In theory could it work like this.
Take the TT voices, split them to ogg files. Then convert these ogg files to wav and use Garmin Voice Studio to convert them to Garmin Voices.
I planned to test this already but it will take huge amount of time