Audio transcription MCP server for Claude, Cursor, Codex and more

Connect the OneStepTranscribe MCP server to your AI assistant and turn audio or video into text without leaving the chat. It is a remote server, so there is nothing to install, no API key, and no account. Just add one URL and ask your assistant to transcribe a file.

Quick start

Pick your tool and add the remote server. The configuration is the same everywhere: point your MCP client at https://onesteptranscribe.com/api/mcp.

claude mcp add --transport http onesteptranscribe https://onesteptranscribe.com/api/mcp

{
  "mcpServers": {
    "onesteptranscribe": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://onesteptranscribe.com/api/mcp"]
    }
  }
}

{
  "mcpServers": {
    "onesteptranscribe": {
      "url": "https://onesteptranscribe.com/api/mcp"
    }
  }
}

{
  "servers": {
    "onesteptranscribe": {
      "type": "http",
      "url": "https://onesteptranscribe.com/api/mcp"
    }
  }
}

[mcp_servers.onesteptranscribe]
command = "npx"
args = ["-y", "mcp-remote", "https://onesteptranscribe.com/api/mcp"]

{
  "mcpServers": {
    "onesteptranscribe": {
      "serverUrl": "https://onesteptranscribe.com/api/mcp"
    }
  }
}

What you can do

Once connected, your assistant gains these tools. In practice you just ask it to transcribe a file and it chains them for you.

check_credits

Check how many transcription credits an email address has. New accounts get a free credit automatically.

prepare_upload

Verify credits and create a one-time, presigned upload link for an audio or video file.

start_transcription

Confirm the file was uploaded, then start the job. Speaker labels and timestamps are included.

buy_credits

Get a secure browser checkout link to top up credits when the balance runs out.

Example: transcribe a file from your assistant

After connecting the server, a typical request looks like this:

How the upload works

The MCP server never receives your file. When you ask to transcribe, it returns a one-time, presigned upload link and your assistant uploads the file directly from your computer to secure storage. The transcription runs in the background, the result is emailed to you, and the uploaded file is deleted from storage once the transcript has been sent.

Frequently asked questions

What is the OneStepTranscribe MCP server?

It is a remote Model Context Protocol (MCP) server that lets an AI assistant such as Claude, Cursor, or Codex transcribe audio and video files for you. You connect it once with a URL, then ask your assistant to transcribe a file in plain language.

How do I transcribe audio in Claude?

Add the OneStepTranscribe MCP server to Claude (Claude Code or Claude Desktop) with the server URL, then ask Claude to transcribe a local audio file and give it your email address. Claude uploads the file straight from your machine, and the transcript is emailed back to you as PDF, DOCX, Markdown, and CSV.

Is there an MCP server for speech to text?

Yes. OneStepTranscribe is a remote MCP server for speech to text and audio or video transcription. It works with any MCP-capable assistant, including Claude, Cursor, Codex, VS Code, and Windsurf, needs no API key, and returns speaker labels and timestamps.

Do I need an API key or an account to use it?

No. There is no API key and no account. You identify yourself with an email address so the finished transcript can be sent to you, exactly like the website. The only thing you configure in your client is the server URL.

Which AI assistants and tools are supported?

Any MCP client that can connect to a remote server works, including Claude Code, Claude Desktop, Cursor, VS Code (Copilot agent mode), the Codex CLI, and Windsurf. It is designed for assistants that can run shell commands, because the file upload runs on your machine.

How does my audio file get uploaded?

Your assistant uploads the file directly from your computer to secure storage using a one-time link, so the file never passes through a middleman server. The transcription runs in the background and the result is emailed to you.

Is it free? How does pricing work?

New accounts receive a free credit, so your first transcription is free with no card required. After that each transcription uses one credit, and you can top up from a secure browser checkout when you run out.

What formats do I get the transcript in?

Each transcript is delivered as PDF, Word (DOCX), Markdown, and CSV in a single email, with speaker labels and timestamps included.