The interaction between individuals and computers would be more natural if computers are ready to perceive and respond to human non-verbal communication like emotions. Although several approaches are proposed to acknowledge human emotions based on facial expressions or speech, relatively limited work has been done to fuse these two, and other, modalities to enhance the accuracy and robustness of the emotion recognition system, [1]. in this project we are developing an emotion recognition model using Deep Networks. Speech being a sequential data model, Recurrent Neural Network [2] can be able to process emotion of an individual's being. There are many of temporal and spectral features which can be extracted from human speech. The features are chosen to represent intended information. we've used the foremost used features that are available in librosa-library [3] including MFCC [4], Chroma [5], MEL Spectrogram Frequency (MEL), [6].
No takes yet. Share an insight, caveat, or question.
Sriharsha et al. (2022) studied this question.