WhatsApp Image Description
Overview
WhatsApp Image Description is an NVDA add-on that allows you to get AI-generated descriptions of images in WhatsApp messages. This add-on works with both the desktop version of WhatsApp and the Microsoft Store version.
Features
- Get detailed descriptions of images in WhatsApp messages
- Support for multiple AI vision services:
- OpenAI (GPT-4 Vision)
- Google Gemini
- Anthropic Claude
- Customizable response length and description language
- Compatible with both desktop WhatsApp and Microsoft Store version
Requirements
- NVDA 2024.2 or later
- WhatsApp Desktop (standard version or Microsoft Store version)
- An API key for at least one of the supported AI services:
- Gemini offers enough generous rate limiting for free.
Installation
- Download the add-on package from the NVDA Add-on Store or the releases page of this repository.
- Open the add-on package with NVDA, which will launch the installation process.
- Follow the on-screen instructions to complete the installation.
- Restart NVDA when prompted.
Configuration
- Go to NVDA menu > Preferences > Settings > WhatsApp Image Description.
- Select your preferred AI service.
- Enter your API key for the selected service.
- Choose your preferred AI model.
- Adjust the maximum response length (in tokens) if needed.
- Select your preferred description language.
- Click OK to save your settings.
Usage
- Open WhatsApp and navigate to a chat.
- Navigate to a message containing an image.
- Press ALT+I to get a description of the image.
- The description will be displayed in a readable window where you can review it at your own pace.
- Press ESC to close the description window when finished.
Troubleshooting
- "This command only works in WhatsApp": Make sure you are in WhatsApp and focused on a message.
- "No image found in this message": Make sure you are focused on a message that contains an image.
- "API Key Required": You need to add your API key in the settings panel.
License
This add-on is licensed under the GNU General Public License v2.
Credits and Acknowledgments
- Thanks to the NVDA community for their support and feedback.