Audio MCP

🎧 音频MCP服务器

A. 模型上下文协议（MCP）服务器 用于专业音频处理、分析和语音音乐同步——由Python（librosa、numpy、Whisper）和TypeScript MCP包装器提供支持。

为AI编码代理构建，以编程方式创建广播就绪的音频内容。

✨ 特性

🎵 核心音频工具（1-8）

🎼 语音音乐同步工具（9-15）

🏗 建筑

audio-mcp/
├── src/
│   └── index.ts          # MCP server — tool definitions + Python runner
├── python/
│   ├── audio_analyze.py          # Audio metadata analysis
│   ├── audio_normalize.py        # LUFS loudness normalization
│   ├── audio_eq.py               # Parametric EQ engine
│   ├── audio_format_convert.py   # Format conversion (via pydub/ffmpeg)
│   ├── audio_beat_grid.py        # Beat tracking (HPSS + librosa)
│   ├── audio_speech_timestamps.py# Word-level timestamps (Whisper)
│   ├── audio_prosody_analyze.py  # Linguistic prosody (n-PVI)
│   ├── audio_dtw_align.py        # Dynamic Time Warping alignment
│   ├── audio_formant_stretch.py  # Formant-preserving stretch
│   ├── audio_beat_drop_detect.py # Energy drop detection
│   ├── artisan_sync_mixer.py     # Prosody-aware music mixer
│   ├── generate_music.py         # Lo-fi music generator
│   ├── generate_ideal.py         # Ambient pad music generator
│   ├── generate_captions.py      # Whisper captioning
│   ├── rhythmic_audio_mixer.py   # Voice+music ducking mixer
│   ├── time_stretch_video.py     # Video time-stretch
│   └── requirements.txt          # Python dependencies
├── package.json
└── tsconfig.json

🚀 设置

先决条件

• Node.js ≥ 18
• python ≥ 3.10
• FFmpeg （用于格式转换）

安装

# Clone the repo
git clone https://github.com/YOUR_USERNAME/audio-mcp.git
cd audio-mcp

# Install Node dependencies
npm install

# Set up Python venv
cd python
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
cd ..

# Build TypeScript
npm run build

配置MCP客户端

添加到您的MCP配置中（例如，双子座/反重力/克劳德）：

{
  "mcpServers": {
    "audio-mcp": {
      "command": "node",
      "args": ["/path/to/audio-mcp/dist/index.js"]
    }
  }
}

📖 用法

连接后，您的AI代理可以使用15个工具中的任何一个。工作流程示例：

1. audio_analyze        → Get BPM, duration of speech clip
2. audio_speech_timestamps → Extract word-level timing
3. audio_prosody_analyze   → Analyze speech rhythm (n-PVI)
4. audio_generate_music    → Create matching background music
5. audio_beat_grid         → Extract beat grid from music
6. audio_dtw_align         → Align speech to music beats
7. audio_formant_stretch   → Stretch speech to fit alignment
8. audio_mix_with_ducking  → Final mix with ducking
9. audio_normalize         → Normalize for target platform
10. audio_format_convert   → Export as MP3

🎯 平台响度标准

📄 许可证

麻省理工学院