作者: Christian Hacker , Elmar Nöth , Stefan Steidl , Heinrich Niemann , Anton Batliner
DOI:
关键词:
摘要: Classification performance for emotional user states found in the few realistic, spontaneous databases available is as yet not very high. We present a database with children’s speech human-robot scenario. Baseline classification seven classes 44.5%, four 59.2%. discuss possible strategies tuning, e.g., using only prototypes (based on annotation correspondence or scores), taking into account requirements and feasibility applications (weighting of false alarms speakerspecific overall frequencies).