Acoustic voice analysis by means of the hoarseness diagram

The hoarseness diagram (Michaelis, Frohlich, & Strube, 1998a) has been prop osed as a new approach to describe different acoustic properties of voices. To test its performance in the analysis of pathologically disturbed and no rmal voices five requirements are suggested that should be met by any acous tic voice-analysis protocol to be used in voice research and clinical pract ice. The hoarseness diagram is then tested with regard to these requirement s. Individual voices are Found to show a satisfactory localization in the d iagram. Aspects of stationarity are discussed in the context of four case s tudies. The different cases illustrate that changes in the acoustic analysi s results are observed if the voice-generation conditions change, whereas r esults are stationary if phonation conditions do not change. Different path ological voice groups defined on grounds of the specific phonation mechanis m are found to map to specific regions of the hoarseness diagram, with diff erences between group locations being significant. All results can be inter preted without exceptions if the two hoarseness diagram coordinates are tak en to reflect the vibrational irregularity of the voice-generation mechanis ms on the one side and the degree of closure of the vibrating structures on the other side. The hoarseness diagram and its underlying algorithms are t hus shown to constitute a useful approach to acoustic voice analysis in res earch and clinical practice. The tests themselves demonstrate several appli cation possibilities, including the quantitative monitoring of individual v oices.