Suche

Italian Emotion Recognition


Description

The aim of this seminar is to classify Italian speech signals according to the emotional state of their speakers. To achieve this, the student will implement and evaluate state-of-the-art emotion recognition systems.

Task

The student is expected to review and implement state-of-the-art methods for emotion recognition applied to Italian utterances. The student is encouraged to investigate system performances using signal and image-based features, and to explore transfer-learning techniques on this problem.

Utilises Python, Keras, scikit-learn
Requirements Good programming skills, previous knowledge in machine learning
Languages English
Supervisor Adria Mallol-Ragolta (adria.mallol-ragolta@informatik.uni-augsburg.de)