Whisper Desktop

Whisper Desktop is an Electron-based application that allows users to transcribe speech to text using OpenAI's Whisper model through the Groq API. It provides a simple interface for recording audio and automatically transcribing it into text, which can then be inserted into any active text input field.

Tl;dr With the magic that is Whisper and the speed of the Groq servers, I thought I'd spend a weekend to make a tool to help me speak globally into my computer.

Features

• Global hotkey (Ctrl+Shift+Space) to start/stop recording
• Real-time audio recording using the system microphone
• Transcription of recorded audio using the Whisper large-v3 model
• Automatic insertion of transcribed text into the active text input field
• System tray integration for easy access

Installation

1. Clone the repository:

   git clone https://github.com/yourusername/whisper-desktop.git
   cd whisper-desktop
   

2. Install dependencies:

   npm install
   

3. Create a .env file in the root directory and add your Groq API key:

   GROQ_API_KEY=your_api_key_here
   

   To obtain your Groq API key, visit https://console.groq.com/keys.

Usage

1. Start the application:

   npm start
   

2. Press Ctrl+Shift+Space (or Cmd+Shift+Space on macOS) to start recording

3. Speak into your microphone

4. Press Ctrl+Shift+Space again to stop recording and initiate transcription

5. The transcribed text will be automatically inserted into the active text input field

Platform-Specific Notes

macOS

• You may need to grant permission for the app to access your microphone. If prompted, allow microphone access in System Preferences > Security & Privacy > Privacy > Microphone.

Linux

• Ensure you have the necessary audio libraries installed. On Ubuntu or Debian-based systems, you might need to run:

  sudo apt-get install libasound2-dev
  

• If you encounter issues with global shortcuts, you may need to install libxtst-dev:

  sudo apt-get install libxtst-dev
  

Development

The main components of the application are:

• main.js: The main Electron process
• renderer.js: The renderer process handling the UI and recording logic
• preload.js: Exposes Electron APIs to the renderer process
• index.html: The main application window

To modify the application:

1. Make changes to the relevant files
2. Restart the application to see the changes

Building

To build the application for distribution:

npm run build

This will create distributable packages for your platform in the dist folder.

Troubleshooting

If you encounter any issues with audio recording or transcription:

1. Ensure your microphone is properly connected and selected as the default input device
2. Check the console logs for any error messages
3. Verify that your Groq API key is correctly set in the .env file
4. You may need to install Java if you don't already have it installed: https://www.java.com/download/ie_manual.jsp

Roadmap

The following are ideas for future development:

• Allow for using different providers (e.g. OpenAI or self-host)
• Include post-processing capability using LLM providers
• Make the UI more customizable
• Add a feature to save and manage transcription history
• Develop a mobile companion app for remote control and syncing
• Implement advanced audio processing for noise reduction and speaker separation

This was done in a weekend, so I don't have any specific plans to implement any of these yet.

License

This project is licensed under the Apache License 2.0. See the LICENSE file for details.