@ww_nero/media 1.0.9 → 1.0.10

package/README.md

@@ -1,6 +1,6 @@
 # Media MCP Server
 
-媒体处理 MCP 服务，提供语音识别等功能。
+媒体处理 MCP 服务，提供语音识别和语音合成功能。
 
 ## 功能
 
@@ -18,13 +18,27 @@
 - 超出限制需要外部分段处理后逐段识别
 
 **输出：**
-- 识别结果保存到工作目录下的 `transcribe.srt` 文件
+- 识别结果保存到工作目录下的 `asr_<timestamp>.srt` 文件
+
+### tts - 语音合成
+
+将文本转换为音频文件。
+
+**参数：**
+- `working_directory`: 工作目录的绝对路径，合成的音频文件将保存到此目录
+- `text`: 需要合成语音的文本内容
+- `voice`: （可选）音色模型，默认 `sambert-zhimiao-emo-v1`
+- `format`: （可选）输出音频格式，可选 mp3（默认）、wav、pcm
+- `sample_rate`: （可选）采样率，默认 16000
+
+**输出：**
+- 合成结果保存到工作目录下的 `tts_<timestamp>.<format>` 文件
 
 ## 环境变量
 
 | 变量名 | 说明 | 必填 |
 |--------|------|------|
-| `ASR_API_KEY` | 阿里云 DashScope API Key | 是 |
+| `DASHSCOPE_API_KEY` | 阿里云 DashScope API Key | 是 |
 
 ## 安装
 
@@ -41,7 +55,7 @@ npm install
       "command": "node",
       "args": ["/path/to/media/index.js"],
       "env": {
-        "ASR_API_KEY": "your-api-key"
+        "DASHSCOPE_API_KEY": "your-api-key"
       }
     }
   }

package/index.js

@@ -9,185 +9,10 @@ const {
   ListToolsRequestSchema,
 } = require('@modelcontextprotocol/sdk/types.js');
 
-const ASR_API_KEY = process.env.ASR_API_KEY || '';
-const ASR_UPLOAD_URL = 'http://fsheep.com:10808/upload';
-const ASR_SERVICE_URL = 'https://dashscope.aliyuncs.com/api/v1/services/audio/asr/transcription';
+const { asr, SUPPORTED_AUDIO_TYPES, MAX_FILE_SIZE_BYTES } = require('./utils/asr.js');
+const { tts } = require('./utils/tts.js');
 
