Relationships between glottal flow parameters and the perception of the voice quality

• A better understanding of the aerodynamic characteristics of the voice, such as the ones represented by glottal flowwaveform parameters, has been considered valuable to better assess voice quality. However, there is still uncertainty regarding which parameters would best reflect significant differences between perceived voice qualities. Thepurpose of this study is to investigate the relationship between glottal flow parameters and the perception of thevoice quality. Thirteen vocally healthy male choir singers, with various levels of experience, skill, and degree of singing training, were asked to sing ascending and descending chromatic scales of the vowel [a], in a comfortable pitchrange and vocal loudness. They were encouraged to sing the ascending scale in modal register, extending it as far aspossible, and the descending scale in falsetto, extending it as far as possible. Analyses of 104 glottograms of vowelswith the same pitch, extracted from the pitch range where both scales overlapped, were carried out regarding thefollowing parameters: period length (T0) and fundamental frequency (F0); duration of the closed phase (Tcl) andclosed quotient (Qclosed); peak-to-peak airflow amplitude (AC amplitude); maximum flow declination rate (MFDR);amplitude quotient (i.e. the ratio between the AC amplitude and MFDR) and its corresponding normalized amplitude quotient (NAQ); level difference between the two lowest partials in the spectrum (H1 – H2). A panel of 16 voiceexperts classified the voices as being produced in modal or in falsetto register. A correlation analysis was carriedout between the listening test classification and each of the flow glottogram parameters. Tones with a high valueof Qclosed and low values of H1−H2 and of NAQ were typically associated with a high number of voices perceivedas modal voices, and vice versa. NAQ showed the strongest correlation with the perceived voice quality (r=0.830).In addition, a stepwise multiple regression analysis was carried out of the relationships between the number ofvotes for modal voices and the various flow glottogram parameters. The results showed that the Qclosed was thestrongest predictor of the number of votes for modal voices, followed by H1−H2, MFDR and AC amplitude. Finally,a cluster analysis of the flow glottogram parameters showed five clusters in a continuum along the range of valuesfor each of the parameters. Voices were unanimously associated to one or the other voice categories when theircorresponding clusters of parameters were found at the extremes of this continuum, suggesting that: 1. differentglottal flow parameters can be varied continuously and combined differently; 2. specific combinations or clusters ofaerodynamics characteristics are likely to be associated with different perceived voice qualities; 3. a perceived voicequality seems to be easier to be classified the further apart the clusters appear on the continuum, and vice-versa.

Ämnesord

TEKNIK OCH TEKNOLOGIER  -- Elektroteknik och elektronik -- Signalbehandling (hsv//swe)

ENGINEERING AND TECHNOLOGY  -- Electrical Engineering, Electronic Engineering, Information Engineering -- Signal Processing (hsv//eng)

NATURVETENSKAP  -- Data- och informationsvetenskap -- Språkteknologi (hsv//swe)

NATURAL SCIENCES  -- Computer and Information Sciences -- Language Technology (hsv//eng)

HUMANIORA  -- Konst -- Musik (hsv//swe)

HUMANITIES  -- Arts -- Music (hsv//eng)

Publikations- och innehållstyp

ref (ämneskategori)

kon (ämneskategori)