E-mail List Archives
Re: Accuracy rates go captioning and transcription
From: Emily Ogle
Date: Oct 21, 2018 9:39AM
- Next message: Isabel Holdsworth: "Re: Highlighting static HTML using the keyboard"
- Previous message: John Foliot: "Re: WCAG 2.1 SC 1.3.5 Identify Input Purpose - TestingMethodology"
- Next message in Thread: None
- Previous message in Thread: Weissenberger, Todd M: "Accuracy rates go captioning and transcription"
- View all messages in this Thread
You would need to achieve 97% accuracy or comprehension starts to suffer. Additionally automated closed captioning tends to caption one word at a time, which leads to the effect of watching the captions rather than the video.
Deaf/hard of hearing *hate* auto captions(myself included). YouTube provides the option to generate closed captioning, but Google won't index autocaptions given their high rate of inaccuracy; you have to fix them or upload before Google will include them in search results.
The best approach is anything that is autogenerated is checked by a human for accuracy. There's a caveat here, too, though: people can sometimes hear what they see and not realize there's an inaccuracy. Once transcripts have been fixed for accuracy, they should go through a timing program where best practices for caption display are followed.
Development on a generation tool would be ongoing. It depends on your long-term plan whether the costs to develop a tool outweighs the existing vendors out there. Two that I know of use a combination of human/technology: Automatic Sync Technologies and 3PlayMedia. Both have prices per minute 2.25-2.75 and have video description services as well.
Emily
> On Oct 21, 2018, at 7:49 AM, Weissenberger, Todd M < <EMAIL REMOVED> > wrote:
>
> Colleagues,
>
> Occasionally I hear about groups (on my campus and elsewhere) who are working to develop automated captioning and transcription tools.
>
> I'd be curious to hear people's thoughts on automated captioning, any success stories (or cautionary tales), cost/benefit analyses between automated and outsourced captioning, and especially what degree of accuracy would be sufficient (and is currently achievable) for such tools.
>
> I look forward to your thoughts on this.
>
> Todd
>
> T.M. Weissenberger
> IT Accessibility Coordinator
> Information Technology Services
> University of Iowa
> 319-384-3323
> > > >
- Next message: Isabel Holdsworth: "Re: Highlighting static HTML using the keyboard"
- Previous message: John Foliot: "Re: WCAG 2.1 SC 1.3.5 Identify Input Purpose - TestingMethodology"
- Next message in Thread: None
- Previous message in Thread: Weissenberger, Todd M: "Accuracy rates go captioning and transcription"
- View all messages in this Thread