Reproduce “PLCMOS -- a data-driven non-intrusive metric for the evaluation of packet loss concealment algorithms” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Lorenz Diener, Marju Purin, Sten Sootla, Ando Saabas, Robert Aichner, Ross Cutler
Researchers from Microsoft Corporation developed PLCMOS, a non-intrusive, data-driven metric designed to estimate human Mean Opinion Scores for packet loss concealment (PLC) algorithms. This metric achieves a 0.97 Pearson Correlation Coefficient with subjective ratings, significantly outperforming existing objective metrics.
Key insights
- A specialized, data-driven non-intrusive neural network can estimate human MOS for PLC evaluation with high fidelity, surpassing both general-purpose neural and classical objective metrics.
- Utilizing realistic packet loss simulations derived from real network traces is crucial for generating a robust and representative dataset for training PLC evaluation metrics.
- Explicitly modeling individual rater variability through an ID embedding module enhances the stability and overall accuracy of the subjective quality prediction.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.