Harmonizing Emotions: A Novel Approach to Audio Emotion Classification using Log-Melspectrogram with Augmentation

This study article explores the field of audio emotion categorization, using a unique method that involves log-Mel spectrogram with augmentation. The research shows a respectable accuracy rate of 63%, which is significantly lower than MFCC. However, it also highlights the significant potential for development. The study utilises a sophisticated model, namely a 2D CNN that resembles the VGG19 architecture. In this model, the audio characteristics are treated as a 30x216 pixel picture. By using datasets such as TESS, CREMA-D, SAVEE, and RAVDESS, specialised functions are used to extract a wide range of characteristics. The importance of promptly identifying emotions via auditory cues is underscored. The 2D CNN model demonstrates encouraging outcomes, with the capacity to exceed conventional approaches. A specialised class is used to contain result characteristics in order to facilitate faster interpretation. This study not only enhances the expanding area of audio emotion classification but also creates opportunities for effective and precise identification, bridging the divide between audio and image-based methods in emotion analysis.