Local Speech-to-Text Model Context Protocol (MCP) Server 🚀

What is Local Speech-to-Text Model Context Protocol (MCP) Server?

It is a high-performance Model Context Protocol (MCP) server that uses whisper.cpp to convert spoken language into text. Designed for Apple Silicon devices, it processes audio entirely on-device, preserving privacy while providing fast transcription.

How to use Local Speech-to-Text Model Context Protocol (MCP) Server?

Clone the repository, then install dependencies via Homebrew (brew install cmake whisper). Build the project with CMake and make, then run the server with ./local-stt-mcp. Send an audio file as a POST request to http://localhost:8080/transcribe to receive the transcribed text in a JSON response.

Key features of Local Speech-to-Text Model Context Protocol (MCP) Server

• Optimized for Apple Silicon for high performance.
• Full local processing – no data leaves your device.
• Simple RESTful API for easy integration.
• Supports transcription in multiple languages.
• Customizable via a config.json file (language, model, max length).

Use cases of Local Speech-to-Text Model Context Protocol (MCP) Server

• Real-time transcription of meetings or lectures.
• Voice command recognition in desktop applications.
• Generating subtitles for locally stored media.
• Accessibility tools for converting speech to text offline.

FAQ from Local Speech-to-Text Model Context Protocol (MCP) Server

What hardware is required?

Apple Silicon (M1 or later) is required for optimal performance.

What dependencies are needed?

You need cmake and whisper (via Homebrew) to build and run the server.

Does the server