Microsoft announced a new accessibility feature for its browser to improve the user experience for visually impaired people. Edge now automatically generates image captions so that screen readers can describe visual content for those who depend on assistive technologies. Image captions or "alt text" is a crucial part of the accessible web that often gets ignored, and Microsoft wants to fix this problem by providing auto-generated captions.
Microsoft uses Azure Cognitive Services to analyze and describe pictures with missing "alt text." When Edge detects an unlabeled image, it sends it to Microsoft's servers for processing. Machine learning algorithms can work with the most common formats, such as JPEG, PNG, GIF, WEBP, and others, to provide descriptive summaries in 5 languages. Besides, Microsoft can analyze text on images in more than 120 languages.
As with other automated systems, auto-generated captions in Microsoft Edge are not perfect, and Microsoft warns that the quality will vary. For that reason, the company urges web developers and content managers to make sure images on their websites have user-generated alt text.
Also, Microsoft will not generate descriptions for decorative pictures, images smaller than 50 x 50 pixels, huge images, and explicit content (pornographic, sexually suggestive, or gory content).
If you want to try auto-generated captions in Microsoft Edge, update your browser to the latest version and navigate to Settings > Accessibility, then toggle on the "Get image descriptions from Microsoft for screen readers" feature. For now, it is available in Microsoft Edge for Windows, Linux, and macOS.