Audio Related Quality of Experience Evaluation in Urban Transportation Environments With Brain Inspired Graph Learning

The fast advancement of urban transportation systems in the recent decades has on one hand improved efficiency in traffic control and management, yet on the other hand brought new obstacles and interferences in audio related services in transportation systems, which is one of the dominating components in urban transportation systems, such as end-to-end Voice over Internet Protocol (VoIP) communications, risk alerting, and personalised recommendation services. The movement of vehicles/trains and the growing complexity of transportation infrastructures has become a big threat to the audio related services. Hence it is crucial to evaluate the Quality of Experience (QoE) of audio related services. Different from traditional algorithms which use digital signal processing to evaluate the QoE of mobile users, in this paper, we propose a two-stage brain-alike neural network aided graph learning algorithm to evaluate the QoE of audio signals with the aid of EEG feature extraction. The results are evaluated by newly-collected on-site data in public transportation environments and are examined by a branch of human experts to show that our algorithm outperforms other benchmark algorithms in term of human perception and accuracy of classification.