Skip to main content
25 May 2026

Promotional graphic for Centre for Accessibility Australia’s “Ask the Expert” series. The image features the organisation’s logo, the heading “Ask the Expert,” and a circular photo of a person smiling while using a smartphone. Text at the bottom reads, “Do you have any questions for our team?” In our returning series “Ask an Expert”, members of the CFA Australia team answer common questions about digital accessibility.

If you have any questions about all things digital accessibility, drop an email through to admin@accessibility.org.au or connect with us on social media.

Question from Jasper, one of our supporters:

Are automated captions good enough for accessibility?

Chris, our Senior Digital Accessibility Analyst, says the quality of automated captions can vary greatly depending on the original audio quality as well as the technology being used to generate them.

Tools such as Whisper and Vosk are capable of producing highly accurate captions in many situations. However, automated systems can still struggle with context, technical terminology, accents, multiple speakers, background noise, or unclear audio. While AI-generated captions continue to improve, they may not always provide the same level of accuracy or understanding as a human transcriber.

For this reason, automated captions such as those in PowerPoint, Zoom or Teams should ideally be reviewed and corrected by a person wherever possible. Human review can help identify errors and ensure captions accurately reflect the intended meaning of the content.

In some situations, such as live-streamed events or user-generated content, manual review may not always be practical. In these situations, automated captions can still offer an important accessibility benefit, particularly when the alternative is having no captions at all.

Automated captions are a useful accessibility tool, but they work best when combined with human review and good quality source audio.