-const SUPPORTED_AUDIO_TYPES = ['.mp3', '.wav'];
-const MAX_FILE_SIZE_BYTES = 120 * 1024 * 1024; // 120MB
-const LANGUAGE_HINTS = ['zh', 'en', 'ja'];
-
-/**
- * 将毫秒转换为 SRT 时间格式 HH:MM:SS,mmm
- */
-const msToSrtTime = (ms) => {
-  const hours = Math.floor(ms / 3600000);
-  const minutes = Math.floor((ms % 3600000) / 60000);
-  const seconds = Math.floor((ms % 60000) / 1000);
-  const milliseconds = ms % 1000;
-  return `${String(hours).padStart(2, '0')}:${String(minutes).padStart(2, '0')}:${String(seconds).padStart(2, '0')},${String(milliseconds).padStart(3, '0')}`;
-};
-
-/**
- * 将 ASR 识别结果转换为 SRT 字幕内容
- */
-const asrToSrt = (asrData) => {
-  const srtEntries = [];
-  let subtitleIndex = 1;
-
-  for (const item of asrData) {
-    const transcription = item.transcription || {};
-    const transcripts = transcription.transcripts || [];
-
-    for (const transcript of transcripts) {
-      const sentences = transcript.sentences || [];
-
-      for (const sentence of sentences) {
-        const beginTime = sentence.begin_time || 0;
-        const endTime = sentence.end_time || 0;
-        const text = (sentence.text || '').trim();
-
-        if (text) {
-          const startStr = msToSrtTime(beginTime);
-          const endStr = msToSrtTime(endTime);
-
-          srtEntries.push(`${subtitleIndex}\n${startStr} --> ${endStr}\n${text}\n`);
-          subtitleIndex++;
-        }
-      }
-    }
-  }
-
-  return srtEntries.join('\n');
-};
-
-/**
- * 上传音频文件到服务器
- */
-const uploadAudio = async (uploadUrl, audioPath) => {
-  const fileBuffer = fs.readFileSync(audioPath);
-  const fileName = path.basename(audioPath);
-
-  const formData = new FormData();
-  formData.append('file', new Blob([fileBuffer]), fileName);
-
-  const response = await fetch(uploadUrl, {
-    method: 'POST',
-    body: formData,
-  });
-
-  if (!response.ok) {
-    throw new Error(`上传失败: ${response.status} - ${await response.text()}`);
-  }
-
-  const data = await response.json();
-
-  if (!data.success || !data.data || !data.data.path) {
-    throw new Error(`上传响应格式错误: ${JSON.stringify(data)}`);
-  }
-
-  return data.data.path;
-};
-
-/**
- * 根据上传接口 URL 和文件路径构建静态资源 URL
- */
-const getStaticUrl = (uploadUrl, filePath) => {
-  const url = new URL(uploadUrl);
-  return `${url.protocol}//${url.host}${filePath}`;
-};
-
-/**
- * 提交 ASR 转写任务
- */
-const submitAsrTask = async (fileUrls, apiKey) => {
-  const response = await fetch(ASR_SERVICE_URL, {
-    method: 'POST',
-    headers: {
-      'Authorization': `Bearer ${apiKey}`,
-      'Content-Type': 'application/json',
-      'X-DashScope-Async': 'enable',
-    },
-    body: JSON.stringify({
-      model: 'paraformer-v2',
-      input: { file_urls: fileUrls },
-      parameters: {
-        channel_id: [0],
-        language_hints: LANGUAGE_HINTS,
-      },
-    }),
-  });
-
-  if (!response.ok) {
-    const errorText = await response.text();
-    throw new Error(`ASR 任务提交失败: ${response.status} - ${errorText}`);
-  }
-
-  const data = await response.json();
-
-  if (!data.output || !data.output.task_id) {
-    throw new Error(`ASR 响应格式错误: ${JSON.stringify(data)}`);
-  }
-
-  return data.output.task_id;
-};
-
-/**
- * 轮询等待 ASR 任务完成
- */
-const waitForTaskComplete = async (taskId, apiKey, timeoutMs = 5 * 60 * 1000) => {
-  const taskUrl = `https://dashscope.aliyuncs.com/api/v1/tasks/${taskId}`;
-  const startTime = Date.now();
-  const pollInterval = 500;
-
-  while (Date.now() - startTime < timeoutMs) {
-    const response = await fetch(taskUrl, {
-      method: 'GET',
-      headers: {
-        'Authorization': `Bearer ${apiKey}`,
-      },
-    });
-
-    if (!response.ok) {
-      throw new Error(`查询任务状态失败: ${response.status}`);
-    }
-
-    const data = await response.json();
-    const status = data.output?.task_status;
-
-    if (status === 'SUCCEEDED') {
-      return data.output.results || [];
-    } else if (status === 'FAILED') {
-      throw new Error('ASR 识别任务失败');
-    }
-
-    await new Promise(resolve => setTimeout(resolve, pollInterval));
-  }
-
-  throw new Error('ASR 识别超时');
-};
-
-/**
- * 获取转写结果详情
- */
-const fetchTranscriptionResults = async (results) => {
-  const allTranscriptions = [];
-
-  for (const result of results) {
-    if (result.subtask_status === 'SUCCEEDED' && result.transcription_url) {
-      const response = await fetch(result.transcription_url);
-      if (response.ok) {
-        const transcriptionData = await response.json();
-        allTranscriptions.push({
-          file_url: result.file_url,
-          transcription: transcriptionData,
-        });
-      }
-    }
-  }
-
-  return allTranscriptions;
-};
+const DASHSCOPE_API_KEY = process.env.DASHSCOPE_API_KEY || '';
 
 /**
  * 将 Windows 路径转换为 WSL 路径，或反之
@@ -286,59 +111,10 @@ const resolveAudioFile = (workingDir, rawPath) => {
   return resolved;
 };
 
