Closed-captioning service available through central source

Iowa State has another tool to help make video content more accessible to hard-of-hearing and deaf viewers. Cyndi Wiley, digital accessibility coordinator in information technology services (ITS), collaborated with the other state Board of Regents institutions to secure a closed-captioning service contract this summer. 

The captioning service -- for recorded videos, not live broadcasts -- is provided by Rev, a company with headquarters in Austin, Texas, and San Francisco. Rather than generating automated captions, Rev's closed captioning work is done by freelancing experts. The service is available 24/7 and the company guarantees a minimum 99% accuracy.

With the combined buying power of the regent institutions, Wiley secured a prepaid bank of service minutes at a discounted price. ISU's $1.06 per minute rate is 15% less than Rev's advertised cost. Departments and units need a worktag to submit a request to use the Rev minutes. Directions for using the closed-captioning service are on the ITS knowledge base site

Automated vs. people-powered captioning

Wiley said speech recognition programs that auto-generate text require editing before publishing. While cost-effective (in many cases free), the time it takes to upload, generate and edit the automated captions isn't the best approach for videos more than 20 minutes in length.

"The larger and longer the video, the more time it will take to upload, process, run auto-generated captions and human edit before publishing. File sizes for videos tend to be large and will take more time to upload depending on length and resolution of the video. A typical amount of time needed for editing and adding captions is two to three times the length of a video," Wiley said. "In cases where the videos are longer than 10-20 minutes, it might be helpful to outsource the captions to a service like Rev."

Automated speech recognition (ASR) platforms have difficulty with multiple speakers, low-quality microphones, nonnative English speakers and content with equations and formulas. "Some ASR platforms run on more advanced artificial intelligence and the error rate is less. However, the accuracy never achieves 99%," she said.

Wiley said captioning is an important part of creating video content, pointing to research that shows most faculty, students and staff with and without a disability use it -- leading to an increase in learning when captions are accurate.