v0.11.0 - Modifier-Only Shortcuts & Instant Recording

What's New

Modifier-Only Global Shortcuts

• Support for modifier-key-only shortcuts (e.g., double-tap ⌃⌘) alongside traditional key combinations
• Polling-based detection at 50Hz for reliable global modifier tracking
• Tap-to-toggle and hold-to-record modes both supported
• New shortcut recorder UI that accepts either key combos or modifier-only taps

Instant Recording (Microphone Warm-Up)

• New "Keep microphone ready" option in Settings > Shortcuts
• Pre-initializes audio hardware via prepareToRecord() for near-instant recording start
• Eliminates the 50-200ms cold-start delay when triggering recording
• Privacy-respecting: no audio is captured while idle — only hardware buffers are pre-allocated
• Automatically re-warms after each recording and on microphone device changes
• Opt-in toggle, disabled by default

Dark Mode Recording Button Redesign

• Dark frosted glass background that blends naturally with the dark window
• Deep wine-red idle state with brighter mic icon for better contrast
• Enhanced amplitude ring glow and halo opacity for dark backgrounds
• Subtle gradient edge stroke replaces invisible drop shadows
• Rich forest-green "copied" state adapted for dark mode

Download

Download the DMG below and drag OpenSuperWhisper to your Applications folder.

v0.10.0 - LLM Post-Processing & Custom Dictionary

What's New

LLM Post-Processing Editor

• Two-pass editing flow: STRICT → REPAIR → deterministic fallback
• Strict JSON contracts for reliable LLM output parsing
• StructureGuard validates JSON structure with flexible key matching
• ModeGuard validates content per mode (bullet count, length limits, banned prefixes)
• TranscriptPostProcessor provides deterministic fallback when LLM fails
• Per-mode DiffGuard thresholds (verbatim: 5%, clean: 40%, notes: 50%)
• Support for OpenAI-compatible endpoints (OpenAI, Ollama, local servers)

Custom Dictionary

• Add custom terms with aliases for transcription correction
• Dictionary terms included in LLM prompts with "may be misheard as" hints
• Aliases are automatically replaced in deterministic fallback
• Full CRUD management UI in Settings

UI Improvements

• Fixed Settings tab buttons clipping text (now vertical layout)
• Dark mode support for the main record button
• Translated Russian comments to English in accessibility code

Download

Download the DMG below and drag OpenSuperWhisper to your Applications folder.

v0.9.0 - Auto-Paste & Accessibility UI

What's New

Auto-Paste After Transcription

• Automatically paste transcriptions into the active text field when accessibility permission is granted
• Shows "Pasted!" feedback distinct from "Copied!" in the indicator
• Transcription remains on clipboard after pasting for easy re-use

Redesigned Accessibility Settings

• Clear "Enabled" / "Optional" status badges
• Feature list with icons showing what accessibility enables
• Auto-paste toggle is disabled when "Copy to clipboard" is off, with helper text
• "Open Settings" button for easy permission management

Technical Improvements

• Extracted clipboard logic to shared utility (cleaner code)
• Added proper error logging for paste operations
• Replaced blocking Thread.sleep with async DispatchQueue

Download

Download the DMG below and drag OpenSuperWhisper to your Applications folder.

v0.8.0 - Distil Model & Download UI Improvements

What's New

Features

• Distil large V3.5 model - New faster model option optimized for short-form transcription (1.5x faster than Turbo V3 large)
• Model descriptions - Each model now shows a helpful description explaining its use case and trade-offs
• Download cancel button - You can now cancel model downloads in progress

Improvements

• Better model card styling - Improved visual distinction between downloaded, not downloaded, and selected models with subtle borders and better state highlighting
• CoreML availability refactor - Cleaner, more maintainable approach using explicit hasCoreML flag instead of filename heuristics

Installation

1. Download the DMG file below
2. Open the DMG and drag OpenSuperWhisper to Applications
3. Launch from Applications (right-click → Open on first launch to bypass Gatekeeper)

v0.7.0 - Quick Language Switching

What's New

Features

• Language selector in main window - Quick access to language switching from the bottom bar, displayed as a globe icon with language code (EN, ZH, AUTO, etc.)

Fixes

• Click-through for recordings behind mic button - Items in the recording list are now clickable even when scrolled behind the floating mic button
• Empty prompts now supported - Users can clear the initial prompt for any language (previously required non-empty text)
• Updated English default prompt - Changed from "Hello." to "I mean." for better transcription accuracy

Installation

1. Download the DMG file below
2. Open the DMG and drag OpenSuperWhisper to Applications
3. Launch from Applications (right-click → Open on first launch to bypass Gatekeeper)

v0.6.0 - Resumable Downloads & Language-Specific Prompts

What's New in v0.6.0

📥 Resumable Model Downloads

• Resume interrupted downloads: If a model download is interrupted, it automatically resumes from where it left off
• Progress persistence: Download progress is saved, so you can quit the app and continue later
• Clean architecture: Robust download management with proper error handling

🎨 Redesigned Model Settings

• Unified card layout: All models displayed in a clean, consistent card format
• At-a-glance status: Easily see which models are downloaded, downloading, or available
• CoreML status: Clear indication of Neural Engine acceleration availability

🌍 Language-Specific Initial Prompts

• Per-language prompts: Set different initial prompts for each language to improve transcription accuracy
• Smart defaults: Built-in default prompts for 20 languages that help guide Whisper's transcription
• Auto-detect support: When using auto-detect mode, the app pre-detects the language and uses the appropriate prompt
• Easy customization: Edit prompts for any language in the Transcription settings tab
• Reset to default: One-click reset to restore the default prompt for any language

