Cleanvoice MCP server
Clean up your audio inside Claude and ChatGPT.
Pick your app and connect Cleanvoice. Then ask for what you want: remove the background noise, cut the ums, make your voice sound like a studio.
Can ChatGPT and Claude edit audio?
Neither app edits audio by itself. Connect Cleanvoice and both can: you ask in plain words, Cleanvoice does the editing, and the cleaned file plays in the chat with a download button.
They can transcribe your recording through Cleanvoice too. Ask for a transcript and it comes with the cleaned audio, plus a summary or show notes if you want them.
What you can ask for
“Remove the background noise”
The voice without the traffic, hum, fan or room noise behind it.
“Make it sound like a studio”
Studio Sound: less echo from the room and a fuller voice, for recordings made at home or on a phone.
“Cut the ums and long pauses”
The same talk with filler words, stutters, hesitations and dead air cut out.
“Remove breaths and mouth clicks”
Breathing and lip smacks taken out or turned down.
“Even out the volume”
One steady level for the whole file, at a loudness target such as -16 LUFS if you name one.
“A transcript and show notes”
Text you can read in the chat, with a summary and social posts.
“Fix the sound of a video”
The video back with clean sound and the picture as it was.
“The cuts for my editor”
Every cut as labels or markers for Audacity, Adobe Audition or Reaper, or as an EDL.
- Files
- Audio or video, up to 5 GB each and up to four per edit: MP3, WAV, FLAC, M4A, MP4, MOV, MKV, WebM and other common formats.
- Cost
- 1 credit per minute of audio. New accounts get 30 free minutes; plans start at $11 a month for 10 hours.
- Approval
- Cleanvoice tells you the cost first and waits for your yes before anything runs.
- Free preview
- Hear 20 seconds of the result before you pay, up to five previews a day.
- Privacy
- Files are processed in the EU, deleted after 7 days and never used to train models.
- Apps
- Claude, ChatGPT, Claude Code, Cursor, Codex, VS Code and other apps that support MCP servers.
Start from a ready prompt.
Each one opens a new chat with the request filled in, or copies a prompt that loads the Cleanvoice skill into your coding agent.
- Interviews
Remove background noise
Takes the traffic out of a street interview and keeps the voice.
“Using Cleanvoice, remove the background noise from this street interview: https://cleanvoice.ai/mcp/samples/street-interview.mp3”
- Podcasts
Cut the ums and pauses
Cuts the ums, uhs and dead air from a podcast clip.
“Using Cleanvoice, cut the ums, uhs and long pauses from this podcast clip: https://cleanvoice.ai/mcp/samples/podcast-with-ums.mp3”
- Voice & video
Make it sound like a studio
A recording from a spare room comes back with Studio Sound: less echo, a fuller voice.
“Using Cleanvoice, make this home recording sound like it was made in a studio: https://cleanvoice.ai/mcp/samples/home-recording.mp3”
- Developers
Clean a whole folder
Your coding agent cleans every file in a folder and saves each one next to its original. One cost check covers the batch.
“Read https://cleanvoice.ai/mcp/skills/cleanvoice-batch/SKILL.md and follow it using the Cleanvoice connector to: clean every .wav file in ./raw: remove the background noise and the ums, and save each cleaned file next to its original”
- Podcasts
Transcript and show notes
You get the cleaned episode, a transcript and show notes in one reply.
“Using Cleanvoice, clean up this episode and give me the transcript, a summary and show notes: [link to your episode]”
- Voice & video
Clean the sound of a video
Send a video link. It comes back with the noise and echo gone and the picture untouched.
“Using Cleanvoice, remove the background noise and the echo from this video and give me the video back: [link to your video]”
- Podcasts
Check the cuts in Audacity
Every cut from your last edit as labels to import into Audacity and check.
“Using Cleanvoice, export the cuts from my last edit as Audacity labels.”
- Developers
Clean a file on your computer
Claude Code or Cursor uploads ./interview.wav, cleans it and saves the result beside it.
“Read https://cleanvoice.ai/mcp/skills/cleanvoice/SKILL.md and follow it using the Cleanvoice connector to: remove the background noise from ./interview.wav and save the cleaned file next to the original”
Watch the whole flow.
Connecting an app, approving the cost, listening to the result and adjusting it.