voice-mcp
Provides text-to-speech synthesis using ElevenLabs, allowing AI assistants to generate realistic multilingual speech (29 languages) with configurable cloned or pre-made voices and inline audio playback.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@voice-mcpSay 'Hello, welcome to our demo' in a friendly tone"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
🎙️ voice-mcp (ElevenLabs Edition)
An MCP server for AI voice synthesis, deployed on Cloudflare Workers. Give your AI assistant a realistic cloned voice — works on both desktop and mobile.
Based on garan0613/voice-mcp, swapping MiniMax TTS for ElevenLabs to get multilingual, high-quality voice cloning.
English
Features
🎤 ElevenLabs TTS — Multilingual v2 model, supports 29 languages including Chinese & English
🔊 Inline Audio Player — WeChat-style player with waveform animation & dark mode
⚡ Cloudflare Workers — Serverless, free tier friendly, globally fast
📱 Works on Mobile — Audio plays directly in claude.ai on any device
📝 Transcript Toggle — Show/hide the spoken text below the player
Quick Start
1. Clone
git clone https://github.com/yukilain007/voice-mcp-elevenlabs.git
cd voice-mcp-elevenlabs2. Install
npm install3. Configure Secrets
You'll need an ElevenLabs account with a voice (cloned or pre-made).
npx wrangler login
npx wrangler secret put ELEVENLABS_API_KEY
npx wrangler secret put VOICE_ID
npx wrangler secret put BOT_NAME # Optional, defaults to "AI"4. Deploy
npx wrangler deploy5. Connect to Claude.ai
Go to Settings → Connectors → Add Connector
Enter your Worker URL:
https://your-worker.workers.dev/mcpDone! The
speaktool is now available.
Configuration
Variable | Required | Description |
| ✅ | Your ElevenLabs API key |
| ✅ | Voice ID (cloned or pre-made) |
| ❌ | Display name in the player (default: "AI") |
API Endpoints
Endpoint | Description |
| MCP server (SSE protocol) |
| Direct audio file |
| Health check |
How to Get a Voice ID
Go to ElevenLabs
Voices → pick a pre-made voice, or clone your own (upload 10–30s of clear audio)
Click the voice → copy the Voice ID
Differences from the Original
Original (garan0613) | This Fork | |
TTS Provider | MiniMax | ElevenLabs |
Model | speech-2.8-hd | eleven_multilingual_v2 |
Languages | Chinese-focused | 29 languages |
Voice Cloning | MiniMax Console | ElevenLabs Console |
Player UI, Cloudflare deployment, and MCP protocol are unchanged.
Custom Domain
// wrangler.jsonc
{
"routes": [
{ "pattern": "voice.yourdomain.com/*", "zone_name": "yourdomain.com" }
]
}Related MCP server: Drizz Voice Generator
中文说明
一个部署在 Cloudflare Workers 上的 MCP 语音合成服务器。让你的 AI 助手拥有逼真的克隆声音——桌面端和手机端都能用。
基于 garan0613/voice-mcp 修改,将 MiniMax TTS 替换为 ElevenLabs,支持多语言高质量语音克隆。
特性
🎤 ElevenLabs TTS — 多语言 v2 模型,支持中文、英文等 29 种语言
🔊 内联音频播放器 — 仿微信语音条样式,带波形动画和暗色模式
⚡ Cloudflare Workers — 无服务器部署,免费额度内可用,全球加速
📱 手机可用 — 在 claude.ai 的任何设备上都能直接播放
📝 文字折叠 — 播放器下方可展开/收起语音文本
快速开始
1. 克隆仓库
git clone https://github.com/yukilain007/voice-mcp-elevenlabs.git
cd voice-mcp-elevenlabs2. 安装依赖
npm install3. 配置密钥
需要一个 ElevenLabs 账号和一个声音 ID(可以用预设声音,也可以上传录音克隆)。
npx wrangler login
npx wrangler secret put ELEVENLABS_API_KEY
npx wrangler secret put VOICE_ID
npx wrangler secret put BOT_NAME # 可选,默认 "AI"4. 部署
npx wrangler deploy5. 连接 Claude.ai
打开 Settings → Connectors → Add Connector
填入你的 Worker URL:
https://your-worker.workers.dev/mcp完成!在对话中就能使用
speak工具了
配置项
变量 | 必填 | 说明 |
| ✅ | ElevenLabs API 密钥 |
| ✅ | 声音 ID(克隆或预设) |
| ❌ | 播放器显示名称(默认 "AI") |
获取 Voice ID
打开 ElevenLabs
进入 Voices → 选一个预设声音,或上传 10-30 秒清晰录音克隆自己的
点击声音 → 复制 Voice ID
与原版的区别
原版 (garan0613) | 本 Fork | |
TTS 引擎 | MiniMax | ElevenLabs |
模型 | speech-2.8-hd | eleven_multilingual_v2 |
语言支持 | 以中文为主 | 29 种语言 |
声音克隆 | MiniMax 控制台 | ElevenLabs 控制台 |
播放器界面、Cloudflare 部署方式、MCP 协议部分不变。
Tech Stack
Cloudflare Workers — Serverless runtime
MCP SDK — Model Context Protocol
ElevenLabs — Voice synthesis
ext-apps — Inline UI rendering
License
MIT © 2026
Credits
Original project: garan0613/voice-mcp
TTS: ElevenLabs
Runtime: Cloudflare Workers
This server cannot be deployed
Maintenance
Related MCP Connectors
ElevenLabs in natural language: generate speech in any language, create and manage voices, compose m
Generate AI images, videos, music, SFX & speech in any AI assistant. Results appear inline in chat.
Manage ElevenLabs voice agents and generate speech, music, sound effects, images, and video.
- ChamadeOAuthio.chamade
Voice and chat for AI agents — Discord, Teams, Meet, Slack, Zoom, Telegram, WhatsApp, NC Talk, SIP
Related MCP Servers
- AlicenseAqualityFmaintenanceEnables text-to-speech conversion using ElevenLabs API with voice management, streaming support, and multiple models.51MIT
- FlicenseAqualityDmaintenanceEnables text-to-speech audio generation using ElevenLabs voices directly from Claude conversations, supporting single and batch conversion, voice listing, and voice ID lookup.5-
- FlicenseNot gradedqualityDmaintenanceEnables Claude to speak text with an embedded audio player, supporting 54 voices, voice cloning, and playback controls, all running locally.2-
- AlicenseNot gradedqualityCmaintenanceEnables AI assistants to generate speech with custom cloned voices via DashScope or ElevenLabs, with an inline audio player and visualizer panel.MIT