This might result in jammed audio for human listeners, but recovering the original audio seems like a fairly mundane signals extraction problem subject to the standard signal/noise ratio issue.