15th International Congress of Phonetic Sciences (ICPhS-15)

Barcelona, Spain
August 3-9, 2003

Speaker Age: A First Step from Analysis to Synthesis

Susanne Schötz

Lund University, Sweden

A first tentative step towards synthesis of age was made in three pilot studies of speaker age. The first was a perceptual study, where results showed that some speakers are more typical for their age than others, and that listeners are able to age estimate single word stimuli considerably better than chance. In the second study an auditive and acoustic analysis of the same material indicated that older voices contain more variability, that some cues, including spectral quality and VOT, may be more important than others (e.g. F0), and that combinations of several cues are probably used for age perception. In the third study, natural, synthesized and resynthesized stimuli were used in a second listening test. Results indicated that spectral information and duration are more important than F0 for age perception. The results from these studies will be used in future research with more data and formant synthesis.

Full Paper

Bibliographic reference.  Schötz, Susanne (2003): "Speaker age: a first step from analysis to synthesis", In ICPhS-15, 2585-2588.