🌊 WisprWave

The Intelligent, Private, Fast Speech-to-Text for macOS.

WisprWave brings OpenAI's powerful Whisper models directly to your Mac menu bar. Experience class-leading transcription speed and accuracy without sending a single byte of data to the cloud. It's completely free, works offline, and injects text into any application.

   

Key Features

• ⚡️ Blazing Fast: Powered by WhisperKit and optimized for Apple Silicon (CoreML). Transcription happens locally and instantly.
• 🔒 100% Private: Your voice never leaves your device. Works completely offline—airplane mode approved.
• 🤖 State-of-the-Art Accuracy: Supports the latest Whisper Large-v3 models (and smaller variants) for nuanced, accurate transcription.
• ⌨️ Universal Dictation: Injects text directly into your active application (Notes, Slack, Obsidian, VS Code, etc.).
• 💻 VM Compatible: Special legacy mode designed for virtual machines (VMware, Parallels, UTM).

Performance Modes

WisprWave offers three modes to fit your environment:

🚀 Boost Mode

Maximum Speed. For native macOS environments, Boost Mode streams audio in the background and pre-transcribes as you speak using WhisperKit's clipTimestamps — each pass only decodes new audio past the last confirmed segment, so it never re-processes the same speech twice.

• Instant Results: When you stop speaking, only the final 1–2 seconds need processing. Text appears almost immediately.

🎙️ Standard Mode

Simple & Reliable. Records all audio first, then transcribes the full buffer in one pass after you stop recording.

• Good For: Short dictations where the slight delay after stopping is acceptable.
• Optimized: Uses tuned DecodingOptions (zero-temperature, prefill cache, no timestamps) for fast single-pass transcription.

🐢 Legacy Mode (VM Support)

Maximum Compatibility. Some virtualized environments restrict direct hardware audio access required for streaming.

• Dependable: Records to a temporary file first, then transcribes.
• Ideal For: Use this if you are running macOS inside a VM or experience audio driver issues.

Download

Don't want to build from source? Download the latest ready-to-use app:

Download WisprWave.zip

1. Download the zip file.
2. Unzip and drag WisprWave.app to your Applications folder.
3. Follow the launch instructions below.

Recommended Model: Wisper Large V3 632T New

Installation from Source

WisprWave is open-source. Build it directly from the codebase in minutes.

Prerequisites

• macOS 14.0+ (Sonoma or later recommended)
• Xcode 15+ (for Swift 6 support)

Steps

1. Clone the Repository:

   git clone https://github.com/yourusername/WisprWave.git
   cd WisprWave

2. Build and Package: We've included a helper script to build and sign the app bundle automatically.

   ./package_app.sh

3. Launch: Open the generated app in the project directory:

   open WisprWave.app

   ⚠️ Note: Since this app is not signed with an Apple Developer ID, macOS (Gatekeeper) may block it.

   • System Settings: Go to System Settings > Privacy & Security and click "Open Anyway" near the bottom.
   • Alternative: Right-click the app in Finder, select Open, and confirm in the dialog.

4. Permissions: On first launch, grant Accessibility (for text injection) and Microphone permissions when prompted. The menu bar icon will appear, ready for action!

Contributing

Contributions are welcome! Feel free to open issues or submit pull requests to make WisprWave even better.

License

MIT License. Free for everyone.