@nobuhojimichaan Would this work for any kind of audio conversation? And would the result always be smaller than the original or could it be more verbose? I’ve listened to your audio clips and I’m curious whether they could be transcribed in a way that would carry the meaning?