The Student Computer Fee Committee helped Accessible Technology Services purchase a DocSoft system. This system creates a voice profile from videos. Individuals who do significant recording can also upload a vocabulary of commonly used words to increase voice recognition accuracy.
Once a voice profile is created, new videos can be run against the profile for voice recognized captioning. While it is true that YouTube auto-recognized captioning has improved, YouTube does not have the ability to work with a vocabulary of commonly used words.
While results vary with mic and voice quality, settings and topic, Randy Skelton has been producing many videos for his classes. And, comparing 20 minutes of DocSoft voice-recognized captions to what was actually said, DocSoft was 94.6% correct. The words that were missed were either unique anthropology vocabulary or easy to understand errors such as substituting 'the' for 'a' or homonyms such as here and hear.
If you record lectures or create videos for your course and want to learn more about DocSoft's abilities, please call ATS at 243-2082.