Native Swift implementation of OpenAI's Whisper running entirely on-device for iOS/macOS - no cloud calls, with real-time streaming and voice activity detection

Privacy-conscious developers, rejoice. WhisperKit brings OpenAI’s powerful Whisper speech recognition directly to Apple devices without sending audio data anywhere. While most speech-to-text solutions require internet connectivity and raise privacy concerns, this runs completely offline on Apple Silicon, leveraging the Neural Engine for efficient processing.

Beyond basic transcription, WhisperKit delivers real-time streaming transcription, word-level timestamps, and voice activity detection - features typically reserved for cloud services. It supports everything from iPhone to Vision Pro, with optimized models for different device capabilities. The 5,500+ stars reflect strong adoption among iOS developers who need reliable speech recognition without the latency, costs, or privacy trade-offs of cloud APIs.

Perfect for voice note apps, accessibility features, or any iOS app needing speech recognition. The Swift Package Manager integration means you can add it to existing projects in minutes, and the TestFlight demo app lets you experience the quality firsthand before committing.

⭐ Stars: 5557
💻 Language: Swift
🔗 Repository: argmaxinc/WhisperKit