Back to news overview

Human-level performance for VokaturiPlus

16 January 2017

Vokaturi has reached a milestone: the software is now just as good at recognizing emotions from the voice as human listeners are. VokaturiPlus scores 76.1% correct (cross-validated) on two emotion databases, while humans guess the speaker's intended emotion correctly in 76 percent of the cases (Sanaul Haq, Philip J.B. Jackson & James Edge 2008: "Audio-visual feature selection and reduction for emotion classification").