When a voice input contains multiple languages, so there is a "Language 1" sentence (let's call this language part 1), followed by a "Language 2" sentence (language part 2), language part 1 is ignored. There seems to be a voice input level processing that occurs which can translate all of the input to English, or when multiple languages are used, and where English is followed by another language, only the other lnaguage is used. Happy to help with any bug fix and testing.
#GPT Multilingual voice input - sections of input ignored.
prime trout