Textual Supervision for Visually Grounded Spoken Language Understanding
This work addresses the challenge of building effective spoken language understanding systems for low-resource languages where transcriptions are scarce or costly.
The paper tackled the problem of improving visually-grounded spoken language understanding models by comparing end-to-end and pipeline approaches when transcriptions are available, finding that the pipeline approach performs better with sufficient text data and that translations can substitute transcriptions in low-resource settings but require more data.
Visually-grounded models of spoken language understanding extract semantic information directly from speech, without relying on transcriptions. This is useful for low-resource languages, where transcriptions can be expensive or impossible to obtain. Recent work showed that these models can be improved if transcriptions are available at training time. However, it is not clear how an end-to-end approach compares to a traditional pipeline-based approach when one has access to transcriptions. Comparing different strategies, we find that the pipeline approach works better when enough text is available. With low-resource languages in mind, we also show that translations can be effectively used in place of transcriptions but more data is needed to obtain similar results.