Cross-Linguistic Distinctions Between Professional and Non-Professional Speaking Styles

This work investigates acoustic and perceptual differences in four language varieties by using a corpus of professional and non-professional speaking styles. The professional stimuli are composed of excerpts of broadcast news and political discourses from six subjects in each case. The non-professional stimuli are made up of recordings of 10 subjects who read a long story and narrated it subsequently. All this material was obtained in four language varieties: Brazilian and European Portuguese, standard French and German. The corpus is balanced for gender. Eight melodic and intensity parameters were automatically obtained from excerpts of 10 to 20 seconds. We showed that 6 out of 8 parameters partially distinguish professional from non-professional style in the four language varieties. Classification and discrimination tests carried out with 12 Brazilian listeners using delexicalised speech showed that these subjects are able to distinguish professional style from non-professional style with about 2/3 of hits irrespective of language. In comparison, an automatic classification using an LDA model performed better in classifying non-professional (96 %) against professional styles, but not in classifying professional (42 %) against non-professional styles.