Synthesis of sung Spanish vowels in lyrical singing by sopranos using the Fant source-filter theory and the Rosenberg glottal pulse model