Enhancing Arabic Audio Quality through Neural Network-Based Upscaling
Mohammed Alsuhaibani · 2024
The ubiquity of data and its capture methods play a pivotal role in information communication. With the advent of online platforms, organizations extensively use compression techniques to mitigate costs and reduce network traffic, often at the expense of audio quality. This paper addresses this challenge by utilising an approach to audio data upscaling through neural network. The method specifically targets the generation of missing high-frequency components, facilitating speech super-resolution and bandwidth expansion. A key distinction of this work is its focus on Arabic audio content, a domain that has seen limited exploration in the context of neural network-based audio enhancement. Along with the development and introduction of an Arabic audio dataset curated for the purposes of this study. Experimental results validate the efficacy of the proposed approach, demonstrating marked improvements in audio quality.