CNN+LSTM Architecture for Speech Emotion Recognition with Data Augmentation
This work addresses emotion recognition in speech, which is important for applications like human-computer interaction, but it appears incremental as it combines established CNN and LSTM components with standard augmentation methods.
The paper tackled speech emotion recognition by designing a CNN+LSTM neural network with data augmentation techniques, achieving competitive results of 64.5% weighted accuracy and 61.7% unweighted accuracy on the IEMOCAP dataset for four emotions.
In this work we design a neural network for recognizing emotions in speech, using the IEMOCAP dataset. Following the latest advances in audio analysis, we use an architecture involving both convolutional layers, for extracting high-level features from raw spectrograms, and recurrent ones for aggregating long-term dependencies. We examine the techniques of data augmentation with vocal track length perturbation, layer-wise optimizer adjustment, batch normalization of recurrent layers and obtain highly competitive results of 64.5% for weighted accuracy and 61.7% for unweighted accuracy on four emotions.