Dirk Michaelis

Selection and combination of acoustic features for the description of pathologic voices

The glottal to noise excitation ratio (GNE) is an acoustic measure designed to assess the amount of noise in a pulse train generated by the oscillation of the vocal folds. So far its properties have only been studied for synthesized signals, where it was found to be independent of variations of fundamental frequency (jitter) and amplitude (shimmer). On the other hand, other features designed for the same purpose like NNE (normalized noise energy) or CHNR (cepstrum based harmonics-to-noise ratio) did not show this independence. This advantage of the GNE over NNE and CHNR, as well as its general applicability in voice quality assessment, is now tested for real speech using a large group of pathologic voices (n = 447). A set of four acoustic features is extracted from a total of 22 mostly well-known acoustic voice quality measures by correlation analysis, mutual information analysis, and principal components analysis. Three of these measures are chosen to assess primarily different aspects of signal aperiodicity, while the fourth one indicates the noise content of the signal. All analysis methods lead to the same feature set that consists of a measure of period correlation, jitter, shimmer, and GNE. The two-dimensional projection of this set named hoarseness diagram allows a graphical illustration of voice quality that can be easily interpreted.

SIM—simultaneous inverse filtering and matching of a glottal flow model for acoustic speech signals

A new method simultaneous inverse filtering and model matching (SIM) is proposed that allows one to calculate voice source measures without any user interaction. It is based on the discrete all-pole modeling (DAP) technique for inverse filtering (IF), which is modified to include a model of the glottal flow as integral part [LF model, Fant et al., STL-QPSR (Stockholm) 4/1985, 1-13 (1986)]. As the correct LF parameters are initially unknown, they are estimated in an iterative procedure using multi-dimensional optimization techniques that are initialized according to the results of an exhaustive search. The error criteria applied reflect how well the IF is performed after the spectral contribution of the glottal flow has been removed. The resulting optimal LF parameter constellation serves as the basis to calculate 11 voice source measures. The performance was evaluated using synthesized signals and recordings of natural utterances. For the synthesized signals, the accuracy to reproduce the original parameters was high (correlations exceeding 0.88) for measures where the starting point of the glottal cycle did not enter explicitly. Errors were smaller compared to conventional estimation methods where the measures were estimated from the IF signal. The analysis of natural utterances indicates that problems still exist with regard to robustness, but that under advantageous conditions the open quotient, the speed quotient, the closing quotient, the parabolic spectral parameter, and the negative peak amplitude of the glottal flow derivative can indeed be determined automatically by the SIM method.

Acoustic "breathiness measures" in the description of pathologic voices

One important perceptual attribute of voice quality is breathiness. Since breathiness is generally regarded to be caused by glottal air leakage, acoustic measures related to breathiness may be used to distinguish between different physiological phonation conditions for pathological voices. Seven breathiness features described in the literature plus one self-developed measure (the glottal to noise excitation ratio, GNE) are compared for their distinguishing properties between different well-defined pathological phonation mechanisms. It is found that only GNE allows a distinction between all the pathological groups and both the normal and aphonic reference group. Furthermore, GNE is among the measures showing the most significant distinctions between the different pathologic phonation mechanism groups. Therefore GNE should be given preference over the other features in the independent assessment of glottal air leakage or breathiness for moderately or highly disturbed voices.

Objektive Beschreibung der Stimmgüte unter Verwendung des Heiserkeits-Diagramms

ZusammenfassungDie Bedeutung der objektivierbaren Stimmgütebeschreibung spiegelt sich im zunehmendem Einsatz akustischer Analsyemethoden in der klinischen Diagnostik und Forschung wider. Das Heiserkeits-Diagramm ermöglicht die akustische Stimmgütebeschreibung auch hochgradig gestörter Stimmen bis hin zur Aphonie. Die Objektivierbarkeit der Stimmqualitätsverbesserung bei postoperativer Stimmtherapie wird an 3 Fallbeispielen verdeutlicht. Des weiteren wird die Differenzierbarkeit pathophysiologisch definierter Gruppen basierend auf den Ergebnissen der akustischen Analyse untersucht. Dabei werden einerseits verschiedene Phonationsmechanismen nach Tumorresektion, andererseits verschiedene laryngeale Lähmungsformen einander gegenübergestellt. Die Ergebnisse führen zu der Interpretation, daß die Koordination des Heiserkeits-Diagramms direkt die Irregularität der Stimmlippenschwingung einerseits und die Qualität des glottalen Schlusses andererseits widerspiegeln. In diesem Zusammenhang ist auch der potentielle Nutzen des Heiserkeits-Diagramms in der klinischen Anwendung und Forschung zu sehen.SummaryThe need for an objective assessment of voice quality can be seen in the increasing use of acoustic analysis methods for clinical diagnosis and research. The hoarseness diagram allows clinicians to objectively describe even highly disturbed or aphonic voices. Its application possibilities are illustrated in three case studies in which changes in voice quality were monitored during voice rehabilitation. The distributions of voice groups that were defined on the basis of specific pathophysiological phonation conditions were then compared for their acoustic differences. The groups comprised various phonation conditions after the resections of laryngeal tumors and different types of laryngeal paralyses. Interpretation of the results suggests a direct correspondence of the hoarseness diagram coordinates and the irregularity of vibration on the one side and the degree of glottal closure on the other. This illustrates the potential usefulness of the hoarseness diagram in a clinical context.

