Edge TTS MCP

import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js"; import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js"; import { randomUUID } from "crypto"; import { z } from "zod"; import { EdgeTTS } from "@andresaya/edge-tts"; import symphonia from "@tropicbliss/symphonia" // Create an MCP server const server = new McpServer({ name: "edge-tts-bun-mcp", version: "1.0.0" }); const generateSpeechArgsSchema = { input: z.string().describe("the text to generate audio from"), }; // Add an addition tool server.tool( "speech_text_aloud", "generate lifelike audio from text, and having it read aloud. If there are no special requirements for emotions and tones, pitch and other options remain default", generateSpeechArgsSchema, async (args: any) => { try { const tts = new EdgeTTS(); const input = args.input; const isSaveAudio = process.env.SAVE_AUDIO || false; const voice = process.env.VOICE || "zh-CN-XiaoxiaoNeural"; const rate = process.env.RATE || "0%"; const volume = process.env.VOLUME || "0%"; const pitch = process.env.PITCH || "0Hz"; await tts.synthesize(`${input}`, voice, { rate, volume, pitch }); const audioBase64 = tts.toBase64(); const audioBuffer = Buffer.from(audioBase64, 'base64') symphonia.playFromBuf(audioBuffer, { isBlocking: true }); // Play the audio buffer // 如果process.env.SAVE_AUDIO为true，则保存音频文件 if (isSaveAudio === 'true') { try { const uuid = randomUUID(); await tts.toFile(`${uuid}`); } catch (error) { console.error("Error saving audio file:", error); } } return { content: [{ type: "text", text: "Audio generated and played successfully" }] }; } catch (error) { return { content: [{ type: "text", text: "Error generating audio" }] }; } } ); // Start receiving messages on stdin and sending messages on stdout async function main() { const transport = new StdioServerTransport(); console.log("Starting server..."); await server.connect(transport); } main().catch((error) => { console.error("Fatal error in main():", error); process.exit(1); });