In a PDF each word and sometimes letter is individually placed on the page. Some will be better than others depending on how the page was produced and if there were any edits performed on the page and other factors. One technique that sometimes works is to print the document to a PDF printer. There are some free ones out there and then convert that. Otherwise the best technique is to OCR the pages and correct any errors. A PDF is an end use document. It is not a source file and cannot reliably be converted to anything else. Some times it will work but another document will fail.