What happened to AVSpeechSynthesizer between 12.0 and 12.1

Hello
Using AVSpeechSynth. It's wonderful, easy to use. I'm wondering what has happened between 12.0 and 12.1 such that the default voice no longer takes punctuation into account and adjusts pronunciation accordingly. Previously, "great job!" sounded exciting and encouraging. It was a distinctly different phoenetic experience from "great job." In 12.1, "great job!" and "great job" are aurally identical. What happened?
In 12.0 I didn't bother to set the utterance's voice. It used whatever the default voice in 12.0 was. I think that voice may have been Siri? I really don't know though. A couple users upgraded to 12.1 and started complaining, the voice was "wrong." I have since tried setting the voice on the utterance, but that isn't making much happen either.
We're building a reinforcement app so encouragement is an extremely fundamental component. We went from a really great UX, to something that's borderline mocking, disheartening, and a terrible UX with no understanding as to why that's happened and no way to revert things to the way they were.
Any suggestions are more than welcome.

Up vote post of nickneedsaname Down vote post of nickneedsaname
933 views
Posted by
nickneedsaname
Reply
Add a Comment

Replies

Seeing the same. Not a change for the better.

Posted by
snydersaurus
Up vote reply of snydersaurus Down vote reply of snydersaurus
Add a Comment

Well, the only way to get around this problem is to select the voice that you think works best for your use case and then force your user to load the voice first in the settings app.


You can do this by checking the return value of your 

init?(identifier: String)
method. If it's nil the voice is not installed on the device yet and you can promt the user to download the voice first by stopping your voice output.


The different intonations you can achieve by punctuation and other little tricks are not synthesizer specific but rather voice specific, what works on one voice may not on another voice. What works in one language won't in another.


The only uncertainty that will still remain is that we don't know whether Apple is alternating or improving individual voices which might break your desired intonation.


I haven't been able yet to get into contact with a voice engineer in order to find out whether there are updates or version numbers available for the voices and how often they change things there.

Posted by
mramosch
Up vote reply of mramosch Down vote reply of mramosch
Add a Comment