Improving Automatic Speech Recognition Performance for Low-Resource Languages With Self-Supervised Models

Speech self-supervised learning has attracted much attention due to its promising performance in multiple downstream tasks, and has become a new growth engine for speech recognition in low-resource languages. In this paper, we exploit and analyze a series of wav2vec pre-trained models for speech recognition in 15 low-resource languages in the OpenASR21 Challenge. The investigation covers two important variables during pre-training, three fine-tuning methods, as well as applications in End-to-End and hybrid systems. First, pre-trained models with different pre-training audio data and architectures (wav2vec2.0, HuBERT and WavLM) are explored for their speech recognition performance in low-resource languages. Second, we investigate data utilization, multilingual learning, and the use of a phoneme-level recognition task in fine-tuning. Furthermore, we explore what effect fine-tuning has on the similarity of representations extracted from different transformer layers. The similarity analyses cover different pre-trained architectures and fine-tuning languages. We apply pre-trained representations to End-to-End and hybrid systems to confirm our representation analyses, which have obtained better performances as well.

Low-resource languages occupy a large proportion of the languages in the world as 94% of languages are spoken by fewer than 1,000,000 people [1]. It is urgent and necessary to pay attention to the research on these languages to conserve the languages as well as the corresponding cultural heritages. Automatic Speech Recognition (ASR) in low-resource languages remains challenging. There are a series of studies focusing on the low-resource problem [2]–[7]. Compared with common languages, it is much more challenging to build an applicable ASR system for low-resource languages due to the lack of transcribed speech data, language scripts and pronunciation lexicons.