VTyper
Changelog

What's new in VTyper

VTyper is under active development. We ship improvements regularly, so the product keeps getting better.

Latest: v0.9.1 · 2026-10-11

v0.9.1

Latest
  • The floating status window now appears next to the text cursor and stays close as the cursor moves.
  • The status window comes in small, medium, and large sizes, with clearer text and an opaque background.
  • Compose mode has a new pause setting: choose how long to wait after you stop speaking before a passage is finished.
  • AI replies stream in as they are generated. Long replies can be scrolled, and incomplete replies are never shown.
  • Fixed text in compose mode sometimes appearing in the input preview area.
  • Renamed the speech pause and idle-disconnect settings to clearer names, with added explanations.

v0.9.0

  • Dictation mode: type as you speak. Text goes straight into the focused input field. Hands-free mode is now part of dictation.
  • Compose mode: speak a full passage first, then optionally send it to an AI service before the text is inserted. You can configure several compose presets.
  • Mode picker: tap Shift to switch quickly between dictation and any compose preset.
  • Status window: a new way to see voice status, including recognition and processing progress.
  • The preset list in settings shows each preset's input mode.
  • Fixed several issues when stopping recognition.

v0.8.0

  • A new first-run setup guide checks your services and microphone, and lets you use either online services or local models.
  • Experimental support for the Fcitx5 input method on Linux.
  • Two new speech recognition services, one of which supports compressed audio for faster transfer.
  • Presets can be reordered by dragging, or from the right-click menu.
  • Speech recognition services that have been idle for a while stop automatically to save resources.
  • The settings window has been redesigned for a more consistent look and feel.
  • New invitation-based beta application.
  • Fixed duplicate results at the end of recognition.

v0.7.0

  • The input method and recognition sessions are now managed separately, so stopping and finishing recognition is more reliable.
  • Custom keyboard shortcuts for the input method, and a choice of how recognition finishes when you stop.
  • History supports data management.
  • Live recognition has a silence timeout: it ends automatically after a long period without sound.
  • Private production release, with installers distributed through private channels.
  • Dependencies upgraded.

v0.6.0

  • New offline local speech recognition. It adapts to both streaming and offline models, with optional punctuation and text normalization.
  • Redesigned version and update window.
  • The web console and admin dashboard are merged into one interface.
  • Fixed streaming recognition dropping the last word.

v0.5.0

  • A new low-latency option for voice activity detection (detecting when you are speaking), now the default.
  • Two new cloud speech recognition services.
  • Limits how many recognitions can run at the same time, to avoid running out of resources.
  • Accounts support paid plans and subscriptions.

v0.4.0

  • Presets: import, export, duplicate, and search. Cloud accounts can sync presets.
  • New offline local speech recognition with GPU acceleration.
  • Voice detection parameters can be adjusted directly in settings.
  • The interface is available in multiple languages.
  • Speech recognition services run as separate processes, so one failing service does not affect the main app.
  • Installers for major Linux distributions.

v0.3.0

  • Setup wizard: checks and guides installation of the input method components, and enables VTyper.
  • History page: browse, search, and play back the audio and text of each recognition, with optional password protection.
  • Custom keyboard shortcuts.
  • Statistics page: usage data such as recognition count and word count.
  • Accounts and device authorization: sign in on the web to link your computer to your account.
  • Website redesign with FAQ, blog, and early access sections.
  • A new cloud speech recognition service.

v0.2.0

  • Speech recognition services use a plugin design and can be enabled as needed.
  • First version of the settings window: service list, configuration dialogs, and volume control.
  • Custom sound effects, with adjustable volume.
  • A new real-time speech recognition service.

v0.1.0

  • First usable version: runs as a Linux input method, with start and stop for voice input. Press Esc to cancel.
  • Supports multiple speech recognition services, both online and local.
  • Voice activity detection that automatically detects when you start and stop speaking.
  • Officially named VTyper.

Have a feature request or found a bug?

Your feedback shapes what we build next.

Contact us