Recap is an open-source, macOS-native application designed to automatically transcribe and summarize audio from meetings and other applications, prioritizing user privacy by processing data locally. It targets macOS users who need to focus on meetings without sacrificing productivity, offering a privacy-first alternative to cloud-based solutions.

How It Works

Recap captures system audio and optionally microphone input using macOS Core Audio and AVAudioEngine. It leverages WhisperKit (MLX) for local speech-to-text transcription and then utilizes Ollama (or OpenRouter as a cloud fallback) for generating summaries. This local-first approach ensures sensitive meeting data remains on the user's device.

Quick Start & Requirements

• Installation: Compile from source using Xcode 15.0+. Clone the repository and build the project.
• Prerequisites: macOS 15.0+, Xcode 15.0+. For local processing, Ollama is required. Apple Silicon Macs (M1 or newer) are recommended. Intel Macs are not officially supported.
• Setup: Requires setting Hugging Face and optionally OpenRouter API keys via Xcode Scheme Editor. Ollama setup involves downloading and pulling models.
• Links: GitHub Repository

Highlighted Details

• macOS-native design using Swift and SwiftUI.
• Local processing via WhisperKit (MLX) and Ollama for enhanced privacy.
• Supports system audio capture and optional microphone input.
• Roadmap includes meeting detection, custom prompts, and live transcription.

Maintenance & Community

The project is actively seeking contributions to stabilize core features. Users can watch the repository for release notifications.

Licensing & Compatibility

The software is provided under a custom license that permits any use, modification, and deployment (commercial or otherwise), but prohibits redistribution of the original or modified project without express prior written consent.

Limitations & Caveats

Recap is currently in an incomplete, proof-of-concept state and not recommended for production or daily usage. Pre-built releases are not yet available, and Intel Macs are not officially supported. Some features like meeting detection are still in progress.