jevbooks

← All projects

jev-voice-browser

moritzkremb/jev-voice-browser

Voice control for a real browser, deciding intent and target with Jev.

What it is

A Node app that drives a headed Chromium window via Playwright using spoken commands. It streams partial Web Speech API transcripts to a local server, which asks Jev for typed probabilities and acts, waits, asks, or ignores.

How it uses Jev

On each partial transcript, one Jev request asks a dozen typed questions: intent, target element, site, completeness, whether it is addressed to the browser, destructiveness, scroll amount, and candidate text/URL spans. Code uses the returned probabilities to decide whether to act, wait, ask, or ignore.

Primitives:choicescorenoul

Technique worth stealing

Extract candidate spans in code and let Jev only pick one, copying it verbatim rather than generating text.

Try it

Clone the repo, npm install, npx playwright install chromium, set TYPESAFE_API_KEY, then ./run.sh and open http://localhost:8787 in Chrome.

View on GitHub

judged by Jevjev-1.13.0

Evidence

Each line is one question put to Jev about the README. ≥ 0.60 reads as yes, ≤ 0.40 as no; in between Jev is not making a call.

  • Jev-centricyes0.95
  • Shows a System One patternyes0.98
  • Handles uncertaintyno0.11
  • Measuredyes0.62
  • Runnableyes0.93
  • Worth recommendingyes0.66
  • Model replicano0.04
  • Problem scopescore on a 0–2 scale0.41
  • About Jevyes0.99

Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 20 Sept 2026.