BLUF: when I tell ChatGPT to “end conversation”, ChatGPT should at least have the ability to actually invoke the “end conversation” function, or whatever it is called in the API.
Give ChatGPT the ability to access some of these basic API calls to control the app through the voice mode during a chat conversation. “Pause conversation”, “end conversation”, “mute microphone”, “switch model”, etc, should all be available as voice commands during a voice chat.
I already hear the comments about “mute microphone”. The app should be intelligent enough to mute the microphone as a function inside the conversation, not the actual physical mic being used. ChatGPT should then only be listening for a command such as “unmute microphone” to then allow ChatGPT to listen for full speech recognition. This isn’t too complicated.