Automatic identification of on-screen text

Requirements: Enterprise subscription, private beta (this feature deducts minutes equal to the video length from your allowance, any subsequent subtitle exports will also be deducted from your allowance)

CaptionHub offers the ability to automatically identify any on-screen text, and create captions at the same position, and at the same starting point as that identified text. This makes subsequent localisation of on-screen text much faster and easier.

Notion image

On-screen text identification is available when editing original captions. To use:

  1. Navigate to the drop-down cog on the Edit page, and choose the “Identify on-screen text” option
  1. You’ll receive confirmation that the process has been initiated
  1. A short while later any identified text will appear as captions in Track 2


  • The time taken to identify on-screen text is approximately equal to the running time of the video being analysed.
  • At the moment, a manual refresh of the page is required in order for the captions to appear, but you will receive an emailed notification when the on-screen text identification is complete.
  • We’re currently have no data as to when text disappears from the screen, so whilst the starting point should be accurate, you’ll need to manually adjust the end point.
  • Only one pass of the automatic process is currently possible.
  • Please note that this action is not un-doable.
Did this answer your question?