DEV Community

Cover image for Open-Source Local AI Meeting Assistant - Yogurt
GitHubOpenSource
GitHubOpenSource

Posted on

Open-Source Local AI Meeting Assistant - Yogurt

TL;DR: Yogurt records call audio and microphone input locally, running transcription via whisper.cpp and Apple SpeechAnalyzer, while generating summaries through llama.cpp. It avoids injecting third-party bots into your meetings by capturing system audio directly.
Best for: Developers and privacy-conscious users looking for a local-first meeting assistant.
Repository snapshot: ⭐ 1 493 stars · 🍴 81 forks · TypeScript · MIT
Daily picks: Follow GitHub Open Source on Telegram


What it is

Yogurt is a free, open-source desktop application that records, transcribes, and summarizes meetings entirely on your local machine. Built with TypeScript, it serves as a local-first alternative to cloud-dependent meeting assistants.

How it works

Instead of injecting a third-party bot into your video calls, the application captures system audio and microphone input directly at the operating system level. It routes the recorded audio through local engines like whisper.cpp and Apple SpeechAnalyzer for transcription, and processes the text through llama.cpp to generate structured summaries using open-weight models.

Key features

  • Direct system audio capture without meeting bots
  • Offline transcription supporting 99 languages
  • Local summary generation using open-weight models
  • Optional local model context protocol server integration
  • Cross-platform support for Mac, Windows, and Linux

Getting started

Clone the repository from GitHub and ensure you have the necessary runtime dependencies installed for local execution. Follow the setup instructions in the readme to configure your preferred local models.

When not to use it

Skip this tool if your computer lacks the hardware resources required to run local transcription and open-weight language models efficiently, or if you rely strictly on cloud-hosted enterprise suites.


Discuss

Which local model size do you usually run for everyday summarisation tasks?

Curated from the repository README and metadata. No hands-on test is claimed; evaluate it in a sandbox before production use.

Top comments (0)