Voice-Swap MCP Server
Create speech, swap voices and separate stems just by asking your AI assistant. It runs on your Voice-Swap account, voices and credits.
What it can do
Text to speech
Turn a script into speech with any of your voices or our public roster.
Voice to voice
Swap the voice on a vocal or spoken recording while keeping the performance.
Stem separation
Split a track into vocals, bass, drums and other.
Instant voice clones
Create a new voice from a short, clean recording.
Browse voices
Search your own voices and our public roster to find the right one.
Set it up
Connect in a minute.
You’ll need a Voice-Swap account. Sign in or create one first.
MCP server URL
https://api.voice-swap.ai/mcpOAuth Client ID
OzMCa7LBcYm13lRQIrOB494PAhwykixtA public identifier, not a password. Leave the Client Secret blank.
Claude on the web or desktop
- Open Claude and go to Settings → Connectors → Add → Custom → Web.
- Name it "Voice-Swap" and paste the server URL above.
- Set Authentication to "Always required".
- Set OAuth client to "Use your own OAuth client", paste the OAuth Client ID above and leave the Client Secret blank.
- Leave Transport as Streamable HTTP and click Add.
- Click Connect, sign in to your Voice-Swap account and accept the permissions.
See it work
Just ask.
The Voice-Swap MCP server is a remote Model Context Protocol server. Your assistant picks the right Voice-Swap tool, runs it on your account and returns the audio.
- I want to swap the voice on a vocal I recorded. Give me an upload link.”
- Separate the vocals from this track so I can use them in a remix.”
- List my Voice-Swap voices.”
Example conversation in Claude.
Tool reference
The tools.
The tools your assistant can call. You never need to name them, just describe what you want.
| Tool | What it does |
|---|---|
voiceswap_ | Creates audio: text to speech, voice to voice or stem separation.You provide: text or audio, and a voice |
voiceswap_ | Checks on a generation and returns links to the finished audio.You provide: nothing, it’s automatic |
voiceswap_ | Lists your voices and the public roster.You provide: an optional search |
voiceswap_ | Gives you a link to upload a recording from your computer.You provide: nothing |
voiceswap_ | Creates an instant voice clone.You provide: a short, clean recording |
voiceswap_ | Gives you links to license a roster voice.You provide: a roster voice |
voiceswap_ | Shows your remaining characters and credits.You provide: nothing |
Security
Your account. Your control.
Built so you stay in control of your account.
OAuth sign-in
You sign in on a Voice-Swap page. Your password and API keys are never pasted into your assistant.
Your account, your credits
Every generation runs on your Voice-Swap account and shows in your usage history.
Disconnect any time
Remove the connector in your assistant’s settings to disconnect it.
Licensed voices
Roster voices are licensed from the artists, and every use pays them.
Credits & limits
Free to connect.
The MCP server uses the same credits and TTS characters as the Voice-Swap web app, at the same rates. There is no extra fee to connect.
- Text to speech
- Uses your TTS characters, same as the web app
- Voice to voice
- Per job, same as the web app
- Stem separation
- Per job, same as the web app
- Instant voice clones
- No credits; limited by your plan
Troubleshooting
Quick fixes.
"Authorization failed" right after signing in+
Sign in once at voice-swap.ai with your Voice-Swap account, then remove the connector, add it again and reconnect. If it still fails, contact support.
Sign-in opens, then loops back to the start+
Remove the connector and add it again, making sure the Client ID is pasted exactly and the Client Secret is empty. Pop-up blockers can also stop the sign-in window from returning.
"Server not found" or "Failed to connect"+
Check the URL is exactly https://api.voice-swap.ai/mcp with no trailing slash, and that the transport is Streamable HTTP (not SSE or stdio).
The assistant doesn’t use Voice-Swap+
Make sure the connector is switched on for this chat, then ask by name, for example "Use Voice-Swap to read this aloud".
"Not enough credits" or "not enough characters"+
Top up or change plan in your Voice-Swap dashboard. Credits and TTS characters are shared with the web app.
Tools are missing after an update+
Disconnect and reconnect the server so your assistant reloads the latest tool list.
What is the Voice-Swap MCP server?+
It connects your AI assistant to Voice-Swap using MCP (Model Context Protocol), an open standard for plugging tools into AI assistants. Once connected, you can create speech, swap voices and separate stems just by asking.
Which AI assistants does it work with?+
Claude (web and desktop), Claude Code and Cursor are tested and supported. Other assistants that support remote MCP servers over Streamable HTTP with OAuth sign-in should also work. Use the Other tab above to set them up.
Do I need to know how to code?+
No. In Claude or Cursor you add it from Settings in about a minute, then just ask in plain language. Claude Code takes one command.
Do I need a Voice-Swap account?+
Yes. The server runs on your own Voice-Swap account, so your voices, credits and history are all there. If you don’t have one yet, create one at voice-swap.ai and sign in once before connecting.
Does it cost extra?+
No. Connecting is free. Anything your assistant creates uses the same credits and TTS characters as the Voice-Swap web app, at the same rates.
What can I ask it to do?+
Turn a script into speech, swap the voice on a recording, split a track into vocals, bass, drums and other, create an instant voice clone or search for a voice. Not sure where to start? Ask "What can I do with Voice-Swap?"
How do I give my assistant an audio file?+
Ask your assistant for an upload link. It gives you a secure Voice-Swap page where you upload the recording, then it carries on with the task. Each link expires after 30 minutes.
How do I get the finished audio?+
Your assistant replies with a link to the finished file so you can play or download it.
Can I use my own custom voices?+
Yes. Voices in your Voice-Swap account, including your custom models and instant voice clones, are available alongside the public roster.
Can I use my organisation’s voices?+
Not yet. The connector works with your personal account only.
Can I use the audio commercially?+
Audio made with your own voices and instant voice clones follows your plan, the same as the web app. Roster voices are for personal use until you license them: ask your assistant for a licence link.
What permissions does it ask for?+
Reading your voices and balance, generating audio (text to speech, voice to voice and stem separation), uploading audio and creating instant voice clones. Your assistant can only act on your own account.
Is it secure?+
Yes. You sign in on a Voice-Swap page using OAuth, so your password and API keys never go into your assistant. The Client ID on this page is a public identifier, not a secret.
How do I disconnect Voice-Swap?+
Remove the Voice-Swap connector in your assistant’s settings. In Claude, go to Settings → Connectors. In Cursor, remove it from Cursor Settings → MCP. In Claude Code, run claude mcp remove voice-swap.