Carnegie Mellon University

a line of HTML code showing alt text that reads "image descriptions have value"

December 15, 2022

The Surprising Value of Transcription and Text Equivalency

One of the core tenants of digital accessibility is that all meaningful visual information (photos, artwork, videos, etc.) should have equivalent text. Most often, this comes in the form of alt tags—a hidden field that contains information about the image being displayed on the page—or in the form of a written transcript that describes both audio and visual information from a multimedia presentation or video. Alt tags and transcripts are readable by screen readers, allowing people without sight to hear the information read aloud. 

According to the CDC, more than 4 million people in the US alone have an uncorrectable vision impairment, so these tools are vital for many people, but the benefits of text equivalency extend far beyond the disabled community in some surprising ways. Drivers often benefit from the power of audiobooks or hands-free, voice-activated calling. Children who can’t yet read can look at the pictures while an electronic narrator reads the story in a picture book. Students who are learning a second language get a more comprehensive understanding when they see a picture of an object, can read its name, and hear the word spoken aloud. 

But the biggest benefit of alternate text may be searchability. Videos, until relatively recently, were largely unsearchable. Locating specific video content required extensive time stamping or a lot of rewinding. With text equivalency, you gain the ability to search for a keyword or phrase correlated to a precise moment in the video. 

Google Image Search was introduced more than twenty years ago in 2001 after search queries for a dress worn by Jennifer Lopez overtook the Internet. Developers soon realized that not all web searches could be answered with text content and so the value of image captions and alt tags grew. Now, through the power of artificial intelligence, computers can even begin to auto generate image descriptions. People, trees, and other common objects are recognizable to AI after averaging millions of images and their associated captions. The alternative text that we write manually now helps to inform tomorrow’s technology. 

Learn more about how to write good alternative text for searchability and accessibility with the following resources: