Use
Model and agent routing
Industry
Productivity
Form
Open-source tool
Stage
In production
Plugs into
Windows
Listed
2026-09-20
Found via
github
Repository
sypsyp97/light-whisper
Stars
69
Forks
9
Last push
2026-10-11
Language
Rust
License
GPL-3.0

The README opens with

Local and cloud speech-to-text for Windows. Hold a hotkey, speak, release. The text goes into your active app.

Download for Windows   ·   Get started   ·   简体中文

1. Install. Download the Windows 10/11 x64 x64-setup.exe from Releases. The recognition runtime is bundled; no separate Python or build tools are needed. 2. Choose an engine. Open Settings, select your microphone, then download a local model or configure a cloud API key and region. 3. Speak. Focus the app you want to type into, hold F2 , speak, then release. You can change the hotkey or switch to toggle recording in Settings.

Badge

For the project's own README, linking back here:

Listed in Jev in production

Also used for model and agent routing