Currently OCR is limited to using it to extract subtitles in DVB image format which are embedded as a separate subtitle track. It does this automatically. We are working on an option to use OCR to also extract from burnt in subtitles (which your sample upload appears to be using).
If you have some that working, I would recommend uploading that original file and the corresponding logs so we can compare the differences between the ones that are working and the ones aren’t working.
Regarding the track, you can select the audio language option in the Conversion settings, on your case setting it to spa for Spanish and it will ignore the other tracks and only extract that track (if available). Again the original video and log files will help us see what’s going on in those that are working vs those that aren’t.