Machine Learning and Deep Learning Approaches for Accent Recognition: A Review

Accent recognition has attracted immense research interest owing to the advancements in automatic speech recognition (ASR) systems. Accent variations are an essential factor in speech, and can drastically decrease the performance of ASR systems. This study presents an extensive examination of various accent-recognition models. In this paper, we explain various preprocessing techniques, different feature extraction methods, and a detailed methodology based on the machine learning (ML) and deep learning (DL) approaches used for accent recognition. Papers were selected from Google Scholar, IEEE Xplore, ACM Digital Library, ScienceDirect, Springer, Research Gate, Scopus, and the Directory of Open Access. The search returned 103 papers, including (journals and conferences) covering the scope of this study from 2015 to 2023. Detailed discussions on the various characteristics of audio signals are presented for the development of an accent recognition system. This paper provides an overview of various DL techniques to highlight the current state-of-the-art approaches, along with the associated activation and loss functions. It also offers a concise discussion on commonly used datasets for accent recognition systems and examines the challenges posed by different auditory characteristics. Finally, we discuss the research gaps, ideas for improvement, and the current and future trends proposed by different authors.