Audio MCP Server for Claude, Cursor, ChatGPT & any AI agent
Remove vocals, split stems, clean up recordings and more — right inside Claude, Cursor or ChatGPT. Just say what you want and get the finished file back.
Set it up in under a minute — then let your assistant do the work.
Just ask, in plain language
Once connected, type what you want — your assistant calls the right tool and hands back a download link.
Other things you can ask
"Take these 12 practice recordings, convert each to MP3, then remove the vocals" runs end-to-end in a single agent session — convert, separate, poll, return 12 links. No app-switching, no glue code.
What you can do
Everyday audio editing, free
- Convert between MP3, WAV, FLAC, OGG, M4A, AAC and OPUS.
- Trim a clip to an exact start time and length.
- Extract the audio track from any video file.
- Probe a file for its duration, bitrate and format.
Studio-grade AI separation
- Remove vocals or grab the instrumental for karaoke.
- Split a song into stems: vocals, drums, bass, guitar, piano.
- Clean up voice recordings — remove noise and room echo.
- Track progress and grab results without leaving the chat.
Our own state-of-the-art models
Many audio MCP servers ship older, open-source separation models. We run our own models — the same ones powering vocalremover.com — with a restoration pass on top.
Our flagship separation for cleaner vocals and instrumentals on complex mixes.
Splits a track into vocals, drums, bass, guitar and piano stems.
Restores clarity and detail to processed or low-quality audio.
Connect in 60 seconds
The easiest way is a one-click connector in ChatGPT or Claude.ai — sign in and approve, no API key to paste.
Connect to ChatGPT or Claude.ai
Just sign in — nothing to paste-
1
Open connector settings
In ChatGPT: Settings → Apps → enable Developer mode → Create app, then paste the server URL into the Connection field. In Claude.ai: Settings → Connectors → Add custom connector.
-
2
Paste the server URL
Drop this address into the connector URL field:
https://vocalremover.com/mcp/audio -
3
Sign in and approve
You are sent to VocalRemover to sign in (or start your free trial) and approve access — nothing to paste.
-
✓
Just ask
Type "remove the vocals from this track" and your assistant calls the tool and returns a download link. That's it.
Advanced: connect with an API token
For Claude Desktop, Cursor, Windsurf, Cline, Zed and Claude Code — add the server with a Bearer API token.
Start your free trial to create your account, then copy your API token from the API page and paste it into the config below.
Claude Desktop
Add to claude_desktop_config.json, then restart Claude.
{
"mcpServers": {
"vocalremover": {
"command": "npx",
"args": [
"-y", "mcp-remote", "https://vocalremover.com/mcp/audio",
"--header", "Authorization:${AUTH}"
],
"env": { "AUTH": "Bearer YOUR_API_TOKEN" }
}
}
}Cursor · Windsurf · Cline · Zed
Native remote MCP — add to the client's mcp.json.
{
"mcpServers": {
"vocalremover": {
"url": "https://vocalremover.com/mcp/audio",
"headers": {
"Authorization": "Bearer YOUR_API_TOKEN"
}
}
}
}Claude Code (CLI)
One command — native remote MCP, no bridge.
claude mcp add --transport http \
vocalremover https://vocalremover.com/mcp/audio \
-H "Authorization: Bearer YOUR_API_TOKEN"Your audio is fetched from the URL you pass, processed, and deleted — nothing is stored. Your access is scoped to your account and can be revoked any time.
Need details? Read the full documentation
Free to start
Everyday audio tools
Convert, trim, extract and clean up audio as much as you like — completely free.
AI separation
Studio-grade vocal removal and stem splitting, right inside your assistant. Start with a free trial.
Set it up in 60 secondsBuilt for
Turn any song into an instrumental on command.
De-reverb and denoise voice recordings in one step.
Pull clean stems for remixing and sampling.
Batch-process whole folders inside a single agent run.
Frequently asked questions
What is an MCP server for audio?
An MCP (Model Context Protocol) server lets AI assistants like Claude or Cursor use audio tools as native capabilities. You simply type "remove the vocals from this track" and the assistant does it — no app switching, no manual uploads.
Is it free to use?
You can start for free. Lots of audio tools — format conversion, trimming, extracting audio from video and more — are free to use, and you can try the AI tools (vocal removal, stem separation, denoise, de-reverb) with a free trial.
Which AI clients does it work with?
Claude Desktop, Claude Code, Cursor, Cline, Windsurf, Zed and the Anthropic Messages API connect with your Bearer token directly. The same processing is also available through our REST API.
Can I use it with ChatGPT?
Yes. In ChatGPT, open Settings → Apps, enable Developer mode, choose Create app and paste the server URL into the Connection field; Claude.ai adds it under Settings → Connectors. You sign in to VocalRemover and approve access via OAuth — no token to paste.
Is this the official MCP standard?
Yes. It is built on the open Model Context Protocol, so it works in any compatible client — a standard connector, not a proprietary plugin.
What happens to my files?
Your audio is processed and then deleted immediately — we do not store it, whether you pass a link or upload it directly. Your access is scoped to your account and can be revoked any time.
How good is the separation quality?
Our AI separation runs on our own state-of-the-art models — HQ Fusion, Vocals HQ and Multi-stem HQ — with a Vocal Restoration pass for extra clarity. Many other audio MCP servers ship older open-source models.
What audio and video formats are supported?
MP3, WAV, FLAC, M4A, OGG, AAC and OPUS audio, plus common video formats (MP4, MKV, MOV, AVI) for audio extraction. Pass files as a public link or upload them directly (base64), up to 100 MB each.
Can I use it in automated agent pipelines?
Yes — that is a first-class use case. MCP is built for agent orchestration, so you can chain audio tasks across many files in a single agent run, for example "separate vocals from every track in this folder".
Can I make karaoke tracks with this MCP server?
Yes. Ask your assistant to "remove the vocals" or "give me the instrumental" and it returns a karaoke-ready backing track. You can do this for a single song or batch a whole folder in one agent run.
Can it clean up podcasts and voice recordings?
Yes. The denoise and de-reverb tools remove room sound and background noise from voice recordings, which is ideal for podcasters, voiceover artists and transcription pipelines.
Do I need an API key?
Not for ChatGPT or the Claude.ai web app — you add the server as a custom connector and sign in with OAuth, so there is no key to paste. For clients like Claude Desktop, Cursor or Cline you generate a token on the API page and add it as a Bearer token. New accounts get credits to test the AI tools.
Give your AI assistant ears
Connect the server to your AI assistant and let Claude, Cursor or ChatGPT edit audio for you.
Connect in 60 seconds©2026 VocalRemover. All rights reserved.
Company code: 307547815
VAT No.: LT100019494111
Address: Maironio g. 14-11, Kaunas, Lithuania
Email: info@vocalremover.com
VocalRemover Uses Cookies!