>Hey Claude, translate this to Swahili from English. Ok, now translate my response from Swahili to English. Thanks.
We're close to the point where using a human -> stt -> llm -> tts -> human pipeline you can do real time high quality bi directional spoken translation on a desktop.