When automatically generated captions average at 60%-70% accuracy, communication cannot be reasonably assumed to be as effective or equivalent to the audio. Even with higher rates of accuracy, small errors can completely change the meaning or intention of a single sentence. Poor quality, non-time-synced captions such as those produced by automatic technologies can negatively impact students who are relying on captions to access video content.
Examples of recent cases on video accessibility and auto-captions:
- National Association of the Deaf Press Release on Settlement with Harvard
- National Association of the Deaf Press Release on Settlement with MIT
- Office for Civil Rights Letter of Finding: University of California Berkeley