Dysphonic Voice Database and Feature Characterization
This task is devoted to the preparation of a database including both normal and dysphonic voice recordings, and to the characterization of objective features that are required for signal classification and accurate parametric modeling.
The major outcomes in this task include a rich database combining both voiced speech and whispered speech voice recordings, as well as the identification of both established and new objective features correlating well with perceptual parameters, e.g. harshness, tension and asthenia. Due to the high importance of these results for the project, it is expected that this task will motivate one MSc dissertation in the area of Speech Therapy.
Milestones
M1_a: A database of dysphonic and nomal Portuguese phonemes and representative features.
Other expected outcomes include a task report and an MSc dissertation.
Leader: Univ. Aveiro