Quantitative analysis of professionally trained versus untrained voices

Transcription

1 36 Quantitative analysis of professionally trained versus untrained voices Clinic of Othorhinolaryngology Kaunas University of Medicine Hospital, Lithuania Keywords: quantitative voice assessment, voice range profile, speaking voice, coefficient of sound, voice training. Summary. The aim of this study was to compare healthy trained and untrained voices as well as healthy and dysphonic trained voices in adults using combined voice range profile and aerodynamic tests, to define the normal range limiting values of quantitative voice parameters and to select the most informative quantitative voice parameters for separation between healthy and dysphonic trained voices. Three groups of persons were evaluated. One hundred eighty six healthy volunteers were divided into two groups according to voice training: non-professional speakers group consisted of 106 untrained voices persons (36 males and 70 females) and professional speakers group of 80 trained voices persons (21 males and 59 females). Clinical group consisted of 103 dysphonic professional speakers (23 males and 80 females) with various voice disorders. Eighteen quantitative voice parameters from combined voice range profile (VRP) test were analyzed: 8 of voice range profile, 8 of speaking voice, overall vocal dysfunction degree and coefficient of sound, and aerodynamic maximum phonation time. Analysis showed that healthy professional speakers demonstrated expanded vocal abilities in comparison to healthy non-professional speakers. Quantitative voice range profile parameterspitch range, high frequency limit, area of high frequencies and coefficient of sound differed significantly between healthy professional and non-professional voices, and were more informative than speaking voice or aerodynamic parameters in showing the voice training. Logistic stepwise regression revealed that VRP area in high frequencies was sufficient to discriminate between healthy and dysphonic professional speakers for male subjects (overall discrimination accuracy 81.8%) and combination of three quantitative parameters (VRP high frequency limit, maximum voice intensity and slope of speaking curve) for female subjects (overall model discrimination accuracy 75.4%). We concluded that quantitative voice assessment with selected parameters might be useful for evaluation of voice education for healthy professional speakers as well as for detection of vocal dysfunction and evaluation of rehabilitation effect in dysphonic professionals. Introduction Different procedures have been developed for the assessment of voice performance in phoniatrics diagnostics. Besides the estimation of spectral properties, the examination of vocal abilities using VRP seems to be helpful for evaluation of voice education (1, 2). N. Siupsinskiene and V. Uloza (1) compared trained and untrained children voices and concluded that VRP parameters seemed to be more sensitive for the assessment of voice training than acoustic perturbation measurements. This was in agreement with other studies (3, 4). The combined (spectral) VRP both measured speaker s ability to produce maximum and minimum vocal intensities at his or her maximum frequency range and sound pressure level (SPL) in the singers formant range ( khz) at maximum SPL gave additional, important information with respect to voice quality and detection of trained voices (2, 4 6). Quantitative voice assessment is important for nowadays phoniatrics, especially in clinical practice to test voice possibilities, to quantify the degree of dysphonia, and to assess the results of treatment or vocal training (7 9). The importance of quantitative voice assessment is predetermined by the following tasks: simple, reliable and inexpensive registration of voice production, determination of voice quality standard, uniform interpretation of the collected data, and Correspondence to N. Ðiupðinskienë, Clinic of Othorhinolaryngology, Kaunas University of Medicine Hospital, Eiveniø 2, 3007 Kaunas, Lithuania. Norai_S@hotmail.com

