How to recognize different types of text?

Question

iPad

Good time.
There was a problem of recognition of different types of text - from handwritten to scanned from books.
The tasks are as follows:
For handwriting (its recording), the iPad is used, where several applications have already been tried, but each could not satisfy the needs for one reason or another. As a solution, the initial vector text recording with formulas is considered, and then the recognition of this data set on a computer or in the tablet itself, but not in the process, but after it. The problem is further aggravated by the fact that only the first model of the tablet is currently on hand, which is why using monsters like MyScript cannot be done.
As an additional feature, the ability to decrypt arithmetic records is desirable (as it is implemented in some applications from MyScript).
As a last resort, how can this be implemented and transferred as a third-party application to a tablet? (there is a jailbreak)
For printed scanned text, an application is interesting that can digitize an image into text not only while preserving the formatting, but also transferring special characters in the form of their images to the final document (for some reason, FineReader's implemented this with a stretch - for example, curly a bracket that refers to several lines, he can decipher as a chaotic set of characters).
And finally, a question about FIneReader: why doesn't learning by looking at uncertainly recognized characters give proper results? (even after the 100th refinement of the same character/pair of characters, it still cannot recognize them with confidence)
Thank you for your attention, I will be grateful for intelligible answers.