Decoding Sentiment: An Enhanced PCA in Tagalog Speech Using Prosodic Suprasegmental Features

Prosodic Suprasegmental Features (PSF) and Principal Component Analysis (PCA) are techniques used in speech sentiment analysis that are capable of capturing the emotional content of speech sentiment and providing increased accuracy. However, PSF is prone to high dimensional features of data and PCA is sensitive to outliers. This study explores the effectiveness of enhanced PCA sentiment analysis in Tagalog speech, focusing on prosodic suprasegmental features such as pitch, duration, energy/intensity, and intonation. Using the optimized PCA, the retrieved prosodic features' dimensionality was decreased, and it was then used to train a Support Vector Machine (SVM) classifier, which divided speech into three categories: positive, negative, and neutral. The model achieved an accuracy of 82%, with an f1-score of 81% and recall of 82% as compared to no PCA enhancement and the simple speech sentiment analysis having an accuracy of 64% and 14% respectively. The results demonstrate the effectiveness of prosodic suprasegmental features and enhanced PCA in capturing emotional nuances within spoken Tagalog and provide a solid foundation for sentiment analysis in speech, providing insightful information on the emotional content of Tagalog speech.