If you couldn't process multiple streams ( I see nothing special about speech - note parsing speech is potentially different from thinking over the top - ie the difference between hearing and listening ) simultaneously it's hard to see how you'd last 5 minutes.