Automatic glottal inverse filtering with non-negative matrix factorization

This study presents an automatic glottal inverse filtering (GIF) technique based on separating the effect of the glottal main excitation from the impulse response of the vocal tract. The proposed method is based on a non-negative matrix factorization (NMF) based decomposition of an ultra short-term spectrogram of the analyzed signal. Unlike other state-of-theart GIF techniques, the proposed method does not require estimation of glottal closure instants. The proposed method was objectively evaluated with two test sets of continuous synthetic speech created with a glottal vocoding analysis/synthesis procedure. When compared to a set of reference GIF methods, the proposed NMF technique shows improved estimation accuracy especially for male voices.