2 Quantitative analysis of professionally trained versus untrained voices 37 clinical application. In 2000, the Committee of Phoniatrics of the European Laryngology Society (ELS) proposed a basic multidimensional (five-dimension) protocol of quantitative voice assessment for functional results of voice therapy (10). First clinical implementation showed large variations in the interindividual and interdimensional results of the voice therapy that indicated low redundancy of five considered dimensions. Another approach to quantify the voice quality is to compose the combined scores and indexes, by grouping data according parameter validity. L. Heylen et al. (1998), J. F. Piccirilo et al. (1998), F. L. Wuyts et al. (2000), and others have proposed multidimensional indexes for adult and children voices classification and evaluation of treatment effectiveness (7 9, 11). However, all these studies were completed only with non-professional speakers. It is known that voice training expands voice abilities as well as stability of acoustic parameters (12, 13). Consequently, the normal range limits of healthy trained and untrained voices parameters may be different as well as other relevant parameters could discriminate healthy from pathological trained voices. The aim of this study was: 1) to assess and to compare healthy trained and untrained voices as well as healthy and dysphonic trained voices in adults using combined voice range profile and aerodynamic tests; 2) to define the normal range limiting values of quantitative voice parameters; 3) to select the most informative quantitative voice parameters for separation between healthy and dysphonic trained voices. Material and methods One hundred eighty six healthy volunteers and 103 dysphonic professional speakers evaluated in Kaunas University of Medicine Hospital in were selected. Healthy voice was established after phoniatric examination: no organic pathology on the vocal cods, and no history of voice surgery. Professional voice users were defined as persons who received practical voice education (not less 2 hrs/week) for about 2 years (14). Only subjects after mutation were enrolled. According to the voice training, healthy voice subjects were divided into two groups. Non-professional speakers group (healthy non-professionals) was composed of 106 persons, aged years (median 29.5 yrs); 36 (33.9 %) males and 70 (66.1%) females. Professional speakers group (healthy professionals) was composed of 80 persons aged years (median 22.0 yrs); 21 (25.9%) males and 59 (74.1%) females selected of professional choirs and conservatoire students. Clinical group consisted of 103 dysphonic professionals aged years (median 27.0 yrs); 23 (22.3%) males and 80 (77.7%) females with various voice disorders. Tested groups were matched with regard to age and gender proportions, smoking habits; professional speakers as well to voice classes. Voice assessments Voice range profile was registered in an ordinary 5 3 m room (the noise level did not exceed 40 db (A)), in a classical way (by hand), according to the recommendations of the Union of European Phoniatricians (15). The pitch range was measured with the help of electronic keyboard (Fujiyama 3A) in a range of four octaves in a manner of half tone step. Sound pressure level (SPL) was determined from a sound pressure level meter (VEB Robotron), using the slow meter damping and an A-weighted frequency curve (db (A)). Only precisely corresponding sounds, sustained for at least 2 sec. were registered. Simultaneously from the SPL measurement of loudest phonation, the sound level in the frequency region of the singer s formant was estimated using a khz band pass filter (roll off 18 db/octave). A microphone (MV 102, Messelektronik Dresden) was used at a constant distance of 30 cm from the mouth. Speaking voice was evaluated in the same conditions in a counting (from 1 to 30) manner. We used modified by squares registration form (1). The parameters assessed include the following: Voice range profile parameters: 1. Pitch range (PR) the distance from the lowest to the highest singing tone, measured in semitones (st), 2. Low frequency limit (LFL) the lowest tone sung, expressed as the numeral classification of semitones from W. Vennard (st No.) (16), 3. High frequency limit (HFL) the highest tone sung (st Nr.), 4. Maximum-minimum intensity range (max.-min. IR) the distance between the softest and the loudest SPL registered by meter, measured in db (A), 5. Maximum voice intensity (max. VI) the highest voice intensity, registered by the meter, db (A), 6. Minimum voice intensity (min. VI) the lowest voice intensity, registered by the meter, db (A), 7. Total area in squares (A Total ) area calculated by square method, measured in cm 2, 8. Area in range of high frequency (A High ) part of the total area, calculated from Hz for male and Hz for female (mean register change place), measured in cm 2. Overall vocal dysfunction degree (VDD) was cal-

