ABSTRACT
Acoustic Scene Classification (ASC) is an area of growing relevance,
with applications ranging from assistive devices, such as hearing
aids, to advanced wearable technologies (hearables). This paper
presents a Systematic Literature Review (SLR) that analyzes the
main adaptive and machine learning-based methods used in ASC,
with a focus on hearing devices. The challenges related to computational
resource limitations, energy consumption and real-time
operation, especially in dynamic environments, are discussed. The
review highlights recent advances, such as the use of generative
probabilistic models and convolutional neural networks, as well as
hybrid approaches that combine cloud computing and edge computing
for greater efficiency. The results show that, despite significant
progress, there are still important technical barriers, such as the
need for more efficient, customizable and robust algorithms to operate
in real conditions. This study contributes by identifying gaps
in the literature and suggesting future directions to improve the
integration of ASC in hearing devices.