Anthropic Splice MCP

Make sound part of the conversation

Splice is now inside Claude—so you can find, shape, and build with real sounds using nothing but words.

Get started

Describe it. Hear it. Use it.

Just describe what you’re looking for and get sounds that match.

Create with context

Claude understands your project and helps you build from there.

Build from a prompt

Generate stacks, kits, and starting points—ready to drop into your workflow.

FAQs

What is the Splice MCP server?

The Splice MCP Server lets you search Splice’s sample catalog, build stacks, and download sounds directly inside AI tools like Claude — without opening the Splice app or switching tabs.

Just describe what you’re looking for in plain language, and the AI handles the rest.

What can I do with it?

Search by feel: Describe a sound in natural language ("dark lo-fi guitar loop around 80 BPM") and get matching results

Download samples: Use your credits to save sounds directly to your computer

Build full stacks: Describe a vibe and generate a multi-layer arrangement (drums, bass, keys, etc.), saved to your library

Build Stacks from a sample: Use a loop as a starting point and generate a full stack around it

Edit stacks: Rename, change BPM, add or remove layers, or swap sounds

Share stacks: Generate a public link anyone can open in a browser

Do I need to be a developer to use this?

Nope. If your AI tool supports MCP, you just add the server URL and log in.

What AI tools does it work with?

Any AI tool that supports HTTP-based MCP servers can connect using the same URL.

Is this free to use?

Searching and building stacks is free. Downloading requires a paid subscription and uses credits.

How do I get access?

The Splice MCP Server is currently in beta and available to all users with a Splice account.

Does this work with ChatGPT?

Not at this time.

Read more here: