We present a simple but effective approach to pitch modification of speech based on a harmonic model. Building on our time-scaling algorithm [1], pitchmodification applies to a harmonically coded glottal wave estimate derived via a simple inverse filtering technique [3]. The modified glottal wave subsequently serves as input to an LPC vocal tract filter and the pitch-scaled speech is generated. Shape invariance is maintained in the glottal wave by exploiting the harmonic nature of the sine waves used to code each frame thus avoiding the need for \pitch pulse onset time" estimation. Furthermore, given its smooth shape it is not necessary to resample the glottal wave spectrum at the new harmonic frequencies. The original spectrum is merely compressed/expanded to produce the desired pitch change.