GemVDA - Google Gemini AI for NVDA

Summary

GemVDA integrates Google Gemini AI capabilities directly into NVDA, providing blind and visually impaired users with powerful AI assistance. The add-on supports various Gemini models including Gemini 3, Gemini 2.5 Pro and Flash variants for chat, image description, video analysis and more.

Features

• AI Chat: Have conversations with Gemini AI directly from NVDA
• Screen Description: Capture and describe the entire screen
• Object Description: Describe the current navigator object
• Video Analysis: Record screen video and have Gemini analyze it
• Attach Images: Attach images from files for AI description
• Conversation History: Maintain context across multiple messages
• Multiple Models: Choose from various Gemini models based on your needs
• Customizable Settings: Configure temperature, tokens, streaming and more

Requirements

• NVDA 2023.1 or later
• Google Gemini API key (free tier available)
• Internet connection

Setup

Getting an API Key

1. Visit Google AI Studio
2. Sign in with your Google account
3. Create a new API key
4. Copy the key to use in the add-on

Configuring the API Key

1. Press NVDA+N to open the NVDA menu
2. Go to Preferences > Settings
3. Select the "Gemini AI" category
4. Click "Configure API key..."
5. Paste your API key and press OK

Keyboard Shortcuts

| Shortcut | Action | |----------|--------| | NVDA+G | Open Gemini AI dialog | | NVDA+Shift+E | Describe the entire screen | | NVDA+Shift+O | Describe the navigator object | | NVDA+V | Start/stop video recording for analysis |

Using the Gemini Dialog

When you open the Gemini dialog with NVDA+G:

1. Model: Select which Gemini model to use
2. System Prompt: Optional instructions on how Gemini should respond
3. History: View the conversation history
4. Message: Type your message or question
5. Send: Send your message to Gemini
6. Attach Image: Add an image file for Gemini to analyze
7. Clear: Clear the conversation history
8. Copy Response: Copy the last response to the clipboard

Dialog Tips

• Press Enter in the message field to quickly send
• Use Tab to navigate between controls
• History updates automatically as you chat
• Attached images are sent with your next message

Settings

Access settings via NVDA menu > Preferences > Settings > Gemini AI:

• Default Model: Choose your preferred Gemini model
• Temperature (0-200): Controls response creativity (0=focused, 200=creative)
• Maximum Output Tokens: Maximum length of responses
• Stream Responses: Display responses as they arrive
• Conversation Mode: Include chat history for context
• Remember System Prompt: Save your custom prompt
• Block Escape Key: Prevent accidental dialog closure
• Filter Markdown: Remove markdown formatting from responses

Audio Feedback

• Play sound when sending request: Audio confirmation when message is sent
• Play sound while waiting: Progress sound during AI processing
• Play sound when response received: Notification when response arrives

Available Models

• Gemini 3 Pro (Preview): Most capable model with reasoning capabilities
• Gemini 3 Flash (Preview): Fast model with reasoning capabilities
• Gemini 2.5 Pro: Powerful production-ready model
• Gemini 2.5 Flash: Fast and efficient for most tasks
• Gemini 2.5 Flash-Lite: Lightweight and faster responses
• Gemini 2.5 Flash Image: Optimized for image-related tasks

Image and Video Features

Screen Description (NVDA+Shift+E)

Captures your entire screen and sends it to Gemini for a detailed description. Useful for:

• Understanding unfamiliar interfaces
• Getting an overview of visual content
• Identifying elements that NVDA cannot describe

Object Description (NVDA+Shift+O)

Captures only the current navigator object. Useful for:

• Describing specific interface elements
• Understanding images or icons
• Getting details about focused controls

Video Analysis (NVDA+V)

1. Press NVDA+V to start recording
2. Perform the actions you want to analyze
3. Press NVDA+V again to stop
4. Wait for Gemini to analyze the video

Useful for:

• Understanding visual workflows
• Getting step-by-step descriptions
• Analyzing dynamic content

Troubleshooting

"Google GenAI library not installed"

Run the dependency installer: 1. Navigate to %APPDATA%\nvda\addons\GemVDA 2. Run install_deps.bat or install_deps.py 3. Restart NVDA

"No API key configured"

Configure your API key in Settings > Gemini AI > Configure API key

Responses are too short or cut off

Increase the "Maximum Output Tokens" setting

Responses are too random

Lower the Temperature setting (try 50-100)

Privacy Notice

• Your messages and images are sent to Google's Gemini API
• API keys are stored locally in your NVDA configuration
• No data is shared with the add-on developer
• Review Google's AI privacy policy for more details

Support

• Report issues: GitHub Issues
• Source code: GitHub Repository

License

This add-on is released under the GNU General Public License v2.

Author

Oriol Gomez Sentis