Research project

Differentiated Objective Analysis of the Speech Quality of Chronically Hoarse Patients

For differentiated diagnostics of functioning and evaluation of distorted voice (i.e. for chronic hoarseness), there are currently no validated objective approaches. Established methods evaluate voice quality only by processing sustained vowels. Important speech criteria, such as intelligibility, are not analyzed. This project served for the development of adequate methods in the context of a cross-sectional study with 87 adult patients. The first topic was modeling clinical evaluation criteria by objectively obtained parameters. The analysis is done text-based. Evaluation of speech quality is performed by automatic speech recognition, prosodic analysis, phonemic, and phonological features. The correlation between perceptual and automatic evaluation reached more than r=0.8. Additionally, mathematical visualization methods, such as the Sammon Transform, allow for intuitive graphical depiction of the individual impairment and its extent. The second topic was the quantification of hoarseness which also includes speech-related parameters for the first time. In this way, an objective clinical evaluation is created.
The automatic analysis that has been developed in this project is the basis for future telemedical applications for the control of the progress of voice rehabilitation. Furthermore, this analysis will serve as objective addition to the established subjective voice and speech evaluation in clinical practice. The approach is another important step towards evidence-based diagnostics in Phoniatrics.

Project manager:
Prof. Dr.-Ing. Michael Döllinger

Project participants:
PD Dr.-Ing. Tino Haderlein, Prof. Dr.-Ing. Elmar Nöth

Keywords:
hoarseness; pathologic voice; automatic speech processing

Duration: 15.7.2012 - 14.11.2014

Sponsored by:
Else Kröner-Fresenius-Stiftung

Contact:
Haderlein, Tino
Phone +49 9131 85 67337, Fax +49 9131 85 28781, E-Mail: studienberatung-medizintechnik@fau.de

Publications

Haderlein, Tino ; Middag, Catherine ; Martens, Jean-Pierre ; Döllinger, Michael ; Nöth, Elmar: Untersuchung von Sprachaufnahmen heiserer Stimmen mittels phonologischer und phonemischer Merkmale. In: Gross, Manfred ; Schönweiler, Rainer (Ed.) : Aktuelle phoniatrisch-pädaudiologische Aspekte 2014 (31. Wissenschaftliche Jahrestagung der Deutschen Gesellschaft für Phoniatrie und Pädaudiologie (DGPP) zusammen mit dem 5. Pädakustiker-Symposium der Akademie für Hörgeräte-Akustik Lübeck 18.-21.09.2014). 2014, pp 126-128.

Haderlein, Tino ; Middag, Catherine ; Maier, Andreas ; Martens, Jean-Pierre ; Döllinger, Michael ; Nöth, Elmar: Visualization of Intelligibility Measured by Language-Independent Features. In: Sojka, Petr ; Horák, Aleš ; Kopecek, Ivan ; Pala, Karel (Ed.) : Proc. Text, Speech and Dialogue; 17th International Conference, TSD 2014 (Text, Speech and Dialogue; 17th International Conference (TSD 2014) Brno, Czech Republic 08.-12.09.2014). Cham, Heidelberg, New York, Dordrecht, London : Springer, 2014, pp 547-554. (Lecture Notes in Artificial Intelligence Vol. 8655) - ISBN 978-3-319-10815-5
[doi>10.1007/978-3-319-10816-2]

Haderlein, Tino ; Nöth, Elmar ; Döllinger, Michael ; Schwemmle, Cornelia ; Ptok, Martin: Vergleich der subjektiv-auditiven RBH-Beurteilung mit apparativer prosodischer Analyse und Irregularitätsberechnung. In: Gross, Manfred ; Schönweiler, Rainer (Ed.) : Aktuelle phoniatrisch-pädaudiologische Aspekte 2013 (30. Wissenschaftliche Jahrestagung der Deutschen Gesellschaft für Phoniatrie und Pädaudiologie Bochum 20.-22.09.2013). 2013, pp 146-148. - ISBN 978-3-00-043305-4
[doi>10.3205/13dgpp51]

Haderlein, Tino: Automatische Bewertung von Stimmstörungen aus Textaufnahmen. In: Gross, Manfred ; Schönweiler, Rainer (Ed.) : Aktuelle phoniatrisch-pädaudiologische Aspekte 2012 (29. Wissenschaftliche Jahrestagung der Deutschen Gesellschaft für Phoniatrie und Pädaudiologie Bonn 21.-23.09.2012). 2012, pp 82-84.

Haderlein, Tino ; Nöth, Elmar ; Döllinger, Michael: Automatische Bestimmung der Verständlichkeit expressiver Sprache. In: German Society of Medical Physics (Ed.) : 43. Jahrestagung der Deutschen Gesellschaft für Medizinische Physik (43. Jahrestagung der Deutschen Gesellschaft für Medizinische Physik Jena 26.-29.09.2012). 2012, pp 147-150.

Haderlein, Tino ; Moers, Cornelia ; Möbius, Bernd ; Nöth, Elmar: Automatic Rating of Hoarseness by Text-based Cepstral and Prosodic Evaluation. In: Sojka, Petr ; Horak, Ales ; Kopecek, Ivan ; Pala, Karel (Ed.) : Proc. Text, Speech and Dialogue; 15th International Conference, TSD 2012 (Text, Speech and Dialogue; 15th International Conference (TSD 2012) Brno, Czech Republic 03.-07.09.2012). Berlin, Heidelberg : Springer-Verlag, 2012, pp 573-580. (Lecture Notes in Artificial Intelligence Vol. 7499) - ISBN 978-3-642-32789-6
[doi>10.1007/978-3-642-32790-2]