The words and symbols that are seen as valid are determined by the language data loaded (spa.traineddata in this instance). By default, Tesseract.js loads .traineddata files provided by the main Tesseract project--we use integerized versions of the tessdata_best data. If you were to find or create a .traineddata file that does not have this issue, you can use it by setting the langPath argument.

Issues related to language data are outside of the scope of this repo. The goal of Tesseract.js is to bring the Tesseract OCR engine to the browser--we do not make any edits to the recognition engine or .traineddata provided by Tesseract. If you are interested in learning more about how language data works, and what tools exist to modify it, you should look at the documentation provided by the main Tesseract project. Their website is here and repo is here.