When I have come across newspaper articles, obituaries and books that contain important information I prefer to take both an image copy of the item as well as capture the actual text.  Having the actual text allows for searching as well as copy-and-paste.   My problem has always been the getting the actual text off of the paper and into my computer.  What is the best way?

Re-Typing is Painful

Even though a simple obituary is usually short typing it in can be really tough because of all the proper names and places that are often in them.  But still, manually transcribing them is something that I have done a lot of.   If I can retype the text in less time than it takes to use other options I will do it.

Newspaper articles and books are usually more time consuming to type because they can be quite long. 

OCR is Imperfect

In the past I thought that the ultimate answer was going to be to scan the text and then have the computer perform Optical Character Recognition (OCR) to extract the text from the image.  I have done this in many cases and there certainly have been instances where the OCR performs flawlessly and in those cases this is the best option.  

It's not a magic bullet of course.  I could go on at length about why it doesn't work well but suffice it to say that there will be some instances where the OCR fails so badly that you end up retyping the article almost entirely to fix all of the mistakes.  This has happened in too many cases for me to keep trying except in those cases where the type on the page is clear and clean.

Speech Recognition Works for Me

Voice to Text

I now prefer to use voice-to-text.  I read the documents into a microphone connected to the computer and the computer converts my speech into text.  I find that though voice-to-text can struggle with names and places the end result has far less errors in it than OCR.  Also, the errors are far more obvious, for example when I speak the name "Vasile Banda" to my computer it usually types-in "the silly panda".  Pretty obvious.  OCR might see "Vasile Banda" and give you "Vasiie Bamda" for example.  Harder to see.

When OCR fails it can fail to correctly recognize every letter in every word whereas voice-to-text has only a single chance to fail on each word.  With the microphone I use and the pronunciation I have I find this is a huge time saver.  Your mileage may vary!

I no longer fear the amount of time it will take to capture text within a Source or for an anecdote to attach to a person.  When things are easier people have a greater chance to doing them.  Having more stories and quotations in your family history documents means they can be richer and more valuable for your consumers.