The focus should be on generating high-quality audio and video using less expensive equipment like cheap microphones and a bit of intentional movement, which provides enough information for excellent subpics or deep learning. The same approach used for super resolution in generating high-quality images could also work for audio.