so, here is Mi Dica. it's a combination of erel's voice recognition example plus my groundbreaking work on MLKit using b4a as a driver. speak in your device's default language (or change it to something else if you like), have it translated into some language spoken by a person you want to talk to, and have the device speak your utterance in their language.
flip a switch, and that person can speak to you in their language, and you hear it in yours. what could be simpler?
here are a couple screen captures, and there is a short video on youtube
there seems to be a problem recording audio from both the microphone and the device's speakers at the same time. there is a setting that's supposed to allow that, but i couldn't get it to...