urusaqqrun/jarvis-voice is a dictation tool that captures speech from the macOS menu bar and turns it into text in the input field of the app you are currently using. FACTS lists no publicly announced latest release, and the repository description and README introduce transcription, AI sentence polishing, automatic input target detection, and voice command features. The repository is written in JavaScript and uses the MIT license. The repository has 50 stars.
The goal of this tool is to put what you say directly into the input field of the app you are working in. Transcription can use an on-device engine or services such as OpenAI, and the transcribed sentences can be polished with OpenAI, DeepSeek, or Claude. Auto focus is a feature that finds the input target on screen and focuses the input field, and it also supports controlling apps with voice commands. So it does more than convert speech into characters: the setup bundles input target selection and post-processing into a single menu bar app.
That said, FACTS did not confirm whether Korean speech synthesis is supported, and neither speech recognition accuracy nor real-world app integration stability was independently verified. The app only inserts text into the input field and does not send it, so the user has to press Enter manually to send the content. Based on the currently confirmable evidence, the scope of the dictation and assistive features can be described, but it is hard to judge recognition quality and integration reliability across multiple apps.