Spectral Analysis of Laryngeal High-Speed Videos: Case Studies on Diplophonic and Euphonic Phonation

Laryngeal high-speed videos (LHSVs) are analysed in order to provide an objective diagnostic criterion for the detection of diplophonia. Diplophonia is a significant subset of irregular phonation and characterized by the presence of two pitches in the voice sound. In current clinical practice diplophonia is detected by auditive perceptual rating, which should be assisted by objective analysis whenever the presence of diplophonia is doubtful. Ten cases of diplophonic and ten cases of euphonic phonation are analysed by means of spectral video analysis (SVA). The Non-unimodality measure í µí±µí µí±¼í µí±ís introduced and serves as a classification feature. Estimates for the sensitivity (í µí±ºí µí±¬) and specificity (í µí±ºí µí±·) of the presented classification paradigm are í µí±ºí µí±¬: 90% (95%-í µí±ªí µí±°: [55.5%, 99.7%]) and í µí±ºí µí±·: 80% (95%−í µí±ªí µí±°: [44.4%, 97.4%]). The estimators reflect promising results. Taking into account the confidence intervals (í µí±ªí µí±°s), increasing the sample size must be considered. As a consequence from these results, LHSVs should be investigated in clinical studies more intensively in order to develop and establish solid interpretation guidelines for clinical issues.