Analyzing the Relationship between Formants and Pitch for Singing Voice

Abstract

Singing voice synthesis is an interesting research due to its wide application in entertainment systems. The generation of singing voice normally replies on techniques of changing the fundamental frequency of existing vocal recording. When the fundamental frequency of vocal signal is changed to match the intended pitch contour of the song, the melody is correctly implemented. However, if there is a great change over the pitch of a vocal signal, the spectrum of the signal should also change accordingly so as to generate a natural singing voice. Unfortunately, currently the spectrum differences for singing voice at different pitch level are unclear. Therefore, the relationship between the spectrum and pitch needs to be examined. As formants are important elements of spectrum, therefore in this research we look at the relationship between formants and pitch. For a start, we first examine the sound of 'ah' at different pitch levels for both male and female voices.