An elementary waveform speech model (EWSM) is defined and some capabilities are demonstrated for the modification of localized time-frequency events. The elementary waveforms allow for modelling the local spectro-temporal maxima of energy inside the speech signal by simple mathematical functions. EWSM parameters axe estimated using a frame by frame processing: spectral modelling and segmentation using short-time Fourier transform and LPC spectrum, Fourier filtering according to this segmentation, waveforms spotting in each channel waveform modelling with simple functions. The EWSM parameters are relevant according to the classical theory of speech production, and their modifications yield well-localized time-frequency transformations, including frequency compression/expansion, pitch, formant, noise modifications.