Research project

Oral squamous cell carzinoma belongs to the 10 most frequent malignoma among humans. Their sugery and even their presence can let to a reduced intelligibility and cause speech disorders. At this state there are no existing methods that allow an objective evaluation, independent from the medical examiner, to assess the kind and seriousness of specific speech disorders. In previous works we developed an automatic speech recognition (ASR) system that allows an objective analysis of the intelligbility and the speech quality of persons with squamous cell carzinoma. The speech intelligbility was measured by the word recognition rate, i.e., the number of correctly recognized words.

The goal of this project is to improve the speech recognition system in order to use it within the clinical environment. In addition to the intelligibility estimattion of a person, it is also important to identify and assess certain articulation disorders. This allows an objective classification and quantification of speech disorders that are specific for persons with squamous cell carzinoma of the oral cavity. In a second step, a prospective data collection allows the assignment of speech recordings to the according information about specifc speech disorders and tumor staging/localization. The voice can be used to assess the speech quality of a person with a malign tumor after treatment. This assessment can be used as an objective decision criterion which can be included into the decision of a suitable treatment for future therapies.