Instrumental dimensioning of normal and pathological phonation using acoustic measurements

Abstract

The present study deals with the dimensions of normal and pathological phonation. Separation of normal voices from pathological voices is tested under different aspects. Using a new parametrization of voice-quality properties in the acoustic signal, the vowel productions of 534 speakers (267 M, 267 F) without any reported voice pathology and the productions of 534 gender-matched pathological speakers were considered. In a first step, a gender-specific separation of the two groups is supported by a number of significantly different parameter means. In a second step, a clustering technique differentiates three subgroups within each group and gender on the basis of the acoustic parameters. Further, a statistical examination of the correct assignment in the database (DB-classification) as "normal" or "pathological" shows that the two groups overlap to some extent. The overlap of speaker assignment indicates a phonation continuum through the multidimensional space extending from normal to pathological voices. The validity of a categorical distinction of normal and pathological phonation in the sense of an individual or group-orientated labelling of voice quality as "normal" or "pathological", respectively, is discussed.

An insight to the automatic categorization of speakers according to sex and its application to the detection of voice pathologies: A comparative study

“…It also illustrates the difficulty that the voice pathology detection problem encompasses. Indeed, the Saarbrücken database has been used for highlighting the necessity for a differentiated classification of normal and pathological phonation into additional subgroups since a strong overlapping between normal and pathological phonation is evidenced [28].…”

“…The Saarbrücken voice disorders database [28,38] holds a collection of speech registers from more than 2000 normal and pathological German speakers. It contains the recordings of the sustained phonation of vowels produced at normal, high and low pitch, as well as with rising-falling pitch.…”

“…Respecting pathological voice detection, it is evidenced in [27] that a manual segmentation of the speakers' database according to their sex improves accuracy in an automatic pathology detection system. Furthermore, authors in [28] determined a significant sex-specific separation of control and pathological classes in the Saarbrücken database, by using statistical analysis and a series of acoustical and spectral measurements. Above studies suggest that profiting from the a priori information about the sex of speakers might be helpful in normophonic vs. pathological discrimination tasks.…”

Evaluation of Phonatory Behavior of German and French Speakers in Native and Non-Native Speech

“…Therefore, we investigate French native speakers learning German and German native speakers learning French and compare their L1 and L2 productions. We chose a parametrization of phonatory behavior and phonatory quality properties in the acoustic signal based on the procedure developed by Wokurek and Pützer [24][25][26][27]. The following four research questions concerning phonatory behavior of German speakers (GS) and French speakers (FS) in native and non-native speech will be answered by the present study:…”

Stimmprofile zur Normalstimme auf der Grundlage akustischer und elektroglottografischer Analysen

“…Darüber hinaus verweisen die vorhandenen Subgruppen auf ein Phonationskontinuum innerhalb normalstimmlicher Phonation, das durch individuell variierende Stimmqualität charakterisiert wird[ 12 ] . Somit ist jede individuelle Phonationsqualität innerhalb der Parameterregionen von einer der Subgruppen zur männlichen und weiblichen Stimme zu positionieren.…”