-/**
- * ASR 语音识别
- */
-const asr = async ({ working_directory, audio_file }) => {
-  // 验证环境变量
-  if (!ASR_API_KEY) {
-    throw new Error('请配置 ASR_API_KEY 环境变量');
-  }
-
-  const workingDir = resolveWorkingDirectory(working_directory);
-  const audioPath = resolveAudioFile(workingDir, audio_file);
-
-  // 检查文件大小
-  const fileSize = fs.statSync(audioPath).size;
-  if (fileSize > MAX_FILE_SIZE_BYTES) {
-    throw new Error(
-      `音频文件大小 ${(fileSize / 1024 / 1024).toFixed(2)}MB 超过限制（最大 120MB）。` +
-      `请先对音频进行分段处理后再逐段识别。`
-    );
-  }
-
-  // 1. 上传音频文件
-  const filePath = await uploadAudio(ASR_UPLOAD_URL, audioPath);
-
-  // 2. 构建静态资源 URL
-  const audioUrl = getStaticUrl(ASR_UPLOAD_URL, filePath);
-
-  // 3. 提交 ASR 任务
-  const taskId = await submitAsrTask([audioUrl], ASR_API_KEY);
-
-  // 4. 等待任务完成
-  const results = await waitForTaskComplete(taskId, ASR_API_KEY);
-
-  // 5. 获取转写结果
-  const transcriptions = await fetchTranscriptionResults(results);
-
-  if (!transcriptions.length) {
-    throw new Error('未获取到识别结果');
-  }
-
-  // 6. 转换为 SRT 格式并保存
-  const srtContent = asrToSrt(transcriptions);
-  const filename = `asr_${Date.now()}.srt`;
-  const outputPath = path.join(workingDir, filename);
-  fs.writeFileSync(outputPath, srtContent, 'utf-8');
-
-  return filename;
-};
-
 const server = new Server(
   {
     name: 'media',
-    version: '1.0.9',
+    version: '1.0.10',
   },
   {
     capabilities: {
@@ -367,6 +143,36 @@ server.setRequestHandler(ListToolsRequestSchema, async () => ({
         required: ['working_directory', 'audio_file'],
       },
     },
+    {
+      name: 'tts',
+      description: '语音合成工具，将文本转换为音频文件。',
+      inputSchema: {
+        type: 'object',
+        properties: {
+          working_directory: {
+            type: 'string',
+            description: '工作目录的绝对路径，合成的音频文件将保存到此目录',
+          },
+          text: {
+            type: 'string',
+            description: '需要合成语音的文本内容',
+          },
+          voice: {
+            type: 'string',
+            description: '音色模型，可选值如 sambert-zhimiao-emo-v1（默认）、sambert-zhichu-v1 等',
+          },
+          format: {
+            type: 'string',
+            description: '输出音频格式，可选 mp3（默认）、wav、pcm',
+          },
+          sample_rate: {
+            type: 'number',
+            description: '采样率，默认 16000',
+          },
+        },
+        required: ['working_directory', 'text'],
+      },
+    },
   ],
 }));
 
@@ -379,10 +185,29 @@ server.setRequestHandler(CallToolRequestSchema, async (request) => {
       if (!working_directory || !audio_file) {
         throw new Error('必须同时提供 working_directory 和 audio_file 参数');
       }
-      const filename = await asr({ working_directory, audio_file });
+      const workingDir = resolveWorkingDirectory(working_directory);
+      const audioPath = resolveAudioFile(workingDir, audio_file);
+      const filename = await asr({ workingDir, audioPath, apiKey: DASHSCOPE_API_KEY });
       return { content: [{ type: 'text', text: `语音识别完成，字幕文件已保存到工作目录下：${filename}` }] };
     }
 
