A light transformer for speech-to-intent applications

TL;DR

This paper introduces a simplified, lightweight transformer model for speech-to-intent understanding that is efficient, quick to train, and effective across diverse speech conditions, enhancing user-taught SLU systems.

Contribution

The paper presents a novel light transformer with relative position encoding that reduces model size and training time for user-taught speech understanding systems.

Findings

Outperforms existing systems on three challenging datasets.

Uses half the model size and training time of previous models.

Effective in diverse speech conditions and for user-taught learning.

Abstract

Spoken language understanding (SLU) systems can make life more agreeable, safer (e.g. in a car) or can increase the independence of physically challenged users. However, due to the many sources of variation in speech, a well-trained system is hard to transfer to other conditions like a different language or to speech impaired users. A remedy is to design a user-taught SLU system that can learn fully from scratch from users' demonstrations, which in turn requires that the system's model quickly converges after only a few training samples. In this paper, we propose a light transformer structure by using a simplified relative position encoding with the goal to reduce the model size and improve efficiency. The light transformer works as an alternative speech encoder for an existing user-taught multitask SLU system. Experimental results on three datasets with challenging speech conditions prove our approach outperforms the existed system and other state-of-art models with half of the original model size and training time.