Akustische Stimmanalysen zur Dokumentation der Stimmrehabilitation nach laserchirurgischer Larynxkarzinomresektion

ZusammenfassungPatienten nach kurativer minimalinvasiver Laserresektion von Larynxkarzinomen (T1–T3) wurden im Rahmen eines von der Deutschen Krebshilfe geförderten Projekts einer intensiven konservativen Stimmrehabilitation unterzogen. Die Therapie erfolgte nach dem Konzept der funktionalen Stimmtherapie für die Dauer von durchschnittlich 2 Monaten. Prä- und postrehabilitativ wurden die Stimmen der Patienten mit verschiedenen computergestützten Meßsystemen analysiert: 1. dem „multidimensional voice program“ (Fa. Kay) und 2. einem eigenen Softwareprogramm, in dem ein neuer Stimmgüteparameter integriert ist, der eine Jitter- und Shimmer-unabhängige Beschreibung des Geräuschanteils gestattet. Die Ergebnisse zeigten: 1. Die glottische Ersatzphonation erreichte in der Regel die besseren Analyseergebnisse als die supraglottische Ersatzphonation. 2. Nicht alle akustischen Parameter dokumentierten gleichermaßen die Stimmverbesserung nach Stimmrehabilitation. Die Standardabweichung der Grundfrequenz zeigte als einziger Parameter eine signifikante posttherapeutische Verbesserung an. Als für die Stimmanalyse geeignet erwies sich weiterhin der von uns neu eingeführte Stimmgüteparameter. Dieser ist im Unterschied zu analogen Meßgrößen anderer Verfahren unabhängig von der exakten Periodizität der Glottisanregung, und ermöglichte dadurch auch bei stark dysphonen oder aphonen Stimmen noch zuverlässige Aussagen.SummaryFollowing curative minimal-invasive laser resection of T1–T3 laryngeal carcinomas, patients were subjected to an intensive voice rehabilitation. Therapy was effected twice a day for approximately 2 months, utilizing the concept of functional voice therapy. Before and after rehabilitation, acoustic analyses were made by using two different computer-supported measuring systems: (1) the “multidimensional voice program” (Kay Elementrics Corp.) and (2) a novel software program developed in Göttingen that includes a new voice quality parameter based on correlations between frequency bands. Acoustic analyses showed superiority of the glottal versus supraglottal compensatory phonation. Findings showed that not all acoustic parameters equally documented voice improvement after rehabilitation. The standard deviation of fundamental frequency was the only parameter showing a significant post-therapeutic improvement. A further suitable acoustic method proved to be the voice quality parameter that has been newly introduced by us. In contrast to analogous parameters of other methods, this approach is independent of the exact periodicity of the glottal excitation function, thus permitting reliable results to be obtained even with aphonic or heavily dysphonic voices.

Stimmfunktionelle Qualitätssicherung in der kurativen Mikrochirurgie der Larynxmalignome : Postoperative Stimmrehabilitation auf Basis der laryngealen Doppelventilfunktion

ZusammenfassungNach Negus bzw. Pressman bilden die Sphinktersysteme der Stimmlippen und Taschenfalten eine „laryngeale Doppelventilfunktion”. Entsprechend finden wir ein physiologisches Phonationssystem der Glottis und ein pathologisch-kompensatorisches der Supraglottis. Deren Steuerung scheint über einen Regelkreis zu erfolgen mit offenbar der glottischen Phonationsfunktion als Meßfühler. Zur Bestätigung dieser Hypothese werden objektive Stimmanalysen mit glottisrelevanten Parametern von 26 stimmrehabilierten Patienten nach minimal-invasiver Laserchirurgie glottischer Karzinome demonstriert und in ein „Heiserkeits-Diagramm” mit den Koordinaten Rauhigkeit und Behauchung integriert. Unsere Daten belegen mit statistisch abgrenzbaren akustischen Clustern eine qualitative Hierarchie differenter postoperativer Phonationsmechanismen. Sie zeigen den Einfluß der Schwingungsfähigkeit glottischer bzw. supraglottischer Strukturen auf die Qualität des vibratorischen Schlusses. Beide Funktionsparameter bestimmen offensichtlich die resultierende Stimmqualität im Sinne unserer Hypothese.SummaryAccording to Negus and Pressman the sphincter systems of the vocal folds and the ventricular folds form a respiratory ”laryngeal double valve function”. Correspondingly, we found a physiological phonation system of the glottis and a pathological-compensatory one of the supraglottis. They appear to be regulated through an automatic phonatory control system with the glottal phonatory function evidently acting as sensor level. In order to confirm this hypothesis, objective voice analyses with glottal-relevant parameters of 26 voice-rehabilitated patients after minimally invasive laser surgery of glottal carcinomas are presented and integrated into a ”hoarseness diagram” with the coordinates roughness and breathiness. Using statistically deliminated acoustic clusters, our data show a qualitative hierarchy of different postoperative phonation mechanisms. They demonstrate the influence of the vibratory capacity of glottal and supraglottal structures on the quality of the vibratory closure. Both functional parameters evidently determine the resulting voice quality in the sense of our hypothesis.