danc at washington.edu
Thu May 8 12:07:05 PDT 2008
Agreed. We had a demo a few weeks back and while we didn't test the
product in-house, it was quite clear to me that the results are just not
good enough. Even if it's 95% accurate, that's not good enough for
academic content and we saw accuracy rates far below that.
-*- Dan Comden danc at washington.edu
Access Technology Lab http://www.washington.edu/computing/atl/
University of Washington UW Technology Services
On Thu, 8 May 2008, John Foliot - Stanford Online Accessibility Program wrote:
> Wink Harner wrote:
>> What experience/hands-on knowledge might any of you have with Docsoft
>> software used for captioning?
> After some local testing here, I sadly had to abandon any thought of using
> DocSoft to create captions: the accuracy rate simply did not meet my
> expectations. As a tool for key word scraping and search/archive
> possibility it seems to be great, but as a captioning tool/solution you will
> be hugely disappointed.
> I know that DocSoft have worked hard on beefing up their editing tool, and
> they claimed that other institutions have seen success rates of 80%+, but my
> experience was far below that within the limited testing I was able to do.
> The holy grail of speech to text is still outside of our current reach, and
> I suspect we will only see reliable results *after* we see jetcars zooming
> around in the air (ala George Jetson), it is simply too hard to do at this
> Athen mailing list
> Athen at athenpro.org
More information about the athen-list