I don't use the OCR feature very often, but it fails to work about 99% of the time for me. The image below is an example. I don't think the background is so bad that the words cannot be captured correctly.
Is that the case for you as well, or am I doing something wrong?
It doesn't tend to do well with subtitles that have lines around them - does the video have embedded subtitle text rather than images? If so then the regular Screen Reader might actually work in some cases.
Would the OCR in the new updated version be able to solve this issue? I'm surprised that Pleco OCR performs poorly in this regard, especially when OCR by Google Translate or other services can handle it quite effectively.
Probably, but we haven't done much work on the Android side of that yet and we're not 100% sure which OCR engine we'll be using on it. On iOS we're adding the option to use Apple's Vision OCR and that does indeed work better with subtitles:
(I should add here that there isn't that much work to be done to port it to Android - the vast majority of the app is cross-platform - but in this particular instance we're relying on an iOS-only API on iOS and so need to decide on which of the Android equivalents of it to use on Android)