3 38 culated using original algorithm from categorized 4 VRP parameters: PR, max.-min. IR, A Total, A High. Four VDD damage degree from 0 (normal) to 3 (severe) were stated (11). Speaking voice parameters: 1. Fundamental frequency ( ) habitual pitch of the speaking voice, measured in hertz (Hz), 2. Habitual intensity ( ) habitual intensity of the speaking voice, measured in db (A), 3. Loudest speak tone (-max. ) shouting frequency, measured in Hz, 4. Loudest intensity (SI max. ) shouting intensity, measured in db(a), 5. Speak tone range ( -max. ) distance between habitual and shouting frequency, measured in semitones (st), 6. Intensity range ( SI max. ) distance between habitual and shouting intensity, measured in db(a), 7. Location of within VRP pitch range ( /PR), measured in percent, 8. Slope of speaking curve (slope-sc), calculated for 10 db/st. Coefficient of sound (CS), i. e. the quotient of the singing formant level (SFL) and the sound pressure level maximum, expressed in percent (5). It was measured from A to a 1 notes for males and from a to a 2 notes for females. Average of all low, medium and high values was calculated in proposed fashion: SFL CS= 100%. max.vi This method was used for 44 healthy non-professionals (11 male and 33 female), 12 healthy professionals (3 male, 9 female), and 25 dysphonic professionals (8 male and 17 female). Aerodynamic parameter maximum phonation time (MPT) was measured in seconds (sec.) with a stopwatch on the basis of one to three test trials with the vowel /a/, sustained at the subject s habitual pitch and loudness in free field and standing position. Statistical analysis was performed with SPSS 10 for Windows. The Kolmogorov Smirnov test was applied to test the normality of the distribution of quantitative data. ANOVA model with multiple comparison and LSD post hoc test were used in comparison of differences between groups for normally distributed quantitative parameters; non-parametric tests Kruskal Wallis, Mann Whitney U for not normally distributed parameters (LFL, HFL,, -max. ). χ 2 test was used to compare proportions. α level of significance of 0.05 was used. Mean, 5% trimmed mean (for not normally distributed quantitative parameters), difference of mean and 95% confidence intervals of difference were used for description. The effects of age and sex on quantitative parameters for healthy subjects were investigated with the ANOVA model. Normal range of reference intervals (RI) was calculated first, finding limiting values of RI as X ± 2 SD (SD standard deviation) for normally distributed and 2.5 and 97.5 percentiles for not normally distributed parameters, later on, finding the 95% confidence intervals (CI) for lower and upper limits of RI. The 95% CI for 2.5 and 97.5 quantiles, estimated directly from the data by the binomial distribution method (17). Based on threshold value (VDD) data for normal and pathological voices were dichotomized and the sensitivity (the proportion of dysphonic cases correctly identified) and specificity (the proportion of normal cases correctly identified) were calculated. Later on, binary logistic regression was used for final separation of healthy and dysphonic trained voices. A stepwise selection procedure was applied. With classification tables, the sensitivity (the proportion of dysphonic cases correctly identified by the test) and specificity (the proportion of normal cases correctly identified by the test) were calculated for model of selected parameters. Results Demographic and clinical data. Investigated groups: healthy non-professionals, healthy professionals and dysphonic professionals were close by age and gender proportions (p>0.05) (Table 1). The vast majority of tested subjects from all groups were under 45 years 87.7%, 92.5%, 89.3%, respectively (p>0.05). Most of the investigated persons were not active smokers 82.1%, 91.2%, 84.5%, respectively (p>0.05) (Table 1). Figure 1 shows distribution of laryngeal pathology in dysphonic professional speakers group for males and females. Mild organic laryngeal pathology was predominant. ANOVA analysis revealed the statistically significant effect of age only on one quantitative parameter (p=0.02); gender had significant effect on most of tested quantitative parameters. Based on these findings, all data were compared separately for male and female subjects. Normative data. Professional versus non-professional speakers. The results of this study are summarized in Tables 2 and 3. According to the study data, the most significant differences between healthy nonprofessionals and professionals were shown in VRP parameters for both male and female subjects. There

