Context-sensitive multimodal emotion recognition from speech and facial expression using bidirectional LSTM modeling | Synapse