I might be oversimplifying, but isn't this the perfect argument for certain digital voice modes? DMR for example uses TDM to put two voice channels in a 12.5 KHz band.