Cutted

Cutted is an AI-powered audio editor you control with natural language.

• Cut, trim, or adjust volume by typing what you want
• Detect and remove silence or loud parts automatically
• Transcribe audio (with Whisper)
• Undo edits, export as MP3 or WAV

Install

1. Make sure FFmpeg is installed.
2. Install Cutted:

   pip install cutted
   

   With Whisper support:

   pip install cutted[whisper]
   

Usage

python -m cutted

Load an audio file, enter your command in plain English, and export your result.
You can also enable or disable chat history for smarter follow-up actions in the settings menu (gear icon).

Usage Examples

Basic Editing

• Cut a segment:

  cut from 10 to 20
  

  Removes audio between 10s and 20s.

• Remove multiple segments:

  cut from 5 to 7 and from 15 to 18
  

• Change volume:

  make 0 to 5 seconds twice as loud
  

  or

  reduce volume from 10 to 12 by half
  

• Remove silence:

  remove all silent parts
  

  Note: Removing silence is experimental and can be buggy or produce bad results in some cases.

• Undo last edit:
  Click the "Undo" button.

• Export your result:
  Click the "Export" button and choose MP3 or WAV.

Advanced: Using Whisper Transcription

If you install with Whisper support and enable "Send transcript to Gemini" in the UI, you can use commands like:

• Cut by word or sentence:

  cut when I say the word hello
  

  or

  cut the sentence "welcome to the show"
  

  This uses Whisper to transcribe your audio and allows precise editing based on spoken words or phrases.

Audio-to-Gemini (Experimental)

If you enable "Send audio to Gemini" (checkbox), you can also use word-based commands, but accuracy is much lower compared to using Whisper.

Gemini API Key

Set your Gemini API key as an environment variable before running Cutted:

Linux/macOS:

export GEMINI_API_KEY="your-gemini-api-key"

Windows (Command Prompt):

set GEMINI_API_KEY=your-gemini-api-key

Windows (PowerShell):

$env:GEMINI_API_KEY="your-gemini-api-key"