+    if (name === 'tts') {
+      const { working_directory, text, voice, format, sample_rate } = args;
+      if (!working_directory || !text) {
+        throw new Error('必须同时提供 working_directory 和 text 参数');
+      }
+      const workingDir = resolveWorkingDirectory(working_directory);
+      const filename = await tts({
+        workingDir,
+        text,
+        apiKey: DASHSCOPE_API_KEY,
+        voice: voice || 'sambert-zhimiao-emo-v1',
+        format: format || 'mp3',
+        sampleRate: sample_rate || 16000,
+      });
+      return { content: [{ type: 'text', text: `语音合成完成，音频文件已保存到工作目录下：${filename}` }] };
+    }
+
     return {
       content: [{ type: 'text', text: `未知工具: ${name}` }],
       isError: true,

package/package.json

@@ -1,15 +1,18 @@
 {
   "name": "@ww_nero/media",
-  "version": "1.0.9",
-  "description": "MCP server for media processing, including ASR speech recognition",
+  "version": "1.0.10",
+  "description": "MCP server for media processing, including ASR speech recognition and TTS speech synthesis",
   "main": "index.js",
   "bin": {
     "media": "index.js"
   },
   "files": [
-    "index.js"
+    "index.js",
+    "utils"
   ],
   "dependencies": {
-    "@modelcontextprotocol/sdk": "^1.22.0"
+    "@modelcontextprotocol/sdk": "^1.22.0",
+    "uuid": "^13.0.0",
+    "ws": "^8.18.3"
   }
 }

