Improving the Learning Power of Artificial Intelligence Using Multimodal Deep Learning

Computer paralinguistic analysis is widely used in security systems, biometric research, call centers and banks. Paralinguistic models estimate different physical properties of voice, such as pitch, intensity, formants harmonics to classify emotions. The main goal find features that would be robust outliers will retain variety human voice at the same time. Moreover, model must able on a time scale for an effective variability. In this paper based Bidirectional Long Short-Term Memory (BLSTM) neural network described, which was trained vocal-based emotion recognition. advantage architecture each module consists several interconnected layers, providing ability recognize flexible long-term dependencies data, important context vocal analysis. We explain bidirectional model, its advantages over regular networks compare experimental results BLSTM with other models.