I thought it would be useful to turn my work in progress into an audiobook so that I could listen to it during my commute to work. Turns out there is an easy way to do this using Mac OS X text to speech, but tweaking the output with Scrivener’s Compile function can make a big improvement to pronounciation.
Before you begin, go to System Preferences -> Dictation and Speech and make sure the default voice is one you like. You can download new voices here. For example there are male and female voices for many different variants of English as well as other languages.
In Scrivener, select the text you want to convert to speech and then right click, Speech -> Start Speaking.
You may find that the voice you choose mispronounces not only science fiction place names and character names, but also words that a human being would pronounce correctly. You can fix this with Scrivener’s compile.
Make notes of each mispronounced word. In TextEdit (to avoid messing up your work in Scrivener), type in the mispronounced word and then try to make a new spelling that will be pronounced correctly. Select original and new spellings and Edit -> Speech -> Start Speaking. Tweak the new spelling until it’s pronounced the way you want it.
When you have figured out all the spelling changes, use the Scrivener -> Compile -> Replacements function.
Create a new preset with a name (e.g. Mac OS Speech).
Select “Replacements” from the list on the left.
Select “Preset Replacements” rather than “Project Replacements.” - this should make sure the replacements are only made for your audiobook and not for print.
Click + and enter the original and new text to correct each term in your text.
When you have entered all the replacements, click compile (to plain text).
Open the plain text, select all, right click, Services -> Add to iTunes as spoken track.
Do this for each chapter; add the chapters to one iTunes playlist. Now you can play the playlist during your commute.
Note: for some reason, it’s a lot easier to miss an interstate exit or subway/metro/underground stop when you are listening to your own story!
Oh that’s interesting. I hate TextEdit (as an application) but if I could not only tweak the pronnunication with it but also switch voices (as the underlying Mac OS X API allows) then I’d use it to check dialogue. There is nothing like hearing dialogue to make one realise it isn’t actually dialogue at all and it needs to be rewritten to be real speech.
Glad people enjoyed this tip. One more thing to add - Mac OS X Voiceover allows you to create corrections for its pronunciation. (1/3 down this page apple.com/voiceover/info/guide/_1136.html)
The service that speaks the selection or turns the selection into a spoken track in iTunes does NOT use these corrections. It’s one of these oversights that Apple seemingly hasn’t gotten around to fixing. I
If you think you might use different voices with this, you may need to set up more than one compile preset with spelling corrections, as they pronounce words differently.
Side note - having “Daniel” read out emails from my wife is hilarious.
Quick note, another user brought up Apple’s SLA for Mac OS X, which states that the Voices services is for personal use only. Quoted from the 10.10.3 licence:
As I understand things, TTS and VO are discrete OS X services (and have been since they were introduced by Apple). I don’t know of a workflow to create a recording using VO, and the VO pronunciation customisations don’t have any impact on audio files created using TTS. That’s why your workaround is so useful.
Lots of websites and OS X users assume — fairly enough — that TTS and VO must share the pronunciation customisations found in VO. Unfortunately, they don’t.
I don’t have Hazel but I hear a lot about it on the Mac Power Users podcast. Hazel watches a folder and takes action on new files entering the folder. It can watch a folder on Dropbox. I’m wondering if you could compile from iOS to plain text in a Dropbox folder and have Hazel do the conversion to iTunes and store the result in another folder on Dropbox. Then you could pick up the converted files on iOS and listen to them - nearly as good as being able to do the conversion to speech on iOS.
