Free AI Meeting Summarizer for macOS Users – Recap

Recap is an open-source AI-powered meeting summarization tool that processes audio recordings locally on your Mac to generate transcripts and summaries.

The tool records system audio from any application on your Mac, transcribes the content using Apple’s WhisperKit technology, and creates summaries through local AI models.

This approach keeps your sensitive meeting data completely private, never sending information to external servers unless you specifically choose cloud-based processing options.

Features

Recap offers several core capabilities for meeting documentation and analysis:

• System Audio Recording: Captures audio from any macOS application including Zoom, Microsoft Teams, Google Meet, and other conferencing platforms.
• Local Transcription: Uses WhisperKit (powered by Apple’s MLX framework) to convert audio to text without sending data to external servers.
• AI Summarization: Processes transcripts through Ollama for completely private summaries or OpenRouter for cloud-based options when local compute power is limited.
• Dual Audio Support: Records both system audio and microphone input simultaneously, capturing both incoming audio and your own voice contributions.
• Custom Prompts: Future capability to customize summarization prompts based on specific meeting types or organizational needs.

Use Cases

• Product Development Meetings: Record feature discussions, technical decisions, and roadmap planning sessions while maintaining focus on complex technical conversations.
• Client Consultations: Capture detailed requirements, feedback, and project specifications without constantly switching between note-taking and active participation.
• Training Sessions: Document educational content, workshops, and knowledge transfer sessions for future reference and team sharing.
• Strategic Planning: Record executive discussions, business planning meetings, and sensitive strategic conversations with complete privacy assurance.
• Remote Team Coordination: Summarize daily standups, sprint planning, and cross-team collaboration sessions for team members in different time zones.

How To Use It

1. Recap is built for Apple Silicon (M1, M2, etc.) and requires macOS 15.0 or later. It won’t work on Intel Macs. You also need to install Xcode, Apple’s software for building apps.

2. Clone Recap’s source code from its GitHub repository. You’ll do this using the command line.

git clone https://github.com/rawandahmad698/recap.git
cd recap

3. Recap uses models from Hugging Face for transcription. You need a free account to create an access token. Go to the Hugging Face documentation to learn how to get your token.

4. In your terminal, while inside the recap folder, type this command to open the project in Xcode:
bash open Recap.xcodeproj

5. In Xcode, find the active scheme at the top of the window (it should say “Recap”). Click on it and select “Edit Scheme…”.

6. Add the Variable:

• In the window that appears, select the Run tab on the left.Click on the Arguments tab at the top.Under the “Environment Variables” section, click the + button to add a new variable.For the Name, enter HF_TOKEN.For the Value, paste the Hugging Face token you created earlier.

If you plan to use the cloud-based OpenRouter for summarization instead of running a model locally, you would also add OPENROUTER_API_KEY here with your key from OpenRouter.

7. For the best privacy, Recap uses Ollama to run summarization models directly on your Mac.

1. Download Ollama: Go to the Ollama website and download the application for macOS.
2. Install a Model: After installing Ollama, open your Terminal and run the following command to download the Llama 3 model: ollama pull llama3

8. Go back to the Xcode window. Press the “Play” button in the top-left corner or use the keyboard shortcut ⌘+R. Xcode will now compile the source code and launch Recap. The first time you run it, you need to go into the app’s settings to select the Whisper model you want to use for transcription. After that, you’re all set to start recording.

9. Select your target application from the dropdown menu when you’re ready to record. This tells Recap which audio source to capture from your system.

10. Click the record button to begin capturing audio. The application will record system audio automatically and can optionally include your microphone input if you want to capture your own voice contributions.

11. Focus on your meeting without worrying about note-taking. Recap handles all audio capture in the background without interrupting your workflow.

12. Stop recording when your meeting ends. The application saves the audio files and begins background processing for transcription and summarization.

13. Check back later to review your generated transcript and summary. Processing time depends on recording length and your chosen models.

Pros

• Complete Privacy: With the recommended Ollama setup, no audio or text ever leaves your computer.
• No Subscription Fees: As an open-source tool, it’s free to use.
• High-Quality Transcription: It uses WhisperKit, which is optimized for Apple Silicon and provides excellent transcription accuracy.

Cons

• Not for Beginners: The installation process requires using Xcode and the command line.
• High System Requirements: For local processing with Ollama, a powerful Apple Silicon Mac with at least 16 GB of RAM is recommended.
• macOS Only: It is built specifically for macOS and does not support Windows or Linux. Intel Macs are not officially supported.

Related Resources

• Ollama Documentation: Comprehensive guide for setting up and managing local AI models for summarization.
• WhisperKit GitHub Repository: Technical documentation for the transcription engine powering Recap.
• Hugging Face Token Guide: Step-by-step instructions for creating the required authentication token.
• OpenRouter API Documentation: Setup guide for cloud-based summarization when local processing isn’t sufficient.
• Core Audio Programming Guide: Apple’s official documentation for understanding macOS audio capture capabilities.

FAQs

Q: Does Recap work with all video conferencing platforms?
A: Yes, Recap records system audio from any macOS application, including Zoom, Microsoft Teams, Google Meet, Slack, and other platforms. It captures audio regardless of the specific meeting software you use.

Q: How long does processing take after recording?
A: Processing time depends on recording length and your chosen models. Local processing on Apple Silicon typically handles one hour of audio in 10-15 minutes, though this varies based on your Mac’s specifications and selected model sizes.

Q: Is my data safe with Recap?
A: Yes. When using the recommended setup with Ollama, all recording, transcription, and summarization happens locally on your Mac. Your data never leaves your device. If you choose to use OpenRouter, your data will be sent to their servers for processing.