UPPSQA predicts pairwise speech preference by scoring each sample's MOS and applying a sigmoid to the difference, beating baselines in most but not all of 16 scenarios.
Comparison with Baselines Table 1 presents the in-domain experimental results across eight scenarios (i.e., two label conditions×four training-testing sce- narios)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SD 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Universal Preference-Score-based Pairwise Speech Quality Assessment
UPPSQA predicts pairwise speech preference by scoring each sample's MOS and applying a sigmoid to the difference, beating baselines in most but not all of 16 scenarios.