A Next Step Towards Measuring Perceived Quality of Speech Through Physiology

Abstract

For service providers, it is crucial to understand the perceived quality of transmitted speech signals, as this can influence a user's choice of operator. Various methods have been developed in the field of Quality of Experience (QoE) to evaluate perceived quality, including subjective assessments, instrumental models, and physiological measurements. In this study, participants listened to multiple speech samples while their brain activity was recorded using electroencephalography (EEG). After each stimulus, participants provided a subjective quality rating, summarized as a Mean Opinion Score (MOS). A speech quality prediction algorithm was also applied to the audio files for validation. The EEG data were then compared with the subjective ratings and demonstrated promising correlations. Unlike many prior QoE studies using EEG, this study employed a standardized testing protocol proposed by the International Telecommunication Union (ITU).

BibTeX

@inproceedings{voigtantons2014c26,
  author    = {Arndt, S. and Wenzel, M. and Antons, J.-N. and Köster, F. and Möller, S. and Curio and Gabriel},
  title     = {A Next Step Towards Measuring Perceived Quality of Speech Through Physiology},
  year      = {2014},
  booktitle = {Paper presented at the 15th Annual Conference of International Speech Communication Association (INTERSPEECH 2014), Singapore, Singapore},
  url       = {https://www.isca-archive.org/interspeech_2014/arndt14_interspeech.pdf},
}