Automatic Speech Emotion Recognition Using Machine Learning
Different features were extracted → and those are used to predict emotions → raw audio signals might not be the ideal representation of data for classification.
From audio data, → if we are able to extract emotions quite a lot of applications can be developed on top of that web-based application as well as medical applications.
Games as well as educational stuff → or even surveillance. (the emotional state should be considered during educations).
A lot of researchers → like to use a lot of representations. However, those will contain a lot of redundant information. Hence it will be easily overfitted as well as the learning process will be more complicated. (hence not a good idea to use every feature).