In Multimodal Emotion Recognition (SER), Emotional Characteristics Often Appear In Diverse Forms Of Energy Patterns In Spectrograms. Typical Attention Neural Network Classifiers Of SER Are Usually Optimized On A Fixed Attention Granularity. In This Paper, We Apply Multiscale Area Attention In A Deep Convolutional Neural Network To Attend Emotional Characteristics With Varied Granularities And Therefore The Classifier Can Benefit From An Ensemble Of Attentions With Different Scales. To Deal With Data Sparsity,we Conduct Data Augmentation With Vocal Tract Length Perturbation (VTLP) To Improve The Generalization Capability Of The Classifier. We Can Classified Three Various Emotion Detection In Real-time (speech,face,text) Experiments Are Carried Out On The Interactive Emotional Dyadic Motion Capture (IEMOCAP) Dataset. Which, To The Best Of Our Knowledge, Is The State Of The Art On This Dataset.
