Gemini Audio Mcp

io.github.jxoesneonPublished by GitHub account jxoesneonv0.1.0Updated Oct 24, 2026

High-performance audio, music, and voice generation MCP server for Gemini 2.5 and Lyria 3.

VerifiedSTDIODesktop onlyMedia & DesignAI & ML

Overview

AI-generated overview

A local MCP server that lets an assistant generate soundscapes, music and voice narration using Google Gemini 2.5 and Lyria 3 audio models.

What you can do
Generate vocal-free ambient soundscapes from a text prompt, with duration, quality and auto-play options and optional seamless looping. Compose music with tempo, key and intensity controls, optionally guided by a local image. Produce expressive narration and character dialogue with voice direction. Crossfade two soundscapes into a smooth scene transition. Tune output with seed, guidance and duration parameters.
When to use it
Useful when you want an assistant to produce ambient audio, music beds or spoken narration on demand, for example for scenes, prototypes or media projects, and you already have Google AI Studio access.
Requirements
Runs as a local process, typically launched with npx (Node.js 18 or newer) or built from source with the Rust toolchain. FFmpeg must be installed and on the PATH for audio encoding and transcoding. Requires a Google AI Studio API key supplied in the GEMINI_API_KEY environment variable, and network access to Google's Gemini and Lyria services.
Before you install
The GEMINI_API_KEY secret is required and should be injected via the environment, never hardcoded. Prompts, text and any local image path given to the server are sent to Google's Gemini and Lyria services. Lyria 3 models typically require a paid tier or preview access, so usage may incur costs. Generated audio is written locally and may auto-play.

Installation

In SourceWeft

  1. Open Gemini Audio Mcp in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

🎵 Gemini Audio MCP

[Institutional Grade] [Gemini 2.5] [Lyria 3]

Gemini Audio MCP is a high-performance Model Context Protocol (MCP) server engineered for professional-grade audio synthesis. It leverages the Gemini 2.5 Multimodal Live API and Google DeepMind's Lyria 3 models to deliver high-fidelity environmental soundscapes, musical compositions, and expressive narration on-demand.

🛠 Prerequisites

Before deploying the server, ensure your environment meets the following technical requirements:

1. FFmpeg (Core Processing Engine)

Required for high-performance audio encoding, decoding, and transcoding.

  • macOS: brew install ffmpeg
  • Windows: winget install ffmpeg or download from ffmpeg.org.
  • Linux (Ubuntu/Debian): sudo apt update && sudo apt install ffmpeg

2. Rust Toolchain (Compilation)

Required to build the server from source.

  • Install via rustup.rs: curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

3. Node.js & NPM (Runtime)

Required if using the pre-compiled NPM package.

  • Version: node >= 18.0.0

🚀 Installation & Deployment

Global Installation (via NPX)

The fastest way to integrate the server into your MCP client (e.g., Claude Desktop).

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

Manual Build (Optimized)

For maximum performance, build the Rust binary locally:

  1. Clone & Build:
    bash
    git clone https://github.com/mcp-servers/gemini-audio-mcp.gitcd gemini-audio-mcpcargo build --release
  2. Locate Binary: The optimized binary will be in ./target/release/gemini-audio-mcp.

🔑 API Key Management

The server requires a valid Google AI Studio API key.

  1. Obtain your key from Google AI Studio.
  2. Security Best Practice: Never hardcode keys. Inject the key via the GEMINI_API_KEY environment variable.
  3. Tier Note: Access to Lyria 3 (Pro/Clip) models typically requires a Paid Tier or specific preview access in Google AI Studio.

🎮 Tool Usage Guide

1. Environmental Generation (generate_soundscape)

Synthesizes immersive, vocal-free ambient textures.

json
{  "name": "generate_soundscape",  "arguments": {    "prompt": "Deep underwater abyss, low-frequency whale songs, rhythmic air bubbles rising, muffled aquatic pressure.",    "duration": 60,    "quality": "high",    "auto_play": true  }}

2. Professional Music (generate_music)

Generates structural compositions with optional vocal control.

json
{  "name": "generate_music",  "arguments": {    "prompt": "Melancholic solo cello in a vast cathedral with 5-second decay reverb.",    "bpm": 72,    "song_key": "D minor",    "intensity": 4  }}

3. Expressive Voice (generate_voice)

Narration and character dialogue using Gemini 2.5 Native Audio.

json
{  "name": "generate_voice",  "arguments": {    "text": "The artifacts are stable, but the rift remains open.",    "voice_direction": "Gravelly, urgent, whispered"  }}

4. Dynamic Evolution (transition_soundscape)

Crossfades two distinct environments for seamless scene transitions.

json
{  "name": "transition_soundscape",  "arguments": {    "from_prompt": "Quiet library silence.",    "to_prompt": "Sudden heavy rain on a tin roof.",    "transition_duration": 8  }}

⚙️ Advanced Parameters

ParameterTypeDescription
seedIntegerEnsures deterministic, reproducible audio outputs.
image_pathStringMultimodal: Uses a local image to guide the acoustic mood (e.g., resonance).
bpmNumberExplicitly sets the rhythmic tempo (essential for music).
intensityNumber1-10 scale controlling dynamic range and complexity.
guidanceNumber0.0-6.0 scale for prompt adherence (Lyria models).
durationNumberTarget length in seconds. Triggers the Seamless Looping Engine.

🔬 Architecture Overview

Gemini Audio MCP employs a unique Hybrid Engine Strategy:

  • WebSocket Loop: Connects to Gemini 2.5 Live for low-latency, interactive voice and foley tasks.
  • REST Pipeline: Interfaces with Lyria 3 Pro for high-fidelity musical synthesis.
  • PCM Processing: An internal Rust-based loop (decode -> crossfade -> loop -> encode) ensures that short clips are transformed into seamless, infinite soundscapes without audible clicks.

🧪 Troubleshooting

FFmpeg Errors

  • "FFmpeg not found": Ensure ffmpeg is in your system PATH. Run ffmpeg -version in your terminal to verify.
  • Transcoding Failures: Check if you have the necessary codecs (e.g., libmp3lame for MP3). Most standard FFmpeg installations include these.

API Issues

  • 429 Rate Limit: The server implements a semaphore to limit concurrency, but ensure your API tier supports the requested model.
  • Empty Audio Output: Verify your GEMINI_API_KEY is correct and that your account has access to the requested model (especially lyria-3-pro-preview).

📄 License

Licensed under the MIT License. Engineered with precision by the MCP community.

Source: README.md at commit 516a1f7

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v0.1.0LatestSep 16, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts