Preparing audio for speech-to-text with ffmpeg: what to change and what to leave alone

Most recordings can go to a speech-to-text system exactly as they are. A few ffmpeg commands still earn their place: they shrink an upload, keep two speakers apart, or prepare a file for a model you r