🎵 Gemini Audio MCP

Gemini Audio MCP is a high-performance Model Context Protocol (MCP) server that leverages the power of the Gemini 2.0 Multimodal Live API to generate high-fidelity, environmental soundscapes on-demand.

🚀 Mission Statement

Our mission is to provide an immersive, AI-powered audio generation layer for any MCP-compatible environment, enabling the creation of dynamic, seamless, and high-quality environmental audio through simple text prompts.

✨ Key Features

• 🌊 Dynamic Soundscapes: Generate complex environmental audio using the latest Gemini 2.5 Native Audio models.

• 🎵 Professional Music: High-fidelity music production via Google's Lyria 3 models:

  • Lyria 3 Pro: Full song generation with structural coherence ($0.08/req).

  • Lyria 3 Clip: Low-latency clips and rhythmic loops ($0.04/req).

• 🔁 Infinite Looping: Seamless, click-free looping with 100ms micro-crossfades.

• 🔀 Smooth Crossfades: Transition between two different soundscapes with customizable crossfade durations.

• 📂 Universal Formats: Export audio to a variety of formats (WAV, MP3, OGG, FLAC) powered by FFmpeg.

• ▶️ Auto-play Integration: Instantly play generated audio through your system's default player upon completion.

• ⚙️ Persistent Configuration: Fine-tune default bitrates, sample rates, and durations once and reuse them across sessions.

🛠 Installation Guide

Prerequisites

1. FFmpeg: Required for audio conversion and processing.

   • macOS: brew install ffmpeg

   • Ubuntu/Debian: sudo apt install ffmpeg

   • Windows: Download from ffmpeg.org.

2. Rust Toolchain: Required for building the project (cargo).

3. Gemini API Key: Obtain your key from the Google AI Studio.

1. NPM / NPX (Recommended for non-Rust users)

Add the server directly to your MCP client configuration using npx:

{
  "mcpServers": {
    "gemini-audio": {
      "command": "npx",
      "args": ["-y", "gemini-audio-mcp"],
      "env": {
        "GEMINI_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

2. Manual Installation (Rust)

1. Clone the repository:

   git clone https://github.com/mcp-servers/gemini-audio-mcp.git
   cd gemini-audio-mcp

2. Build the project:

   cargo build --release

3. Configure your environment: Set the GEMINI_API_KEY environment variable in your MCP client or system.

🔧 Tool Usage Examples

Generate a Soundscape

Create an immersive 30-second loop of a cyberpunk rainy city.

{
  "name": "generate_soundscape",
  "arguments": {
    "prompt": "Heavy rain on neon-lit cyberpunk city streets, distant hover-car hums, muffled holographic advertisements.",
    "duration": 30,
    "format": "mp3",
    "auto_play": true
  }
}

Transition Between Environments

Seamlessly shift from a peaceful forest to a roaring thunderstorm.

{
  "name": "transition_soundscape",
  "arguments": {
    "from_prompt": "Quiet morning forest with chirping birds and rustling leaves.",
    "to_prompt": "Intense tropical thunderstorm with loud thunder claps and heavy downpour.",
    "transition_duration": 10,
    "auto_play": true
  }
}

Update Server Defaults

Set the default output format to FLAC for higher quality.

{
  "name": "configure",
  "arguments": {
    "default_format": "flac",
    "default_sample_rate": 48000
  }
}

🏛 Architecture Overview

The server is built with a modular Rust architecture designed for efficiency and reliability:

• main.rs: The core MCP protocol engine handling tool registration and request dispatching.

• gemini.rs: Manages low-level WebSocket communication with the Gemini 2.0 Multimodal Live API.

• audio.rs: Handles PCM data manipulation, including seamless looping algorithms and FFmpeg integration for format transcoding.

• mixer.rs: Implements audio processing logic for crossfading and blending multiple audio streams.

• config.rs: Provides a persistent JSON-based configuration layer for user preferences.

📄 License

Distributed under the MIT License. See LICENSE for more information.