v3.6.0 - Purple Oscillator

Purple Oscillator

• New oscillator waveform — smooth canvas-based wave replaces the bar visualizer (no more jank when speaking)
• App icon in recording overlay
• Solid dark background — removed transparent/blurred backdrop for cleaner look
• Purple theme — oscillator color matches the app icon

v3.5.0 - Waveform

v3.5.0 — "Waveform"

---

What's new

• Multi-monitor overlay — Recording overlay now centers on the monitor where your cursor is
• Redesigned waveform — 80-bar rolling waveform mirrored from center, replacing the old 5-bar display
• Improved readability — Bumped text opacity across the menu bar panel, overlays, and transcription history

v3.4.1 - Speak Now

Speak Now

Fixes

• Short utterance capture — Audio stream startup is now synchronized: start_recording() blocks until stream.play() confirms the mic is active. This eliminates the ~50-200ms gap that could lose 2-3 word utterances.
• "Speak now" prompt — Overlay text changed from "Recording..." to "Speak now" so you know exactly when the mic is ready.

v3.4.0 - Silk Road

Faster Paste & UI Refresh

Paste System

• 3x faster paste — happy path reduced from ~300ms to ~105ms
• Clipboard verification instead of blind re-set
• Structured focus restoration with polling (aborts if target app lost)
• PID generation guard prevents rapid double-tap issues
• Paste errors now surface in overlay ("Paste failed — press Cmd+V")
• Blocking sleep moved off async runtime via spawn_blocking

UI Refresh

• Glassmorphic overlays — recording + switch overlays with blur, gradients, inner highlights
• Smooth state crossfades — recording → processing → ready transitions
• New processing spinner — comet-tail arc with pulsing glow
• Animated checkmark — SVG stroke-draw on transcription ready
• Richer waveform — 7 bars with reflections, snappier 60ms animation, sqrt response curve
• Boosted audio sensitivity — 12x RMS multiplier for better speech visualization
• Menu bar polish — gradient background, scroll fade, empty state with mic icon, gradient dividers
• Copy feedback — green checkmark flash on transcription copy
• Switch toast — slide-up entrance animation with color dot

Also

• Expanded filler-word removal list
• Improved text contrast on Settings, Relaunch, Quit, version label

v3.3.0 - Clean Sweep

v3.3.0 — Clean Sweep

---

What's New

• Pinned Tauri dependencies — All Tauri packages pinned to specific versions for reproducible builds
• Faster regex processing — Filler word patterns are now compiled once and cached, not rebuilt on every transcription
• Smarter permission polling — Checks every 3s instead of 1s, and pauses entirely when the settings tab is hidden
• Reliable settings toggles — If a toggle fails to save, it now reverts automatically instead of showing a stale state
• Safer model downloads — 10-minute timeout with 30-second connect timeout (was 1 hour with no connect timeout)
• Dynamic version display — Version in the menu bar panel now always matches the actual build
• Removed scaffolding — Cleaned up leftover Tauri placeholder code

---

How to Upgrade

1. Download Speech.app.zip below
2. Quit the currently running Speech app (click the menu bar icon → Quit)
3. Unzip and drag Speech.app to your /Applications folder, replacing the old version
4. Open Speech from Applications
5. Re-grant permissions — Go to System Settings → Privacy & Security and enable Speech under:
   • Accessibility
   • Input Monitoring
   • Microphone (should prompt automatically on first recording)

Note: Your settings, downloaded models, and hotkey bindings are all preserved. Only macOS permissions need to be re-granted because the app is ad-hoc signed.

---

🤖 Generated with Claude Code

v2.4.0 - Quiet Hands

v2.4.0 — Quiet Hands

---

What's New

• Clipboard preservation — Your clipboard is saved before auto-paste and restored after, so transcription no longer destroys what you copied
• Escape to cancel — Press Escape during recording to abort without transcribing
• Live waveform — Overlay bars now respond to your actual voice instead of random animation
• Filler word removal — Strips "um", "uh", "you know", "I mean" and collapses stutters like "I I I think" → "I think" (toggle in Settings)
• Multi-monitor support — Recording overlay now appears on the screen where your cursor is, not always the main display

---

🤖 Generated with Claude Code

v2.3.4 - Back to Basics

v2.3.4 - Back to Basics

What's Changed

Reverted: Auto-open Permissions Tab Feature

The v2.3.2 and v2.3.3 feature to automatically open Settings to the Permissions tab after upgrades has been rolled back. Settings now returns to the original behavior:

• ✅ Always opens to General tab by default
• ✅ Simple, predictable behavior
• ✅ No version tracking or tab switching logic

If you need to check permissions, simply click the Permissions tab manually in Settings.

Installation

1. Download Speech.app.zip below
2. Unzip and move Speech.app to /Applications/
3. Launch the app

---

Full Changelog: v2.3.3...v2.3.4

v2.3.3 - Permission Guard (Fixed)

v2.3.3 - Permission Guard (Fixed)

What's New

Fixed: Permissions Tab Now Shows After Upgrade 🔐

The v2.3.2 feature to auto-open the Permissions tab after upgrades had a bug where it wouldn't work consistently. This is now properly fixed:

• Version check moved from Settings window to app launch
• Settings now correctly opens to Permissions tab after every upgrade
• Ensures you immediately see which permissions need to be enabled

Why This Matters

After macOS updates or app upgrades, permissions may need to be re-granted. The Permissions tab shows:

• ✅ Microphone (for voice recording)
• ✅ Accessibility (for text injection)
• ✅ Input Monitoring (for keyboard simulation)

Auto-opening this tab makes it much easier to notice when permissions need attention, especially for the auto-paste feature.

Installation

1. Download Speech.app.zip below
2. Unzip and move Speech.app to /Applications/
3. Launch the app (replaces your current v2.3.2)
4. Click Settings to verify it opens to Permissions tab ✨

---

Full Changelog: v2.3.2...v2.3.3

v2.3.2 - Permission Guard

v2.3.2 - Permission Guard

What's New

Auto-Open Permissions Tab After Upgrades 🔐

After every version upgrade, the Settings window now automatically opens to the Permissions tab instead of General. This ensures you immediately see which permissions need to be enabled for the app to work correctly:

• ✅ Microphone (for voice recording)
• ✅ Accessibility (for text injection)
• ✅ Input Monitoring (for keyboard simulation)

This makes it much easier to notice when permissions need to be re-granted after an update, especially for the auto-paste feature.

Installation

1. Download Speech.app.zip below
2. Unzip and move Speech.app to /Applications/
3. Right-click → Open (first launch only, to bypass Gatekeeper)
4. Grant required permissions when prompted

---

Full Changelog: v2.3.1...v2.3.2

v2.3.1 — "Quiet Mic"

v2.3.1 — "Quiet Mic"

---

What's New

Recording UI waits for microphone readiness

• The red mic overlay and recording indicator now only appear after the microphone is confirmed capturing
• Fixes clipped speech at the beginning of recordings — no more missed first words
• Quick press-and-release of the hotkey still works correctly (stop awaits mic startup)

Technical

• startRecording() defers isRecording = true until after AudioRecorder.startRecording() completes
• Added recordingStartTask to guard against double-starts and allow stop to await startup
• Extracted performStopAndTranscribe() for cleaner control flow