Multi-Modal Emotion Detection using Deep Learning Techniques

Accurately detecting emotions and analysing it, especially in real-time, plays a major role in bolstering human-computer interaction across a spectrum of applications like virtual assistants, call centers, and cyber-security. Highly precise emotion recognition can potentially transform user experience and provide some invaluable insights. In this paper, an attempt to explore multimodal emotion detection using audio signals in the form of. wav audio clips and images that have been grey-scaled by employing appropriate advanced deep-learning techniques and previously available machine learning classifiers. By the efficient and succinct usage of various Machine Learning Architectures and techniques, it must be noted that separate datasets have been used for both images and audio. An effective combination of both further strengthens the possibility of achieving superior performance.