Abstract
We present a new method for the estimation of the glottal volume velocity from voiced segments of the radiated acoustic speech pressure waveform. Our algorithm is based on spectral factorization of the signal and is a general purpose procedure. It does not suffer from residual effects or assume constraining models for the vocal tract and the glottal source, as is commonly the case with existing methods. The resulting estimate of the glottal volume velocity is accurate and can be used for modeling and synthesis purposes.