In November 2009, Google introduced automatic captions to YouTube. The company quickly touted the feature as beneficial to people with certain hearing disabilities.
“With YouTube expanding its index at a breakneck speed of about 20 hours of new material uploaded each minute, access to this vast body of video material becomes increasingly challenging,” Google’s research team wrote a month after the feature rollout. Of course, automatically generated captions have limitations — and arguably, they’re not a true accessibility feature.
Related: Becoming More Accessible on YouTube
Automatic captions have an accuracy problem
Captions are a crucial accessibility feature, but in order to be useful, they need to be accurate.
The University of Minnesota at Duluth’s Media Hub estimates that YouTube automatic captions typically provide about 60-70% accuracy, which varies significantly depending on the quality of the audio content. Background noise can worsen the accuracy, and while Google’s machine learning captions have made enormous strides since 2009, the service still has trouble understanding homonyms, acronyms, and speakers with certain accents.
Even at the higher end of that range, a 70% accuracy rate simply isn’t good enough. Imagine if you could only read two out of every three words on this page; could you still understand the content?
Automatic caption issues can be entertaining for some users, but frustrating to others
Since YouTube introduced automatic captions, some content creators have built entire channels around the technology’s limitations. Comedy duo Rhett & Link took a creative approach with their “Caption Fail" series, which features actors performing a basic scene several different times — allowing YouTube’s auto-generated captions to change the script.
But while inaccuracies are frequently amusing, it’s important to recognize the real-world impact of accessibility barriers. According to the Centers for Disease Control and Prevention, about 48 million Americans have some form of hearing loss. For many of these people, accurate captions are an essential feature.
For that reason, the Web Content Accessibility Guidelines (WCAG) specifically require captions for all prerecorded audio content in synchronized media. The authors of WCAG note that automatic captions “do not meet user needs or accessibility requirements, unless they are confirmed to be fully accurate.”
The good news is that accurate captions can directly benefit your content in a number of ways:
- An improved user experience. Many people browse the web without sound — regardless of whether they have hearing-related disabilities. Captions provide these users with another way to enjoy your content.
- Improved search engine positioning. When search engines can scan your content for keywords, they can rank it appropriately. Videos with captions and transcripts are more likely to draw organic traffic.
- Improved compliance. Providing text alternatives can help to fulfill your legal obligation to provide reasonable accommodations under the Americans with Disabilities Act (ADA) and other non-discrimination laws.
Use caution when using any automatic caption feature
So, should you turn off YouTube’s automatic captions?
Not necessarily. While automatic captions can create an embarrassing portrait of your brand, they can also save some time if you’re writing captions from scratch — as long as you read the output carefully. Never publish a video without reviewing the captions, regardless of the length of the video or the quality of the output.
In most cases, automatic captions will need significant edits. YouTube makes this fairly easy for content creators. Remember, your captions should include all dialogue and any important sound cues that could change the context of your content. For more guidance, review Google’s support page for editing or removing captions.
Finally, if you’re spending a lot of time editing automatic captions, consider creating captions and transcripts as part of your drafting process. By prioritizing accessibility when planning your video content, you can save time — and ensure that all users have access to useful text alternatives.