Integrating Self-supervised Speech Model with Pseudo Word-level Targets from Visually-grounded Speech Model
This addresses the challenge of improving semantic understanding in speech models for SLU applications without costly speech-text data, representing an incremental advance in training methodology.
The paper tackles the problem that self-supervised speech models focus on frame-level objectives, which limits semantic comprehension for spoken language understanding tasks, by proposing PW-HuBERT, a framework that integrates pseudo word-level targets from a visually-grounded speech model without needing speech-text paired data. The model shows superiority on four SLU benchmarks, though no concrete numbers are provided.
Recent advances in self-supervised speech models have shown significant improvement in many downstream tasks. However, these models predominantly centered on frame-level training objectives, which can fall short in spoken language understanding tasks that require semantic comprehension. Existing works often rely on additional speech-text data as intermediate targets, which is costly in the real-world setting. To address this challenge, we propose Pseudo-Word HuBERT (PW-HuBERT), a framework that integrates pseudo word-level targets into the training process, where the targets are derived from a visually-ground speech model, notably eliminating the need for speech-text paired data. Our experimental results on four spoken language understanding (SLU) benchmarks suggest the superiority of our model in capturing semantic information.