It almost sounds like you're asking about Text Video Description (TVD) rather than captions.These are described under the alternate content technologies section of Media Accessibility User Requirements - HTML accessibility task force Wiki . When you look at the requirements listed there, Canvas Studio does not support those.
Captions would be very difficult to read for a lot of reasons: audio (screen reader) over audio (video), the time the caption is on screen may not be enough time to read the caption, the caption is dynamically generated, etc.
If you want to screen read a caption, then it is recommended that you supply a transcript and let the screen reader read that. Some software will highlight the transcript as the video plays, but Canvas Studio is not one of those as Canvas Studio doesn't even support transcripts.