Voxloom v0.3.0

Voxloom v0.3.0

This build is signed with a Developer ID and notarized by Apple.

What's new

• One-click download for the local cleanup model (BUS-181). This replaces the old Import model… button. A model you imported in an earlier version still counts as installed.
• Dictation now pastes into Apple Messages (BUS-34).
• The listening pill stays above a visible Dock (BUS-177).
• Dictionary words are spelled in the transcript exactly as you saved them (BUS-138).

Install

1. Download Voxloom_0.3.0_macos_*.dmg (or the zipped .app) below and
   check it against SHA256SUMS: shasum -a 256 -c SHA256SUMS --ignore-missing.
2. Open the dmg and drag Voxloom to Applications.
3. Launch Voxloom from Applications.

Permissions (macOS)

• Microphone: requested on first dictation. Voxloom should appear by name
  under System Settings → Privacy & Security → Microphone.
• Accessibility: needed to insert text into other apps. Enable Voxloom
  under System Settings → Privacy & Security → Accessibility.

Speech model

Transcription runs locally with whisper.cpp. The Whisper model is not
bundled; download it separately as described in the README
("Local STT (whisper.cpp)").

Known limitations

• macOS only (minimum macOS 13.1 (Ventura)); Windows and Linux are not supported yet.
• No auto-update; download new versions from the Releases page.
• The cleanup model download doesn't resume; if it's interrupted, it starts over.

What's Changed

• fix(delivery): paste dictation into Apple Messages (BUS-34) by @KyleMyers001 in #148
• fix(overlay): keep the listening pill above a visible Dock (BUS-177) by @KyleMyers001 in #151
• feat(dictionary): spell dictionary words as saved in the transcript (BUS-138 PR A) by @KyleMyers001 in #149
• BUS-181: one-click download for the local cleanup model by @KyleMyers001 in #152
• BUS-186: remove cleanup model Import (download replaces it) by @KyleMyers001 in #154
• chore: bump version to 0.3.0 by @KyleMyers001 in #153

Full Changelog: v0.2.0...v0.3.0

Voxloom v0.2.0

Voxloom v0.2.0

This build is signed with a Developer ID and notarized by Apple.

Install

1. Download Voxloom_0.2.0_macos_*.dmg (or the zipped .app) below and
   check it against SHA256SUMS: shasum -a 256 -c SHA256SUMS --ignore-missing.
2. Open the dmg and drag Voxloom to Applications.
3. Launch Voxloom from Applications.

Permissions (macOS)

• Microphone: requested on first dictation. Voxloom should appear by name
  under System Settings → Privacy & Security → Microphone.
• Accessibility: needed to insert text into other apps. Enable Voxloom
  under System Settings → Privacy & Security → Accessibility.

Speech model

Transcription runs locally with whisper.cpp. The Whisper model is not
bundled; download it separately as described in the README
("Local STT (whisper.cpp)").

Known limitations

• macOS only (minimum macOS 13.1 (Ventura)); Windows and Linux are not supported yet.
• No auto-update; download new versions from the Releases page.

What's Changed

• Move marketing website to voxloom-site by @KyleMyers001 in #138
• BUS-139: trim trailing silence before Whisper to stop hallucinated "Thank you" by @KyleMyers001 in #132
• Keep soft trail-offs the trailing-silence trim cut (BUS-143) by @KyleMyers001 in #137
• ci(cleanup-worker): stop CMake probing for OpenSSL (fixes red main on macOS) by @KyleMyers001 in #139
• test(worker): widen the prewarm preemption time bound for slow CI runners by @KyleMyers001 in #141
• ci(release): stop CMake probing for OpenSSL in the macOS worker build by @KyleMyers001 in #140
• feat(shortcut): lone-key tokens and recorder capture (BUS-46 slice 1) by @KyleMyers001 in #142
• BUS-46 slice 2: lone-key state machine and macOS listen-only tap (not wired) by @KyleMyers001 in #143
• BUS-46 slice 3: wire the lone-key shortcut (arm from settings, edges into the gesture machine, fix L4) by @KyleMyers001 in #144
• BUS-156: Copied, press ⌘V when Copy happens in a terminal by @KyleMyers001 in #145
• chore: bump version to 0.2.0 by @KyleMyers001 in #146
• feat: paste dictation into WezTerm and other terminals (BUS-157) by @KyleMyers001 in #147

Full Changelog: v0.1.0...v0.2.0