Jev
Jev products and open-source projects, categorized and searchable with JevAll categories (98)
98 of 98 projects
Repository owner imageCategoryAgent workflows | UseVoice-driven browser actions | SourceGitHub |
Ownermoritzkremb | Original projecthttps://github.com/moritzkremb/jev-voice-browser | |
Research evidence
src/jev.js builds typed questions over the transcript, browser state and candidate spans, calls systemOne, and returns the answers with those candidates. The README explains act, wait, ask and ignore policies.
Distinct voice-input workflow, not a rename of jkudish/jev-browser or the baseline Jev Ultrafast.
Browser control and speech recognition were not executed. The model selects candidate text; it is not a speech-to-text or free-text generation model.
documentation source 1 (retrieved 2026-09-18)code source 2 (retrieved 2026-09-18)