Where AudioBud is headed.

The Apple Silicon candidate prepares a signed and notarized macOS DMG, a protected Apple release workflow, and permission, shortcut, settings, input, audio, and model-loading recovery fixes shared with Windows. Publication waits for the clean-Mac, remote candidate, and exact draft-asset gates.

Output routing: pin dictation to a window you chose and keep working somewhere else, or send a single dictation to a chosen window without locking. Deliveries now run off the main thread, a successful delivery names the window it reached, and per-application output profiles override paste behavior for apps that need it. The settings system behind all of it moved to one generic command instead of dozens of one-off ones.

The U.S. Microsoft Store listing now delivers v0.4.4 as a signed NSIS package. New Store installs join AudioBud's signed update feed and check it for later releases by default.

The signed updater now uses a Windows-compatible archive format, completing the end-to-end update path introduced across the 0.4 releases. Microsoft Store now serves this release through the same update channel.

A release-integrity patch: checksums and build provenance on every release, CI that compiles and tests the real Windows transcription engine before it ships, and clipboard history no longer picks up temporary dictation text.

A distribution release with signed and timestamped Windows installers, verified packaged executables, a protected release pipeline, and public publisher-domain pages. Release integrity followed in v0.4.1, and automatic updates in v0.4.2 through v0.4.4.

A dictation-quality release: spoken numbers come through as digits and symbols ($25, 10%, 3.5), a tray switcher flips between formatted and raw output without opening settings, and raw mode gained the groundwork to read spoken punctuation.

A naming release: the tray tooltip, the command line, and the headers AudioBud sends to post-processing providers all name the app rather than the project it forked from.

A translation release: the 19 languages AudioBud ships besides English now cover the whole interface, the app calls itself AudioBud in them rather than Handy, and the learned-words counter agrees with its count.

A stability patch for the default setup: dictation now starts on Windows accounts with non-English names, history settings no longer delete unsaved recordings, pasting keeps what you had copied, and a stuck engine can no longer hang the window.

Opt-in, on-device personalization: AudioBud learns the words you use most from your own dictation history and applies the ones you accept, with view, export, and reset controls.

A self-contained installer with a portable build, the overlay placement grid, tray quick-toggles, and a dictionary that stops correcting words that were already right.

The first working local prototype, forked from Handy with the frog and swamp identity and a Windows-first default engine.