⚙️ Settings Window Improvements

• Movable window: Settings now opens as a proper window that can be moved anywhere on screen
• Custom tab bar: Clean horizontal tabs at the top (fixes macOS dropdown menu issue)
• Settings menu: Access Settings from the app menu with Cmd+, keyboard shortcut
• Scrollable tabs: All settings tabs are now scrollable for smaller screens
• Compact layout: Reduced spacing for better information density
• Done button: Always visible at the bottom of the window

🔧 Bug Fixes

• Fixed code signature crash when launching freshly built app

Installation

1. Download OpenSuperWhisper-0.6.0.dmg
2. Open the DMG and drag OpenSuperWhisper to Applications
3. Note: This is an ad-hoc signed build. On first launch, right-click the app and select "Open" to bypass Gatekeeper

Requirements

• macOS 14.0 or later
• Apple Silicon (arm64)

v0.5.0 - System Audio Capture

What's New in v0.5.0

🎧 System Audio Capture

• Transcribe desktop audio: Capture podcasts, YouTube videos, meetings, and any audio playing on your Mac
• Easy to use: Select "System Audio" from the microphone picker to start capturing
• Non-interrupting: Audio continues playing normally while recording
• Requirements: macOS 14.2+ and Screen Recording permission

🎯 Recording Feedback Improvements

• Amplitude ring halo: Outer halo with slow pulse clearly indicates "recording active" state
• Too-quiet detection: Breathing pulse appears after 1.5s of silence to alert when mic isn't picking up audio
• Signal indicator: Live green/gray dot in mic picker shows when audio is being captured
• Device name display: Mic picker now shows the full device name
• SYS/MIC label: Indicator window shows which audio source is active
• Smoother animation: Meter update rate increased to 60Hz

⚡ Performance & UX

• Model preloading: Whisper model starts loading on app launch for faster first transcription
• Loading status: Indicator window shows "Loading model..." vs "Transcribing..." for clearer feedback
• Asian autocorrect: Auto-applies CJK text formatting for auto-detected Chinese, Japanese, and Korean

Installation

1. Download OpenSuperWhisper-0.5.0.dmg
2. Open the DMG and drag OpenSuperWhisper to Applications
3. Note: This is an ad-hoc signed build. On first launch, right-click the app and select "Open" to bypass Gatekeeper

Requirements

• macOS 14.0 or later (14.2+ for System Audio)
• Apple Silicon (arm64)

v0.4.0 - CoreML Neural Engine Acceleration

What's New in v0.4.0

CoreML Neural Engine Support

• Added CoreML acceleration for ~1.5-2x faster transcription on Apple Silicon
• Neural Engine handles audio encoding, freeing up GPU for other tasks
• Automatic background download of CoreML encoder models
• Graceful fallback to Metal GPU when CoreML is not available

Bug Fixes

• Fixed: Quick double-tap of recording shortcut no longer causes infinite "Transcribing..." state

Build Improvements

• Enabled Link Time Optimization (LTO) for better performance
• Added OpenMP support for parallel processing
• Improved build system configuration

Settings

• New "Neural Engine Acceleration" section in Settings → Model tab
• Download/cancel CoreML encoder with progress indicator
• Manual download instructions with direct Hugging Face links
• "Show Models Folder" button for easy access

Installation

Manual Installation

1. Download OpenSuperWhisper-0.4.0.dmg below
2. Open the DMG and drag OpenSuperWhisper to Applications
3. Launch the app and grant necessary permissions

Requirements

• macOS 14.0 (Sonoma) or later
• Apple Silicon (ARM64) Mac

SHA256

a3c52fa3130d67e5fe1eb659e7c12cba06eabbb75129aeb4c59deb8e7c449c8e  OpenSuperWhisper-0.4.0.dmg

Note: This build is not notarized. You may need to right-click and select "Open" on first launch, or allow it in System Settings → Privacy & Security.

v0.3.0 - Typography System & UI Improvements

What's New

Typography System

• Centralized font definitions in Typography.swift for better maintainability
• Consistent font styling across the app

Recording List Cards

• Redesigned card layout with improved spacing and visual hierarchy
• Smart date/time formatting:
  • Today: shows just time (e.g., "3:23 AM")
  • Yesterday: "Yesterday 3:23 AM"
  • Within 7 days: weekday name (e.g., "Friday 3:23 AM")
  • Older: full date (e.g., "Dec 13, 2025")
• Hover effect on metadata bar for cleaner appearance
• Expandable "Read more/less" for long transcriptions

Bug Fixes

• Fixed transcription edit save not persisting to database
• Fixed UI not refreshing after editing transcription
• Detail window now stays open after saving edits

🤖 Generated with Claude Code

v0.2.0 - Redesigned UI

What's New

Redesigned Recordings List UI

• Floating record button with liquid glass effect - looks premium and floats above the list
• Click-to-view transcriptions - Click any card to open full transcription in a sheet for easy reading and editing
• Calmer, stable UI - Cards no longer expand on hover, keeping the list steady
• Better card separation - Subtle borders help distinguish cards from each other
• Improved record button - Mic icon always visible, with smooth animations when recording

UI Improvements

• Minimized bottom bar - more space for your transcriptions
• Edit transcriptions directly in the detail sheet
• Copy confirmation feedback on buttons

Installation

1. Download OpenSuperWhisper-v0.2.0.dmg
2. Open the DMG and drag OpenSuperWhisper to Applications
3. Launch and grant microphone permission when prompted

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