Towards Reducing the Need for Speech Training Data To Build Spoken Language Understanding Systems
This addresses the data scarcity issue for SLU system builders, offering a practical solution to reduce reliance on costly speech annotations, though it is incremental in improving existing methods.
The paper tackles the problem of limited annotated speech data for spoken language understanding (SLU) by proposing a text representation and training method that leverages abundant text resources, achieving up to 90% of full speech performance with text-only training and 97% with only 10% additional speech data.
The lack of speech data annotated with labels required for spoken language understanding (SLU) is often a major hurdle in building end-to-end (E2E) systems that can directly process speech inputs. In contrast, large amounts of text data with suitable labels are usually available. In this paper, we propose a novel text representation and training methodology that allows E2E SLU systems to be effectively constructed using these text resources. With very limited amounts of additional speech, we show that these models can be further improved to perform at levels close to similar systems built on the full speech datasets. The efficacy of our proposed approach is demonstrated on both intent and entity tasks using three different SLU datasets. With text-only training, the proposed system achieves up to 90% of the performance possible with full speech training. With just an additional 10% of speech data, these models significantly improve further to 97% of full performance.