CleanAudio

CleanAudio MCP vs Descript for prompt-driven audio work

Connect CleanAudio to your agent when you need hosted, account-authenticated audio tools from a prompt, with no self-hosted service or local FFmpeg install

MCP
1

Add CleanAudio to Cursor

Open Cursor Settings → MCP, add a remote server named CleanAudio, paste the connector URL, then click Add and sign in

Cursor MCP docs
2

Connect and start cleaning

After signing in, ask Cursor to clean, cut, or transcribe audio

Start cleaning

Descript is an editing application with its own interface and editorial workflow. CleanAudio MCP is a hosted connector for agents that need to clean noise, cut clips, join takes, remove silence, produce TTS, or transcribe after CleanAudio sign-in.

Use Descript for its product workflow and visual editing features. Use CleanAudio MCP when a supported task should happen directly from an AI agent conversation.

What this MCP server does

  • Clean background noise with clean_noise
  • Cut an uploaded recording with cut_audio
  • Join ordered clips with join_audio
  • Remove pauses with remove_silence
  • Generate speech with text_to_speech
  • Create a transcript with transcribe

Example prompts

Ask Claude, Cursor, ChatGPT, or VS Code in plain language.

  • “Transcribe this interview recording.”
  • “Remove background noise from this narration.”
  • “Remove silence from this podcast segment.”
  • “Cut the first 30 seconds from this voice recording.”
  • “Create a TTS version of this paragraph.”
  • “Join the cleaned intro and interview clips.”

Where CleanAudio MCP differs from Descript

Agent-directed audio tasks

Works inside an agent

Call supported audio tools from Claude, Cursor, ChatGPT, or VS Code instead of moving every task into another editor.

CleanAudio hosted sign-in

Hosted account connection

Authorize CleanAudio in the browser, with no locally hosted audio service to manage.

Podcast transcription workflow

Transcription plus edits

Use transcribe alongside noise cleanup, cuts, joins, silence removal, and TTS.

Use CleanAudio MCP for supported Descript-style tasks

Connect your agent to the hosted URL, sign in, and request a specific cleanup, cut, join, TTS, or transcript.

  1. 1

    Copy the hosted connector URL

    Copy https://mcp.cleanaudio.app/mcp. CleanAudio hosts the MCP server, so there is no local server or FFmpeg installation.

  2. 2

    Connect your AI agent

    Add the URL in Claude Connectors, Cursor Settings → MCP, ChatGPT connectors, or your VS Code MCP client.

  3. 3

    Sign in to CleanAudio

    Complete the browser sign-in with your CleanAudio account. Authorization keeps processing and usage tied to your account.

  4. 4

    Ask for an audio cleanup or transcript

    Upload or reference your audio, then give the agent a concrete instruction. It can choose the relevant CleanAudio tool.

Tools your agent can call

CleanAudio MCP provides clean_noise, cut_audio, join_audio, text_to_speech, remove_silence, and transcribe as focused hosted tools.

clean_noise

Remove background noise from podcast recordings

cut_audio

Trim MP3 files to create ringtones or clips

join_audio

Merge multiple audio files into one

text_to_speech

Convert blog posts to audio narration

remove_silence

Clean up long recordings by removing pauses

transcribe

Convert speech to text automatically

CleanAudio MCP vs Descript FAQ

No. Descript is a full editing product. CleanAudio MCP is a focused, hosted tool connector for agent-driven audio requests.

Yes. The transcribe tool is available after you connect and authorize CleanAudio.

Do not assume filler-word editing from the listed MCP tools. It does provide remove_silence for pauses and transcribe for text.

Yes. Use text_to_speech for supported TTS requests.

No. CleanAudio hosts the MCP connector and processing backend. You connect with your account.

Start prompt-driven audio work from your agent

Add the hosted CleanAudio MCP connector, sign in once, then send a clear audio request from your preferred agent.

Open connector setupMCP overview