When listening to speech sounds,our brains must compute a number of features from the sound wave,including the pattern of so called formant spectral peaks which determine the sounds timbre and identify its phonetic identity,as well as the voice pitch and sound source direction.Furthermore,these computations need to be achieved in a manner which is "noise robust",given that adding background noise to a speech stream disrupts our ability to process speech less than might be expected.Little is known about how these processing steps are accomplished in the human brain,but given that many of the fundamental aspects ofvocal communication are generic across all mammals,we have been conducting a number of studies in ferrets,rats,gerbils and mice which aim to shed light on the neural processing of vocalisations and speech sounds in the ascending auditory pathway,and in this talk I will present some of the highlights of our recent work.