Text this: Audio-visual expression-based emotion recognition model for neglected people in real-time