MOTI: A Motivational Prosody Corpus for Speech-Based Tutorial Systems

Konferenz: Sprachkommunikation - Beiträge zur 10. ITG-Fachtagung
26.09.2012 - 28.09.2012 in Braunschweig, Deutschland

Tagungsband: Sprachkommunikation

Seiten: 4Sprache: EnglischTyp: PDF

Persönliche VDE-Mitglieder erhalten auf diesen Artikel 10% Rabatt

Wolff, Susann; Brechmann, André (Special Lab Non-Invasive Brain Imaging, Leibniz Institute for Neurobiology, 39118 Magdeburg, Germany)

In this paper, we present an evaluated corpus of recorded natural human speech with controlled prosodic modulations. The construction of this corpus aims at an application in experimental setups examining the influence of prosodic modulations in computer-assisted learning environments. To this avail, we recorded a set of spoken feedback comments and pseudo-words in neutral as well as motivational prosodies. The motivational prosodies contained praising, blaming, and commiserative modulations. The recordings were evaluated by 24 untrained listeners and obtained stable recognition ratings between 70% and 95%, depending on prosodic category. In a recent experiment, the MOTI Corpus was put to use for the first time, providing results that point to the usefulness of employing motivational prosody in computer-assisted learning environments.