1 paper
Lara Gauder, Pablo Riera, Andrea Slachevsky +3
We introduce a toolkit for uncovering spurious correlations between recording characteristics and target class in speech datasets. Spurious correlations may arise due to heterogene…