package/utils/asr.js

@@ -0,0 +1,233 @@
+const fs = require('fs');
+const path = require('path');
+
+const ASR_UPLOAD_URL = 'http://fsheep.com:10808/upload';
+const ASR_SERVICE_URL = 'https://dashscope.aliyuncs.com/api/v1/services/audio/asr/transcription';
+
+const SUPPORTED_AUDIO_TYPES = ['.mp3', '.wav'];
+const MAX_FILE_SIZE_BYTES = 120 * 1024 * 1024; // 120MB
+const LANGUAGE_HINTS = ['zh', 'en', 'ja'];
+
+/**
+ * 将毫秒转换为 SRT 时间格式 HH:MM:SS,mmm
+ */
+const msToSrtTime = (ms) => {
+  const hours = Math.floor(ms / 3600000);
+  const minutes = Math.floor((ms % 3600000) / 60000);
+  const seconds = Math.floor((ms % 60000) / 1000);
+  const milliseconds = ms % 1000;
+  return `${String(hours).padStart(2, '0')}:${String(minutes).padStart(2, '0')}:${String(seconds).padStart(2, '0')},${String(milliseconds).padStart(3, '0')}`;
+};
+
+/**
+ * 将 ASR 识别结果转换为 SRT 字幕内容
+ */
+const asrToSrt = (asrData) => {
+  const srtEntries = [];
+  let subtitleIndex = 1;
+
+  for (const item of asrData) {
+    const transcription = item.transcription || {};
+    const transcripts = transcription.transcripts || [];
+
+    for (const transcript of transcripts) {
+      const sentences = transcript.sentences || [];
+
+      for (const sentence of sentences) {
+        const beginTime = sentence.begin_time || 0;
+        const endTime = sentence.end_time || 0;
+        const text = (sentence.text || '').trim();
+
+        if (text) {
+          const startStr = msToSrtTime(beginTime);
+          const endStr = msToSrtTime(endTime);
+
+          srtEntries.push(`${subtitleIndex}\n${startStr} --> ${endStr}\n${text}\n`);
+          subtitleIndex++;
+        }
+      }
+    }
+  }
+
+  return srtEntries.join('\n');
+};
+
+/**
+ * 上传音频文件到服务器
+ */
+const uploadAudio = async (uploadUrl, audioPath) => {
+  const fileBuffer = fs.readFileSync(audioPath);
+  const fileName = path.basename(audioPath);
+
+  const formData = new FormData();
+  formData.append('file', new Blob([fileBuffer]), fileName);
+
+  const response = await fetch(uploadUrl, {
+    method: 'POST',
+    body: formData,
+  });
+
+  if (!response.ok) {
+    throw new Error(`上传失败: ${response.status} - ${await response.text()}`);
+  }
+
+  const data = await response.json();
+
+  if (!data.success || !data.data || !data.data.path) {
+    throw new Error(`上传响应格式错误: ${JSON.stringify(data)}`);
+  }
+
+  return data.data.path;
+};
+
+/**
+ * 根据上传接口 URL 和文件路径构建静态资源 URL
+ */
+const getStaticUrl = (uploadUrl, filePath) => {
+  const url = new URL(uploadUrl);
+  return `${url.protocol}//${url.host}${filePath}`;
+};
+
+/**
+ * 提交 ASR 转写任务
+ */
+const submitAsrTask = async (fileUrls, apiKey) => {
+  const response = await fetch(ASR_SERVICE_URL, {
+    method: 'POST',
+    headers: {
+      'Authorization': `Bearer ${apiKey}`,
+      'Content-Type': 'application/json',
+      'X-DashScope-Async': 'enable',
+    },
+    body: JSON.stringify({
+      model: 'paraformer-v2',
+      input: { file_urls: fileUrls },
+      parameters: {
+        channel_id: [0],
+        language_hints: LANGUAGE_HINTS,
+      },
+    }),
+  });
+
+  if (!response.ok) {
+    const errorText = await response.text();
+    throw new Error(`ASR 任务提交失败: ${response.status} - ${errorText}`);
+  }
+
+  const data = await response.json();
+
+  if (!data.output || !data.output.task_id) {
+    throw new Error(`ASR 响应格式错误: ${JSON.stringify(data)}`);
+  }
+
+  return data.output.task_id;
+};
+
+/**
+ * 轮询等待 ASR 任务完成
+ */
+const waitForTaskComplete = async (taskId, apiKey, timeoutMs = 5 * 60 * 1000) => {
+  const taskUrl = `https://dashscope.aliyuncs.com/api/v1/tasks/${taskId}`;
+  const startTime = Date.now();
+  const pollInterval = 500;
+
+  while (Date.now() - startTime < timeoutMs) {
+    const response = await fetch(taskUrl, {
+      method: 'GET',
+      headers: {
+        'Authorization': `Bearer ${apiKey}`,
+      },
+    });
+
+    if (!response.ok) {
+      throw new Error(`查询任务状态失败: ${response.status}`);
+    }
+
+    const data = await response.json();
+    const status = data.output?.task_status;
+
+    if (status === 'SUCCEEDED') {
+      return data.output.results || [];
+    } else if (status === 'FAILED') {
+      throw new Error('ASR 识别任务失败');
+    }
+
+    await new Promise(resolve => setTimeout(resolve, pollInterval));
+  }
+
+  throw new Error('ASR 识别超时');
+};
+
+/**
+ * 获取转写结果详情
+ */
+const fetchTranscriptionResults = async (results) => {
+  const allTranscriptions = [];
+
+  for (const result of results) {
+    if (result.subtask_status === 'SUCCEEDED' && result.transcription_url) {
+      const response = await fetch(result.transcription_url);
+      if (response.ok) {
+        const transcriptionData = await response.json();
+        allTranscriptions.push({
+          file_url: result.file_url,
+          transcription: transcriptionData,
+        });
+      }
+    }
+  }
+
+  return allTranscriptions;
+};
+
+/**
+ * ASR 语音识别
+ */
+const asr = async ({ workingDir, audioPath, apiKey }) => {
+  // 验证 API Key
+  if (!apiKey) {
+    throw new Error('请配置 DASHSCOPE_API_KEY 环境变量');
+  }
+
+  // 检查文件大小
+  const fileSize = fs.statSync(audioPath).size;
+  if (fileSize > MAX_FILE_SIZE_BYTES) {
+    throw new Error(
+      `音频文件大小 ${(fileSize / 1024 / 1024).toFixed(2)}MB 超过限制（最大 120MB）。` +
+      `请先对音频进行分段处理后再逐段识别。`
+    );
+  }
+
+  // 1. 上传音频文件
+  const filePath = await uploadAudio(ASR_UPLOAD_URL, audioPath);
+
+  // 2. 构建静态资源 URL
+  const audioUrl = getStaticUrl(ASR_UPLOAD_URL, filePath);
+
+  // 3. 提交 ASR 任务
+  const taskId = await submitAsrTask([audioUrl], apiKey);
+
+  // 4. 等待任务完成
+  const results = await waitForTaskComplete(taskId, apiKey);
+
+  // 5. 获取转写结果
+  const transcriptions = await fetchTranscriptionResults(results);
+
+  if (!transcriptions.length) {
+    throw new Error('未获取到识别结果');
+  }
+
+  // 6. 转换为 SRT 格式并保存
+  const srtContent = asrToSrt(transcriptions);
+  const filename = `asr_${Date.now()}.srt`;
+  const outputPath = path.join(workingDir, filename);
+  fs.writeFileSync(outputPath, srtContent, 'utf-8');
+
+  return filename;
+};
+
+module.exports = {
+  asr,
+  SUPPORTED_AUDIO_TYPES,
+  MAX_FILE_SIZE_BYTES,
+};

