For the development of high performance text-to-speech systems, as well as for the implementation of feature based Automatic Speech Recognition systems accurate tracking of speech parameters like fundamental frequency and formant frequencies and bandwidths is essential. Moreover, applications will benefit if the results of the parameter tracking may be interpreted in a way that appeals to phoneticians. In the case of formant parameters this means that 'formants' preferably should mean 'vocal tract resonances', rather than mere spectral maxima.