Does Visual Self-Supervision Improve Learning of Speech Representations for Emotion Recognition?
This work addresses the challenge of improving speech emotion recognition, especially in noisy conditions and with small datasets, by leveraging cross-modal self-supervision, though it is incremental as it builds on existing self-supervised learning frameworks.
The paper tackles the problem of learning speech representations for emotion recognition by investigating visual self-supervision via face reconstruction and proposing an audio-only approach, showing that a multi-task combination outperforms existing self-supervised methods on downstream tasks like discrete emotion recognition and automatic speech recognition.
Self-supervised learning has attracted plenty of recent research interest. However, most works for self-supervision in speech are typically unimodal and there has been limited work that studies the interaction between audio and visual modalities for cross-modal self-supervision. This work (1) investigates visual self-supervision via face reconstruction to guide the learning of audio representations; (2) proposes an audio-only self-supervision approach for speech representation learning; (3) shows that a multi-task combination of the proposed visual and audio self-supervision is beneficial for learning richer features that are more robust in noisy conditions; (4) shows that self-supervised pretraining can outperform fully supervised training and is especially useful to prevent overfitting on smaller sized datasets. We evaluate our learned audio representations for discrete emotion recognition, continuous affect recognition and automatic speech recognition. We outperform existing self-supervised methods for all tested downstream tasks. Our results demonstrate the potential of visual self-supervision for audio feature learning and suggest that joint visual and audio self-supervision leads to more informative audio representations for speech and emotion recognition.