Voice Browser: spoken commands to bounded browser actions

A local Node and Playwright app streams partial speech transcripts, asks Jev for intent and target choices, then applies code-level action gates.

Notes

The README reports roughly 250–350 ms for a Jev decision and 34/34 author integration cases; speech recognition uses the browser's Web Speech API.

source-backed — Public repository, docs or live artifact. About this label

Destinations

Filed under

• #browser
• #voice
• #computer-use
• #playwright

Also on Jevland by Moritz Kremb

• Voice browser in roughly 300 ms demo · author demo
• The early Jev projects thread roundup · commentary

In collections

• Browser & computer use →
• Agent harnesses →