Short Description

Currently available speech technology based systems built to support children with speech pathologies only work well if the child’s speech shows the characteristics of one clearly defined speech pathology, and if the child speaks the standard variety of a well-resourced language. In practice, however, many children have mixed-forms of speech pathologies of different degrees and they speak in their regional dialect. The aim of this thesis is to test different ASR systems on dysarthric child speech and optimize it towards one specific child, who has a mixed form of dysarthria and speaks a southern Austrian dialect. The ASR system will then be integrated into an educational tool, allowing the child to dictate stories and math calculations at high school level.

Your Task

Review of Literature

Data cleaning and preparation

Training and testing different ASR systems

Your Profile (recommended)

Automatic Speech Recognition VO

Spoken Language in Human and Computer-Human Dialogue VU or

Linguistic Foundations of Speech and Language Technology VO

Good Programming Skills (e.g., Python)