Google announced a suite of AI-powered accessibility improvements for Android and Chrome on Thursday, May 15, 2025, significantly enhancing the digital experience for users with visual impairments.
The most notable update comes to TalkBack, Android's screen reader, which now leverages Gemini AI to provide interactive assistance. While TalkBack previously offered AI-generated image descriptions when alt text wasn't available, users can now ask specific questions about images and receive detailed responses. For example, if someone receives a photo of a guitar, they can inquire about its brand, color, or other details. This functionality extends beyond individual images to entire screens, allowing users to ask questions about product materials or available discounts while shopping.
Google has also enhanced its Expressive Captions feature, which now captures not just what people say but how they say it. The new duration feature recognizes elongated words like "amaaazing" or "nooooo" to better convey emotion and emphasis. Additional sound labels for actions such as whistling or throat clearing further enrich the captioning experience. This update is rolling out in English across the U.S., U.K., Canada, and Australia for devices running Android 15 and above.
For Chrome users, Google has introduced Optical Character Recognition (OCR) technology that automatically processes scanned PDFs. This advancement allows screen readers to interact with previously inaccessible documents, enabling users to highlight, copy, and search for text as they would on any standard webpage. Additionally, Chrome for Android now features Page Zoom, which increases text size without disrupting webpage layouts.
These developments showcase how AI can be harnessed to create more inclusive technology. By integrating Gemini's capabilities directly into accessibility tools, Google is addressing significant barriers faced by visually impaired users, who encounter an average of 90 unlabeled images daily according to Google's research. The company continues to expand its accessibility initiatives, including opening its Project Euphonia resources to help developers build speech recognition tools for diverse speech patterns.