IENT» Teaching» Final Theses» Perceived Speech Quality in Real-World Acoustic Backgrounds (Bachelor and Master)

Perceived Speech Quality in Real-World Acoustic Backgrounds (Bachelor and Master)

Speech quality is an important perceptual measure for evaluating speech communication systems, such as telecommunication devices and hearing aids. Most existing studies investigate speech quality under conventional laboratory noise conditions. Relatively few studies have explored how perceived speech quality varies across different real-world acoustic scenes, despite their practical relevance. 

This thesis aims to investigate the impact of diverse real-world acoustic scenes on perceived speech quality. Clean speech signals will be mixed with recordings from different realistic acoustic environments from the TAU 2022 dataset [1]. The speech quality will be evaluated through online listening experiments based on WebMUSHRA [2]. The work will include stimulus generation, listening test design and implementation, and analysis of the experimental results.

Students should have a basic background in signal processing and experience with programming in Matlab or Python.

Suitable for a Master theses. Contact persons: Fan Zhang M.Sc.

References

  1. Heittola, T. (Creator), Mesaros, A. (Creator), Virtanen, T. (Creator), Heittola, T. (Contributor), Laakso, H. (Contributor), Bejarano Rodriguez, R. (Contributor) (1 Jun 2022). TAU Urban Acoustic Scenes 2022 Mobile, Evaluation dataset. Zenodo. https://doi.org/10.5281/zenodo.6591203
  2. Schoeffler, M. et al., (2018). webMUSHRA — A Comprehensive Framework for Web-based Listening Tests. Journal of Open Research Software. 6(1), p.8.