Our findings show a way of using a simple ratio to gather language specific information. It shows that creaky tokens and pitch jumps can be useful and not errors, at least if you're measuring glottal pulses. Come speak to me in Rotterdam to learn more! 😊
/End

...This was especially fun because the two datasets are really different, with one being conversational and the other structured. We didn't do much filtering or adjustment of the data, so with even more controlled datasets we expect even better results.

...F0 Ratio was shown to be useful in language specific situations, with non-stød being identified from stød tokens successfully due to a rising intonation contour. Korean showed that female speakers consistently shift from creaky to modal voice, captured well by the ratio...

Together with @rpuggaardrode.bsky.social , Paul Boersma and Silke Haman, we looked at two languages with interesting creaky voice qualities...

Almost 20 showed up, a miracle haha

Ja is echt niet te geloven 🤦‍♀️