mcp-tts-kokoro mcp

v1.0.1 · MCP Tool · audio · registry.pascalai.org

Text-to-speech via Kokoro local TTS server (OpenAI-compatible API). Fast, offline, no API key required. Runs at http://localhost:8880 by default. Many voices: af_bella, af_sarah, am_adam, bf_emma, bm_george, etc.

ttskokorolocalofflinespeechaudio

Input Parameters

Parameter	Type	Description
textrequired	string	Text to convert to speech.
voiceoptional	string	Voice name: af_bella, af_sarah, am_adam, bf_emma, bm_george, etc. Default: af_bella. Default: `af_bella`.
modeloptional	string	Model name (default: kokoro). Some servers use tts-1, kokoro-v1-0, etc. Default: `kokoro`.
outputPathoptional	string	Save audio to this file path. Returns base64 audio if empty.
speedoptional	number	Speech speed 0.25–4.0 (default: 1.0). Default: `1.0`.
formatoptional	string	Output format: mp3 (default), wav, opus, flac. Default: `mp3`.
apiKeyoptional	string	Unused — Kokoro runs locally without authentication.
baseUrloptional	string	Kokoro server URL (default: http://localhost:8880). Default: `http://localhost:8880`.

Output Fields

Field	Type	Description
ok	boolean
audio	string	Base64-encoded audio (when outputPath not set).
outputPath	string	File path where audio was saved.
format	string
size	integer
error	string

Examples

Synthesize with Kokoro local server

// Input
{
  "text": "Kokoro is a fast offline text-to-speech model with natural-sounding voices.",
  "voice": "af_sarah",
  "speed": 1.0
}

// Output
{
  "ok": true,
  "audio": "//NExAA...",
  "format": "mp3",
  "size": 32768
}

Save to file with British male voice

// Input
{
  "text": "Good evening, sir. Your car is ready.",
  "voice": "bm_george",
  "outputPath": "C:\\audio\\evening.mp3"
}

// Output
{
  "ok": true,
  "outputPath": "C:\\audio\\evening.mp3",
  "format": "mp3",
  "size": 18432
}

Install & Discovery

Install

ppm install mcp-tts-kokoro

Get JSON Schema

GET /v1/packages/mcp-tts-kokoro/1.0.1/schema

Discover by keyword

GET /v1/mcp/discover?q=tts

Discovery hint: Install with ppm install mcp-tts-kokoro or invoke remotely via POST /v1/invoke/mcp-tts-kokoro on the MCP Service.