Vexa provides a self-hosted, multi-user API for real-time meeting transcription, targeting enterprises prioritizing data security and compliance. It offers an alternative to services like recall.ai, enabling programmatic bot integration into Google Meet for live transcripts and translations.

How It Works

Vexa utilizes a scalable architecture to support numerous concurrent transcription sessions. It integrates bots directly into meetings, capturing audio streams for real-time processing. The system leverages Whisper for multilingual transcription and offers translation across 99 languages, aiming for low latency.

Quick Start & Requirements

• Install/Run: Access the public API via self-service keys at vexa.ai. Setup is described as taking under 5 minutes.
• Prerequisites: API key required.
• Links: Vexa Website, Example Client

Highlighted Details

• Real-time transcription and translation for Google Meet.
• Supports 99 languages via Whisper.
• Public API available with self-service keys.
• Self-deployment options for enhanced security.

Maintenance & Community

• Active development with a public roadmap and task board.
• Community engagement encouraged via Discord.
• Key contributor: @grankin_d (LinkedIn, X).

Licensing & Compatibility

• Licensed under Apache License, Version 2.0.
• Permits commercial use and integration with closed-source applications. Vexa name/logo are trademarks of Vexa.ai Inc.

Limitations & Caveats

Currently supports Google Meet; Zoom and Microsoft Teams integrations are planned for April/May 2025. Speaker identification is under development. Future features like LLM processing and RAG are planned.