Skip to content
Yap MCP

Yap MCP

by O-mega Enterprise Inc.
Released
v1.0.0
MIT
Repository Docs
Productivity

Summary

Transcribe local audio and video files on-device from Claude, Cursor or VS Code, and search your Yap dictation history. No upload, no API key.

Features

  • Transcribes local audio and video files of any length
  • Automatic language detection across 25 European languages
  • Optional start-time stamps per passage
  • Runs fully on-device with NVIDIA Parakeet TDT v3, no API key
  • Searches the Yap app dictation history read-only
  • One-click .mcpb install for Claude Desktop

Installation

Set up this MCP server in your favorite AI agent — copy a ready-made configuration below.

Any MCP-compatible agent

Most agents (Claude, Cursor, Windsurf, VS Code, and more) read a standard mcpServers configuration.

  1. Open your agent's MCP configuration file.
  2. Merge the snippet below into it.
  3. Restart the agent — the "Yap MCP" tools will be available.
{
  "mcpServers": {
    "yap-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "github:founden-bot/yap-mcp"
      ]
    }
  }
}

Description

Yap MCP is a free, MIT-licensed MCP server that gives any MCP client the ability to turn audio into text without sending a byte to the cloud. It runs NVIDIA Parakeet TDT v3 through sherpa-onnx on your own CPU, so recordings such as interviews, voice memos, meetings and video soundtracks are decoded in memory and discarded after transcription. The server exposes three tools: transcribe_audio handles local WAV, MP3, M4A, MP4, FLAC, OGG, Opus, WebM and similar files of any length, detects the language automatically among 25 European languages and can add start-time stamps; search_dictations performs read-only searches over the dictation history kept by the Yap desktop app on macOS and Windows; prepare_speech_model downloads and verifies the roughly 670 MB speech model ahead of first use. After that one-time download everything works offline, at around ten times real-time speed on a modern laptop. It needs Node.js 20 or newer (22.13+ for dictation search) and runs on macOS, Windows and Linux. Install it in Claude Desktop with the one-click .mcpb bundle, with claude mcp add in Claude Code, as a Claude Code plugin, as a Gemini CLI extension, or as a plain npx stdio server in Cursor, Windsurf, VS Code, Cline and Zed. A companion yap-transcribe command-line tool does the same job from the terminal. It is a good fit when privacy rules out cloud transcription APIs.

Related MCP Servers

MCP: Filesystem

by Model Context Protocol

Secure local file operations for agents — read, write, search, and edit files within configurable allowed directories.

DevelopersLocal-First

MCP: VoiceStudio

by Palash Deb

Local speech MCP server bundled with VoiceStudio — generate speech, clone a voice from reference audio, and transcribe in 646 languages, all on your own machine with no API key.

Developer Tools

MCP: ElevenLabs

by ElevenLabs

Featured

ElevenLabs' official MCP server — generate speech, clone voices, transcribe audio, and build voice agents.

Developer Tools

MCP: AudioPod MCP

by AudioPod AI

New

Hosted audio-AI tools for agents: speaker-labelled transcription, stem splitting, noise removal, MIDI conversion, speech and music. No install or GPU needed.

Files & Storage
Browse all MCP servers →