Gianluca Panici

Light-SERNet: a fully convolutional neural network for speech emotion recognition

Light-SERNet is a lightweight fully convolutional neural network for speech emotion recognition. With the evolution of the technological scenario where the human-computer interaction has become a daily constant in our society, the detection of human emotions from a speech signal plays an important role in this exchange between humans and machines. Existing benchmarks of speech emotion recognition (SER) methods are mainly comprised of a feature extractor and a classifier to obtain the emotional states. Recently, deep learning techniques are used to extract high level features from raw data and it is shown that they are effective for speech emotion recognition. In this study, convolutional neural networks (CNNs) are used, due to the significant improvements in SER. This kind of tools are useful especially when it is required to divide the information in smaller chunks to analyze them and find out some characteristics that may appear irrelevant to the target task, and so they find their perfect place when the input is a complex unstructured signal, such as image or a speech signal. See the full report HERE

MORE