I’m trying to provide the option for the user to disable/enable voice control such that there is no overhead associated with the controller. I worried that even though I can suspend/resume the controller, it is still consuming CPU time.
My plan now is to create all my OEEventsObserver and proxies first. Then call startRealtimeListeningWithLanguageModelAtPath only if the user has enabled voice control. I can call stopListinging later if the user disables voice input.
I expect to make the final decision to purchase Rapid Ears soon, although the price is daunting for an individual developer.
BTW, I really appreciate the way you have set up the demo version of RapidEars. I’m able to do a complete evaluation before purchasing.
If you are interested, I’m adding voice control (and later Google Cardboard support) to my app, Walkers 3D. http://walkers3d.com