package/utils/tts.js

@@ -0,0 +1,128 @@
+const fs = require('fs');
+const path = require('path');
+const WebSocket = require('ws');
+const { v4: uuidv4 } = require('uuid');
+
+const TTS_WS_URL = 'wss://dashscope.aliyuncs.com/api-ws/v1/inference/';
+
+/**
+ * TTS 语音合成
+ */
+const tts = async ({ workingDir, text, apiKey, voice = 'sambert-zhimiao-emo-v1', format = 'mp3', sampleRate = 16000 }) => {
+  // 验证 API Key
+  if (!apiKey) {
+    throw new Error('请配置 DASHSCOPE_API_KEY 环境变量');
+  }
+
+  // 验证文本
+  if (!text || typeof text !== 'string' || !text.trim()) {
+    throw new Error('合成文本不能为空');
+  }
+
+  const trimmedText = text.trim();
+
+  // 生成输出文件名
+  const filename = `tts_${Date.now()}.${format}`;
+  const outputPath = path.join(workingDir, filename);
+
+  return new Promise((resolve, reject) => {
+    // 清空或创建输出文件
+    fs.writeFileSync(outputPath, '');
+    const fileStream = fs.createWriteStream(outputPath, { flags: 'a' });
+
+    const ws = new WebSocket(TTS_WS_URL, {
+      headers: {
+        Authorization: `bearer ${apiKey}`,
+        'X-DashScope-DataInspection': 'enable'
+      }
+    });
+
+    let resolved = false;
+
+    const cleanup = () => {
+      if (!resolved) {
+        resolved = true;
+        fileStream.end();
+        if (ws.readyState === WebSocket.OPEN) {
+          ws.close();
+        }
+      }
+    };
+
+    ws.on('open', () => {
+      const taskId = uuidv4();
+      const runTaskMessage = {
+        header: {
+          action: 'run-task',
+          task_id: taskId,
+          streaming: 'out'
+        },
+        payload: {
+          model: voice,
+          task_group: 'audio',
+          task: 'tts',
+          function: 'SpeechSynthesizer',
+          input: {
+            text: trimmedText
+          },
+          parameters: {
+            text_type: 'PlainText',
+            format: format,
+            sample_rate: sampleRate,
+            volume: 50,
+            rate: 1,
+            pitch: 1,
+            word_timestamp_enabled: false,
+            phoneme_timestamp_enabled: false
+          }
+        }
+      };
+      ws.send(JSON.stringify(runTaskMessage));
+    });
+
+    ws.on('message', (data, isBinary) => {
+      if (isBinary) {
+        fileStream.write(data);
+      } else {
+        try {
+          const message = JSON.parse(data);
+          const event = message.header?.event;
+
+          if (event === 'task-finished') {
+            cleanup();
+            resolve(filename);
+          } else if (event === 'task-failed') {
+            cleanup();
+            reject(new Error(message.header?.error_message || 'TTS 任务失败'));
+          }
+        } catch (e) {
+          // 忽略 JSON 解析错误
+        }
+      }
+    });
+
+    ws.on('error', (error) => {
+      cleanup();
+      reject(new Error(`WebSocket 错误: ${error.message}`));
+    });
+
+    ws.on('close', () => {
+      if (!resolved) {
+        cleanup();
+        reject(new Error('WebSocket 连接意外关闭'));
+      }
+    });
+
+    // 超时处理（2分钟）
+    setTimeout(() => {
+      if (!resolved) {
+        cleanup();
+        reject(new Error('TTS 任务超时'));
+      }
+    }, 2 * 60 * 1000);
+  });
+};
+
+module.exports = {
+  tts,
+};