Short description:

Tracking of fundamental frequency from microphone recordings of pathological voices is crucial, because many diagnostic procedures rely on its correctness. Special attention must be paid if multiple distinct glottal oscillators are vibrating, which is a manifestation of hoarseness. Aim of the project is to establish a tracker that identifies the correct fundamental frequencies, and perturbations thereof. The procedure is evaluated on the reference fundamental frequencies that are obtained from laryngeal high-speed videos. Your tasks will be to mine the ground truth from the videos, to identify appropriate tracking methods from the literature, to train and to evaluate the procedures as well as to document your work. Optionally, you will produce a publication for a scientific conference and/or Journal.

Your tasks:

• Literature review
• Annotation of the ground truth
• Implementation and evaluation of fundamental frequency tracker in MATLAB
• Documentation

Your profile:

• Very good skills in signal processing and speech communication
• Very good programming skills in MATLAB
• Good English and LaTex skills

Contact:

• Philipp Aichinger (philipp.aichinger@meduniwien.ac.at or 01/40400 11670)
• Franz Pernkopf (pernkopf@tugraz.at)

Reference:

• P. Aichinger, M. Hagmüller, B. Schneider-Stickler, J. Schoentgen, and F. Pernkopf, “Tracking of Multiple Fundamental Frequencies in Diplophonic Voices,” IEEE/ACM Trans. Audio, Speech, Lang. Process., vol. 26, no. 2, pp. 330–341, 2018.