On Building Spoken Language Understanding Systems for Low Resourced Languages

TL;DR

This paper explores methods for building spoken language understanding systems in extremely low-resource settings, demonstrating that phonetic transcriptions outperform speech features in intent classification tasks for low-resourced languages.

Contribution

The study introduces a novel approach using phonetic transcriptions for intent classification in low-resource languages, bypassing the need for extensive ASR data.

Findings

Phonetic transcription-based systems outperform speech feature-based systems by over 12% in accuracy.

Effective intent classification is achievable with as little as one data-point per intent and a single speaker.

The approach works on languages with limited or no written form, like Flemish and English.

Abstract

Spoken dialog systems are slowly becoming and integral part of the human experience due to their various advantages over textual interfaces. Spoken language understanding (SLU) systems are fundamental building blocks of spoken dialog systems. But creating SLU systems for low resourced languages is still a challenge. In a large number of low resourced language, we don't have access to enough data to build automatic speech recognition (ASR) technologies, which are fundamental to any SLU system. Also, ASR based SLU systems do not generalize to unwritten languages. In this paper, we present a series of experiments to explore extremely low-resourced settings where we perform intent classification with systems trained on as low as one data-point per intent and with only one speaker in the dataset. We also work in a low-resourced setting where we do not use language specific ASR systems to transcribe input speech, which compounds the challenge of building SLU systems to simulate a true low-resourced setting. We test our system on Belgian Dutch (Flemish) and English and find that using phonetic transcriptions to make intent classification systems in such low-resourced setting performs significantly better than using speech features. Specifically, when using a phonetic transcription based system over a feature based system, we see average improvements of 12.37% and 13.08% for binary and four-class classification problems respectively, when averaged over 49 different experimental settings.