Spread Lips + Raised Larynx + Higher F0 = Smiled Speech? – An Articulatory Synthesis Approach

We present an initial study on how to model smiled speech with an articulatory speech synthesizer, led by the research question as to what cues are responsible for the effect of an audible distinction of smiled vs. non-smiled speech. In a perception test, we explore the relative contributions of i) spreading of the lips, ii) raising of the larynx, and iii) raising of the fundamental frequency. 36 test subjects assessed isolated synthetic vowel stimuli of /a:, i:, y:, u:/ on a 5-point “smiley scale”. Results indicate that F0 is the main acoustic factor for perceiving smileyness. The other factors depend on the vowel quality, with best results for the unrounded vowels /i:/ and /a:/.