Why not letting the first amplifying stage after the microphoneto be the most linear at low levels and gradualy non linear at higher level?

After all is this not what we do when we listen to Frank Sinatra on YouTube? Or what nature does with human or cat ears-brain in order to extract as much information?