Descripteur de contenu IA
Par Carter Temm and other contributors
This add-on makes it possible to describe the focused control, navigator object, or entire screen using popular AI vision models including OpenAI GPT, Google Gemini, Anthropic Claude, xAI Grok, Mistral Pixtral, local options like Ollama/llama.cpp/LiteLLM, and more. It also supports follow-up questions on any description, camera snapshots, clipboard images, and face positioning via a connected camera. Content descriptions may not always be completely accurate or reflect real-world information. Press NVDA+shift+i to pop up a menu for choosing what to describe, NVDA+shift+u to describe the navigator object, NVDA+shift+y for an image on the clipboard, or NVDA+alt+c to ask follow-up questions about the last description. Other keystrokes are available from the input gestures dialog. By default, usage is free through PollinationsAI with no sign-up or API key required. To use other models, visit the provider's website to obtain an API key, then enter it under the AI Content Describer category in NVDA's settings dialog. See the add-on documentation for more information.
Téléchargements disponibles
Autres détails
- Compatibilité NVDA : de 2023.1 à 2026.1
- Proposé dans l'add-on store : le 06/05/2026 à 21:06
- Code source
- Licence : GPL v2