Speech emotion analysis using convolutional neural network (CNN) and gamma classifier-based error correcting output codes (ECOC)

In this section, after describing the specification of datasets used in the research, the details of the proposed method for emotion analysis in speech signal of these datasets is presented.

Data acquisition

In this research, the speech signals available in ShEMO²² and Berlin Emotional²³ datasets were used as input for experiments. In the following, the specifications of the mentioned datasets are described first, and then the obtained results are analyzed.

The data set used from the Berlin dataset is 535 speech signals in six emotional categories, and a subset including 150 speech signals…