Make Jeff sound like you want.

Test speech input and choose a supported voice for Jeff’s spoken replies. This page runs in your browser; it is not the native Agent Jeff desktop service.

Assistant speaking voice

Browser-supported

Choose a voice available on this device

Speaking rate 1.0×

Preview text

Ready. Voice previews are spoken by your browser’s text-to-speech engine.

Voice commands

Permission required

When supported by your browser, start speech recognition to test simple commands. Microphone access begins only after you press the button.

Privacy: this page does not upload or save microphone recordings. Browser speech recognition may send audio to your browser provider; check its privacy behavior before enabling it.

Microphone is off.

Recognized words

Nothing recognized yet.

Demo commands: “hello Jeff”, “what time is it”, “read this sentence”, or “stop”. This tests transcription and spoken output only; it does not operate your computer.

Only respond to my voice

Unavailable

This setting is intentionally unavailable until a real speaker-verification engine is installed in the native desktop app.

Not secure yet: speech-to-text cannot tell whether the speaker is you. Saving a recording or a name in browser storage would not authenticate you, so this page will not pretend to enroll your voice.

Require my enrolled voice before replying
Unavailable until speaker verification is implemented and tested.

When implemented, enrollment should be explicit, encrypted, revocable, and processed locally where feasible. Voice matching alone must never authorize sensitive actions or replace Windows Hello.