Abstract
The perception of melody in speech depends mainly on the fundamental frequency (f0) which reflects vocal fold oscillation speed. Whisper is defined by the absence of phonation and therefore the lack of f0. Intended melody in whisper, however, seems to be discernible regardless.
This paper presents a perception experiment assessing the discernibility of melody in whispered Czech words and words sung in whisper, which proved that melody in whisper in certain cases can in fact be discerned, along with an acoustical analysis of the effect of intended melody in whisper on formant frequencies, formant to formant ratios, center of gravity (CoG) and spectral slope. The parameters affected by intended melody in whispered speech were F2 and CoG of stopband filtered signal with main formant bandwidths removed. In words sung in whisper, the affected parameters were F2, F3, F2:F1 and F3:F2 ratios, CoG and spectral slope.