Devoicing in European Portuguese fricatives is an extensive phenomenon, especially when compared to other languages. Small scale acoustic studies have shown that devoicing rates and voicing profiles of fricatives are more similar to those of Germanic languages, setting European Portuguese (EP) apart within the Romance language family. The present study tests whether voicing in EP fricatives diverges from its sister languages by using empirically motivated combinations of different languages (EP, Italian, German) acoustic phone models on large EP corpora, allowing an ASR system to choose the best fitting one when force aligning the data. Results confirm that voicing in EP fricatives is more similar to German, suggesting EP voicing patterns are shifting away from classic voicing systems known for Romance languages.