An Attention Self-supervised Contrastive Learning based Three-stage Model for Hand Shape Feature Representation in Cued Speech
This work addresses feature extraction for Cued Speech, a communication system for deaf or hearing-impaired people, by improving recognition accuracy with a novel model, though it is incremental as it builds on existing deep learning approaches.
The paper tackles the problem of noisy annotations in Cued Speech recognition by proposing a three-stage model that uses self-supervised contrastive learning, fine-tuning, and sequential feature learning, achieving over 90% accuracy in hand shape recognition and improving phoneme recognition correctness by 8.75% for French and 10.09% for British English compared to state-of-the-art methods.
Cued Speech (CS) is a communication system for deaf people or hearing impaired people, in which a speaker uses it to aid a lipreader in phonetic level by clarifying potentially ambiguous mouth movements with hand shape and positions. Feature extraction of multi-modal CS is a key step in CS recognition. Recent supervised deep learning based methods suffer from noisy CS data annotations especially for hand shape modality. In this work, we first propose a self-supervised contrastive learning method to learn the feature representation of image without using labels. Secondly, a small amount of manually annotated CS data are used to fine-tune the first module. Thirdly, we present a module, which combines Bi-LSTM and self-attention networks to further learn sequential features with temporal and contextual information. Besides, to enlarge the volume and the diversity of the current limited CS datasets, we build a new British English dataset containing 5 native CS speakers. Evaluation results on both French and British English datasets show that our model achieves over 90% accuracy in hand shape recognition. Significant improvements of 8.75% (for French) and 10.09% (for British English) are achieved in CS phoneme recognition correctness compared with the state-of-the-art.