SLaTE 2015 - Workshop on Speech and Language Technology in Education
We apply a tool originally developed for comparing pathological and healthy speakers to non-native speech. The method works on speakers who produce a given word sequence. Using time-alignment, we can display prototypical loudness contours, local tempo variations, and also spectrograms, together with information on variability and group effect size over time. The system, which will be made publicly available, is able to expose typical differences in a group of German and Italian speakers.
Bibliographic reference. Hönig, Florian / Wankerl, Sebastian / Batliner, Anton / Nöth, Elmar (2015): "Pinpointing the difference visual comparison of non-native speaker groups", In SLaTE-2015, 187.