SDCLASOct 31, 2021

Speech Emotion Recognition Using Quaternion Convolutional Neural Networks

arXiv:2111.00404v150 citations
Originality Incremental advance
AI Analysis

This addresses the challenge of inferring emotion from speech signals for applications in human-computer interaction, though it is incremental as it builds on existing methods with a novel encoding approach.

The paper tackled speech emotion recognition by proposing a quaternion convolutional neural network (QCNN) model that encodes Mel-spectrogram features in an RGB quaternion domain, achieving state-of-the-art accuracy of 77.87% on RAVDESS and comparable results on other datasets like 70.46% on IEMOCAP and 88.78% on EMO-DB.

Although speech recognition has become a widespread technology, inferring emotion from speech signals still remains a challenge. To address this problem, this paper proposes a quaternion convolutional neural network (QCNN) based speech emotion recognition (SER) model in which Mel-spectrogram features of speech signals are encoded in an RGB quaternion domain. We show that our QCNN based SER model outperforms other real-valued methods in the Ryerson Audio-Visual Database of Emotional Speech and Song (RAVDESS, 8-classes) dataset, achieving, to the best of our knowledge, state-of-the-art results. The QCNN also achieves comparable results with the state-of-the-art methods in the Interactive Emotional Dyadic Motion Capture (IEMOCAP 4-classes) and Berlin EMO-DB (7-classes) datasets. Specifically, the model achieves an accuracy of 77.87\%, 70.46\%, and 88.78\% for the RAVDESS, IEMOCAP, and EMO-DB datasets, respectively. In addition, our results show that the quaternion unit structure is better able to encode internal dependencies to reduce its model size significantly compared to other methods.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes