Funnily enough, I am also having similar problems from a different angle.
Part of the problem here is also the fancy unicode apostrophes and hyphens some of the editiors on here like to use. They swap a standard ASCII apostrophe as found on the keyboard to a prettified curly version as found on the printed page in a book. Because these can cause confusions in a file system there is a mechanism to swap these for “standard” characters.
Someone then expanded that substitution within MB to cover ALL accented characters. I believe this was being done for cheapo media players that had to have all ASCII text. (Also add in US based programmers who will have less experience of Hungarian, Turkish, etc)
Go and have a look into the options in Picard. Under METADATA there is an option to “Convert Unicode Punctuation characters to ASCII”. And then in the plugins there are plugins that will do even more character swaps (Non-ASCII Equivalents is an example)
When I avoid the plugins and just tick the Convert Unicode Punctuation characters to ASCII option it looks like I am getting just the punctuation swapped out and the text entered correctly.
I am no expert on this. Just a noob to Picard and its oddities. My ripping with EAC using MB data has led to some odd filenames. So I am only just digging deeper to untangle this.
In a different discussion on this, I was handed which points into the MB source and shows us what is being swapped out.
Even if you don’t know Python, it is possible to look down that list and see what is going on. IMHO this is far too wide. Too much of an all swapped or nothing.
Sorry for waffly post. I need to sit down and look carefully at what option does what as I have been getting confused on this.
(The other discussion - Unicode apostrophe standardization )