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.