Skip to content

Releases: LeoMartinDev/Plume

v0.4.0

Choose a tag to compare

@github-actions github-actions released this 13 Oct 10:52
3fff8ee

Plume 0.4.0

Improvements

  • With a Bluetooth headset, the start of a dictation is no longer cut off. Plume waits until the microphone really captures audio before showing Listening, and shows "Opening mic…" in the meantime.
  • On macOS, Plume appears in the Dock and the app switcher while its settings are open, so the window no longer gets lost behind other apps, for example after the microphone permission prompt.

Fixes

  • On macOS, dictation now inserts text into Chromium and Electron apps such as Claude, Slack and VS Code.
  • On macOS, pasting the transcription no longer types a "v" in some apps instead of the text.

Downloads are native installers: Windows .exe, macOS Apple Silicon .dmg, and Ubuntu 24.04+ .deb. Speech models are downloaded separately.

These installers are not publisher-signed or notarized. macOS builds use a self-signed certificate, so the first launch needs Right-click → Open.

v0.3.2

Choose a tag to compare

@github-actions github-actions released this 13 Oct 08:02
7bd3d23

Plume 0.3.2

Fixes

  • On macOS, Plume no longer quits after a dictation without inserting the text. This happened when the app in front did not report which field had the focus.
  • On macOS, quitting Plume with a Whisper model loaded no longer ends in a crash report.

Downloads are native installers: Windows .exe, macOS Apple Silicon .dmg, and Ubuntu 24.04+ .deb. Speech models are downloaded separately.

These installers are not publisher-signed or notarized. macOS builds use a self-signed certificate, so the first launch needs Right-click → Open.

v0.3.1

Choose a tag to compare

@github-actions github-actions released this 12 Oct 08:40
e79ec62

Plume 0.3.1

Fixes

  • On macOS, updates no longer silently turn off your dictation shortcuts. Plume is now signed with the same certificate in every release, so macOS keeps its Microphone and Accessibility permissions across updates. You need to allow them one last time after installing this version.
  • On macOS, if Plume starts without the Microphone or Accessibility permission, it opens the Permissions step again instead of ignoring your shortcuts.

Improvements

  • Plume writes a log file you can share when something goes wrong: ~/Library/Logs/Plume/plume.log on macOS, %LOCALAPPDATA%\plume\logs\plume.log on Windows and ~/.local/share/plume/logs/plume.log on Linux.

Downloads are native installers: Windows .exe, macOS Apple Silicon .dmg, and Ubuntu 24.04+ .deb. Speech models are downloaded separately.

These installers are not publisher-signed or notarized. macOS builds use a self-signed certificate, so the first launch needs Right-click → Open.

v0.3.0

Choose a tag to compare

@github-actions github-actions released this 12 Oct 07:47
dd2e91f

Plume 0.3.0

New

  • Parakeet TDT v3 from NVIDIA is available as a speech model in Settings → Models. It covers 25 European languages, including French and English.

Improvements

  • On macOS, onboarding now has a dedicated Permissions step before the test. It explains why Plume needs the microphone and Accessibility, and you continue once both are allowed.
  • On Windows, the settings window has a subtle border.

Downloads are native installers: Windows .exe, macOS Apple Silicon .dmg, and Ubuntu 24.04+ .deb. Speech models are downloaded separately.

These installers are not publisher-signed or notarized. macOS uses local ad-hoc signatures only.

v0.2.1

Choose a tag to compare

@github-actions github-actions released this 11 Oct 13:25

Plume 0.2.1

Plume is local push-to-talk dictation: hold a shortcut, speak, and the text is inserted where you type. Speech recognition runs on your computer.

Highlights

  • Choose a speech model during onboarding (Nemotron 3.5 Compact or Whisper) and try dictation before finishing.
  • Hold-to-talk and hands-free shortcuts, with Esc to cancel.
  • A dictation bubble with a live waveform, transcription progress and clear errors.
  • Insert text into the focused field or copy every dictation to the clipboard; when nothing can receive text, it stays on the clipboard.
  • History of recent dictations with retry from the saved audio.
  • Light and dark themes, and an English or French interface.

Updates

  • Updates install from Settings → Updates. Installing an update on Windows now works.

Downloads are native installers: Windows .exe, macOS Apple Silicon .dmg, and Ubuntu 24.04+ .deb. Speech models are downloaded separately.

These installers are not publisher-signed or notarized. macOS uses local ad-hoc signatures only.