Using the gpt voice mode for language learning is arguably it’s most useful use case, and yet for years it lacked both the ability to quite the model from interrupting you while thinking (which happens ALOT learning a language) and understanding your speech in a native way that isn’t basic speech to text conversion (like the basic voice model that does have hold to talk still).
I recall in mid-june the advanced voice mode did have hold to talk functionality then it randomly disappeared. Now both the advanced voice mode and the live mode have no wait to let you finish your thoughts without interruption unless you already know what you’re going to say and how you’re going to say it. The live version is a bit better than the advanced voice mode model but it still doesnt wait for you to finish your conceptual/contextual thought before interrupting.
Why is it raking years to have a basic workaround such as hold to talk employed consistently by default?