Okay, this is not honest and I have not seen any evaluations about this.
Real Time API Voice and Chat GPT Real Time Voice are totally different technologies. The quality difference is incredible in tonality, stressing, naturality. Any Text to Speech Model has better quality than Real Time API. Yes it’s “real time” and latency is incredibly low. But it does not even understand question mark “?” It does not sound like asking a question when you use a question mark at the end of a sentence. This is unaccaptable.
Open Ai needs to release the same exact Chat GPT voices on the API side ASAP.