Sorry for seeing this question earlier! I am terrible person.
All the tts videos I have posted here do happen in realtime. However, when you request tts from CereVoice Cloud, there obviously some amout of delay to get a reply. The reply provides a url for the wav file and a url to a metadata file, which contains timing information of the words in the audio file. The lenght of this delay increases with the word count of the request but it is pretty quick.
More recently I have made it so that all the dialog urls and metadata url are generated before the animation is played back, however this does still download and playback the audio file in as close to realtime as possible.
I just posted here with the alternative option to just enter the words into a Sequencer timeline along side a standard wav file.
Thanks for watching and commenting. I promise to reply much swifter next time…