Add loudreader/loudkit MCP server

data/server-metadata/github-loudreader-loudkit-788794b9.json

@@ -0,0 +1,60 @@
+{
+  "$schema": "../../schemas/server-metadata.schema.json",
+  "id": "github-loudreader-loudkit-788794b9",
+  "source": {
+    "issue": 2310,
+    "projectUrl": "https://github.com/loudreader/loudkit"
+  },
+  "description": "Local stdio MCP server for offline text-to-speech with 28 bundled voices across 10 languages, voice profiles, long-text continuations, and WAV/FLAC/MP3/Opus audio output.",
+  "category": "Multimedia Processing",
+  "links": {
+    "docs": "https://github.com/loudreader/loudkit/blob/main/docs/guides/04-server-and-agents.md"
+  },
+  "install": {
+    "commands": [
+      "uvx --from \"loudkit[mcp,hub]\" loudkit serve --mcp --checkpoint loudreader/loudr-1"
+    ],
+    "env": [],
+    "confidence": "medium"
+  },
+  "transport": [
+    "stdio"
+  ],
+  "auth": {
+    "type": "none",
+    "notes": []
+  },
+  "clients": [
+    "Claude Code",
+    "Claude Desktop",
+    "Cursor",
+    "Cline"
+  ],
+  "tools": {
+    "count": 3,
+    "names": [
+      "list voices",
+      "synthesize speech",
+      "describe loaded model"
+    ],
+    "source": "self_reported"
+  },
+  "license": "Apache-2.0",
+  "verification": {
+    "status": "partial",
+    "notes": [
+      "GitHub repository and PyPI package were reachable during review.",
+      "README documents `loudkit serve --mcp` as a local stdio MCP server."
+    ]
+  },
+  "community": {
+    "maintainedBy": [
+      "@loudreader",
+      "@pepinu"
+    ],
+    "verifiedBy": [
+      "@TensorBlock"
+    ],
+    "claimed": true
+  }
+}

docs/multimedia-processing.md

@@ -221,3 +221,4 @@ Servers focused on generating or manipulating images, processing video, audio tr
 - [ArthurBrioche/cadre-video-editor-plugin](https://github.com/ArthurBrioche/cadre-video-editor-plugin): Official MIT-licensed local Claude plugin and stdio MCP bridge for Cadre, a proprietary Apple Silicon macOS screen recorder and non-destructive video editor. With Cadre open, it exposes 57 tools for recording, timeline inspection, cuts, cinematic zooms, on-device captions, styling, saving, and MP4 export through an authenticated loopback Agent API. Install: download the v1.0.2 plugin ZIP for Claude Desktop or Cowork, or clone the repository for Claude Code. Requires macOS 13+, Node.js 22, 24, or 26, and Cadre Pro only for export.
 - [speakai/speakai-mcp](https://github.com/speakai/speakai-mcp): Official hosted server for Speak AI, transcribing and analyzing audio and video recordings in a Speak AI workspace — speech-to-text in 100+ languages with speaker labels, AI-generated summaries and sentiment, call scoring against custom rubrics, clip creation, and captions. Streamable HTTP at `https://api.speakai.co/v1/mcp`; OAuth 2.1 or `Authorization: Bearer <API key>`. Also installable via `npx @speakai/mcp-server`. TypeScript, MIT.
 - [Synthenova/infiknit-mcp](https://github.com/Synthenova/infiknit-mcp): Local stdio launcher for Infiknit desktop. Agents generate images and videos into a local project and build canvas workflows on the same computer; cloud agents cannot reach it. Install the `infiknit.mcpb` release bundle or clone the repo and run `node plugins/infiknit/bin/infiknit-mcp.cjs` while Infiknit is open. Clients: Claude Code, Codex, Cursor, Grok Build, VS Code, Cline, Copilot, and Gemini CLI. Official registry id `io.github.Synthenova/infiknit`; MIT.
+- [loudreader/loudkit](https://github.com/loudreader/loudkit): Lets agents speak text aloud with a local text-to-speech engine: list the 28 built-in voices (10 languages) and voice profiles cloned with the loudkit CLI, synthesize speech as base64 WAV, FLAC, MP3 or Opus with its duration and a continuation for reading long texts in pieces, and describe the loaded model. Install: `uvx --from "loudkit[mcp,hub]" loudkit serve --mcp --checkpoint loudreader/loudr-1` (stdio; no cloud, no API key). Apache-2.0.