Abstract

Voice quality is a multi-dimensional concept that is closely related to vocal timbre reflecting vocal health status. Creak is a particular voice quality type frequently occurring in healthy speech, sometimes carrying linguistic information. In pathological speech, however, the use or occurrence of creak may be impeded by patients’ lack of targeted voice control. What is currently unknown is how the timing of the onsets and offsets of creaky intervals differs between healthy and pathological creaky speakers. We aim to model the occurence of creak using position- and boundary-conditional probabilities, seeking to enable discrimination of healthy and pathological speakers.

Your Tasks

• literature review
• selecting speech audio data from existing databases
• annotation of the data with respect to the presence or absence of creak, using interval tiers in Praat
• fitting of a statistical model of creak occurence
• health status discrimination
• reporting your results (thesis writing, optional: paper writing)

Your Profile

• good knowledge of sound engineering and/or speech communication
• good knowledge of conditional probabilistic modelling
• good knowledge of Python

Contact

Philipp Aichinger (philipp.aichinger@meduniwien.ac.at) Barbara Schuppler (b.schuppler@tugraz.at)

References

[1] S. Wepner, A. Kelterer, and B. Schuppler, “CREAPY : A PYTHON-BASED TOOL FOR THE DETECTION OF CREAK IN CONVERSATIONAL SPEECH,” in ICPhS, 2023, pp. 1716–1720, [Online]. Available: https://guarant.cz/icphs2023/59.pdf