Speech Emotion Recognition Using Multi-hop Attention Mechanism
This work addresses emotion classification from speech for applications like human-computer interaction, but it is incremental as it builds on existing multimodal fusion methods.
The paper tackled speech emotion recognition by proposing a multi-hop attention mechanism to exploit acoustic and textual data in tandem, achieving a 6.5% relative improvement in weighted accuracy over the state-of-the-art on the IEMOCAP dataset.
In this paper, we are interested in exploiting textual and acoustic data of an utterance for the speech emotion classification task. The baseline approach models the information from audio and text independently using two deep neural networks (DNNs). The outputs from both the DNNs are then fused for classification. As opposed to using knowledge from both the modalities separately, we propose a framework to exploit acoustic information in tandem with lexical data. The proposed framework uses two bi-directional long short-term memory (BLSTM) for obtaining hidden representations of the utterance. Furthermore, we propose an attention mechanism, referred to as the multi-hop, which is trained to automatically infer the correlation between the modalities. The multi-hop attention first computes the relevant segments of the textual data corresponding to the audio signal. The relevant textual data is then applied to attend parts of the audio signal. To evaluate the performance of the proposed system, experiments are performed in the IEMOCAP dataset. Experimental results show that the proposed technique outperforms the state-of-the-art system by 6.5% relative improvement in terms of weighted accuracy.