Speech-Based Emotion Analysis Using Log-Mel Spectrograms and MFCC Features

This study proposes a method for recognizing emotions from speech using Mel spectrograms and MFCC features which capture the spectral features of speech signals. To predict emotions from the extracted features from the dataset, Convolutional Neural Networks (CNNs) and finetune pre-trained models are used. Pre-trained models are fine-tuned with some optimizations and one-dimensional convolutional neural network is constructed. The results demonstrate that the proposed method achieved an accuracy rate of over 80% in predicting emotions from speech and show the effectiveness of the approach in a comparative manner.