In this paper, we propose a novel architecture for multi-modal speech and text input. We combine pretrained speech and text encoders using multi-headed cross-modal attention and jointly fine-tune on the target problem. The resultant architecture can be used for continuous token-level classification or utterance-level prediction acting on simultaneous text and speech. The resultant encoder efficiently captures both acoustic-prosodic and lexical information. We compare the benefits of multi-headed attention-based fusion for multi-modal utterance-level classification against a simple concatenation of pre-pooled, modality-specific representations. Our model architecture is compact, resource efficient, and can be trained on a single consumer GPU card.
@article{arxiv.2204.09227,
title = {Cross-stitched Multi-modal Encoders},
author = {Karan Singla and Daniel Pressel and Ryan Price and Bhargav Srinivas Chinnari and Yeon-Jun Kim and Srinivas Bangalore},
journal= {arXiv preprint arXiv:2204.09227},
year = {2022}
}