Multimodal Speech Emotion and Text Sentiment Analysis
Nur Bengisu Çam, İlknur Dönmez, Ömer Faruk Bitikçioğlu, Fadime Buse Yediparmak, Emre Bektaş, Mehmet Haklıdır · 2023
Today, every industry uses sentiment analysis as machine learning, and artificial intelligence techniques to evaluate client satisfaction with a given product or service. In our study, we investigated how combining voice sentiment analysis with text sentiment analysis influences outcomes compared to employing only one of the analysis methods for small size of data. In our experiments, we had 96% accuracy for the test data of the speech sentiment analysis model and 78% accuracy for the test data of the text sentiment analysis model. By dint of the Multimodal model methods, the combined use of the two models exceeded the sound test results and reached 98% accuracy. This shows the positive effect of the Multimodal model method on sentiment analysis performed with little data.