Speech Emotion Recognition with Multiscale Area Attention and Data Augmentation
This work provides an incremental improvement in speech emotion recognition accuracy for researchers and practitioners working with the IEMOCAP dataset.
This paper addresses Speech Emotion Recognition (SER) by proposing a multiscale area attention mechanism within a deep convolutional neural network to capture diverse emotional energy patterns in spectrograms. Additionally, vocal tract length perturbation (VTLP) is used for data augmentation to combat data sparsity. The method achieved 79.34% weighted accuracy (WA) and 77.54% unweighted accuracy (UA) on the IEMOCAP dataset, setting a new state of the art.
In Speech 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. Experiments are carried out on the Interactive Emotional Dyadic Motion Capture (IEMOCAP) dataset. We achieved 79.34% weighted accuracy (WA) and 77.54% unweighted accuracy (UA), which, to the best of our knowledge, is the state of the art on this dataset.