Well, now yes, increasing the size of my template to 3 Mp made it. I can decode now.
So I found another issue.
String:
Juan Antonio Martínez Rodríguez-Bermejo
c/ Alcañís 40 Logroño 28043
ESPAÑA
 
Is usually very misdecoded and shown in red. But once it [thinks it] gets properly decoded, what was actually interpreted was:
Juan Antonio Martinez Rodriguez-Bermejo
c/ Alcalis 40 Logrofio 28043
ESPANA
 
I may well understand í might be misinterpreted as i (in fact, this not a big deal), but what is unacceptable is ñí being as li, ñ as fi and especially Ñ as N.
This latter is a common unaccuracy on Spanish OCRs and -I can bet my left eye- this is what customers are going to test as soon as they get the app!
This field is defined as OCR, whose settings are: Alphanumeric, NO regexp, NO verification, European languages.
Should I instead specify a custom language with all the possible letters in -in my case- Spanish?
How is this done, i.e, which is the syntax for doing this?
 
Thanks!