Speech Science Forum 13th May - Prof. Mark Huckvale

Title: Automated detection of dysphonia from speech and laryngograph signals

Abstract: 

The Saarbrücken Voice Database contains speech and simultaneous electroglottography recordings of 1002 speakers exhibiting a wide range of voice disorders, together with recordings of 851 controls. Previous studies have used this database to build systems for automated detection of voice disorders, but these have varied considerably in the subset of pathologies tested, the audio materials analyzed, the cross-validation method used and the performance metric reported. This variation has made it hard to determine the most promising approaches to the problem. In this talk I will present two studies that use the SVD to build machine-learning systems for automated detection of voice disorders. The first re-implements three recently published systems that have been trained to detect pathology using the speech recordings in the SVD and compare their performance on the same pathologies with the same audio materials using a common cross-validation protocol and performance metric. We show that under this approach, there is much less difference in performance across systems than in their original publication. The second explores how the laryngograph signals in the database might be used to improve the accuracy of pathology detection. We show that the fusion of predictions made from speech and from laryngographic analysis can lead to improved accuracy.