Investigations on speech recognition systems for low-resource dialectal Arabic–English code-switching speech

Abstract

Code-switching (CS), defined as the mixing of languages in conversations, has become a worldwide phenomenon. The prevalence of CS has been recently met with a growing demand and interest to build CS automatic speech recognition (ASR) systems. In this paper, we present our work on code-switched Egyptian Arabic–English ASR. We first contribute in filling the huge gap in resources by collecting, analyzing and publishing our spontaneous CS Egyptian Arabic–English speech corpus. We build our ASR systems using DNN-based hybrid and Transformer-based end-to-end models. In this paper, we present a thorough comparison between both approaches under the setting of a low-resource, orthographically unstandardized, and morphologically rich language pair. We show that while both systems achieve comparable overall recognition results, the systems have complementary strengths. We show that recognition can be improved by combining the outputs of the two systems. We propose several effective system combination approaches, where hypotheses of both systems are merged on sentence- and word-levels. Our approaches result in overall WER relative improvement of 4.7%, over a baseline performance of 32.1% WER. In the case of intra-sentential CS sentences, we achieve WER relative improvement of 4.8%. Our best performing system achieves 30.6% WER on ArzEn test set.