Advanced sequence learning approaches for emotion recognition using speech signals

The most significant and successful method of human communication is speech, which can also serve as a channel for human-computer interaction (HCI). The use of sensors to identify auditory emotions is an emerging field of HCI research. In the ongoing scenario, emotion recognition is a persistent issue that is crucial to real-time applications. Recognizing human emotions is difficult when analysing and predicting a person's behaviour from a collection of audio clips. This chapter is going to cover the concept of sequence learning for emotion recognition using LSTM, GRUs, and their modifications, such as multi-layer or deep LSTM and bi-directional LSTM networks. The successes and weaknesses of current LSTM/GRUs-based emotion recognition systems will be assessed and discussed. We'll also discuss the drawbacks of traditional RNNs and why LSTM is better than RNNs.