4 Quantitative analysis of professionally trained versus untrained voices 39 Table 1. Baseline characteristics of investigated groups Healthy non- Healthy Dysphonic Characteristics professionals professionals professionals p-value (n=106) (n=80) (n=103) Age groups proportions in % 87.7/7.5/ /5.0/ /7.8/2.9 ns (16 45/46 60/ 61yrs) Female/male (% (n)) 66.0(70)/34.0(36) 73.8(59)/26.3(21) 77.2(80)/22.8(23) ns Smokers (% (n)) 17.9(19) 8.8(7) 15.5(16) ns Voice training year (median) ns Voice classes: 16.3/40.0/ /47.6/25.2 ns (1 bass or altos; 2 baritone or mezzo; 3 tenors or soprano (%) ns not significant. Laryngeal pathology of professional speakers in males 13% 9% 4% Laryngitis l posterior VC noduli VC polypus and cystis Functional f dysphonia 74% Laryngeal pathology of professional speakers in females 36% 31% Laryngitis l posterior VC noduli VC polypus and cystis Functional f dysphonia 4% 29% Fig. 1. Distribution of laryngeal pathology in dysphonic professional speakers group; VC vocal cords

5 40 Table 2. Normative data of quantitative voice parameters for healthy non-professional and professional male speakers Healthy non-professionals (n=36) Healthy professionals (n=21) Parameters Lower Lower Lower Lower p- X Upper Upper X Upper Upper value 95%CI 95%CI 95%CI 95%CI of X-2SD of X+2SD of X-2SD of X+2SD Voice range profile PR st LFL st No. a ns HFL st No. a Min.VI db(a) ns Max.VI db(a) ns Max. min.ir db(a) ns A Total cm ns A High cm Speaking voice Hz a db(a) ns max. Hz a ns SImax. db(a) ns max. st ns SI max. db(a) ns /PR % ns Slope SC 10dB/st ns MPT sec X mean; a 5% trimmed mean; SD standard deviation; 95%CI of X±2SD the 95% confidence intervals of normally distributed parameters reference interval limits (normal range) (2.5% 97.5% percentiles of reference interval were calculated for not normally distributed parameters); ns not significant; significance level p<0.05. were more significant differences in VRP parameters for female subjects. Data of almost all tested VRP parameters except one low frequency limit, differed significantly in comparing professional versus nonprofessional female speakers groups. Both professional male and female speakers in comparison to nonprofessionals speakers demonstrated larger mean pitch range 37.7± (standard deviation) 3.9 st vs. 34.1±3.2 st, males; 34.4±2.9 st vs. 29.6±3.3 st females; higher high frequency limit 65.0± 3.1 st No. vs. 61.7±3.1 st No., males; 70.9±1.7 st No. vs. 66.9±2.4 st No., females; and larger area of high frequencies 8.5±3.2 cm 2 vs. 6.3 ±2.9 cm 2, males; 8.2±2.1 cm 2 vs. 4.8±1.7 cm 2, females (p<0.05). There were no significant differences in VRP intensity parameters for males in opposite to female speakers. Specificity of VDD (VDD=0) for male healthy professionals and non-professionals were 100% and 100%, for females 100% and 98.6%, respectively. Specificity (damage degree=0) of each of four estimated categorized VDD system parameters for male and female healthy professional speakers and nonprofessionals speakers were as follows: PR 95.2% and 88.9%, males; 100% and 87.1%, females; max.- min. IR 95.2% and 88.9%, males; 98.3% and 95.7%, females; A Total 100% and 100%, males; 100% and 98.6%, females; A High 100% and 86,1%, males; 100% and 95.7%, females. Analysis of speaking voice revealed significant difference between healthy professional and non-professional speakers groups only for 1 of 8 assessed parameters for male: speaking fundamental frequency ( ), and 2 of 8 parameters for female subjects: and slope of speaking curve. Speaking fundamental frequency was higher in professional speakers groups for both males and females in comparison to non-pro-

6 Quantitative analysis of professionally trained versus untrained voices 41 Table 3. Normative data of quantitative voice parameters for healthy non-professional and professional female speakers Healthy non-professionals (n=70) Healthy professionals (n=59) Parameters Lower Lower Lower Lower p- X Upper Upper X Upper Upper value 95%CI 95%CI 95%CI 95%CI of X-2SD of X+2SD of X-2SD of X+2SD Voice range profile PR st LFL st No. a ns HFL st No. a Min.VI db(a) Max.VI db(a) Max. min.ir db(a) A Total cm A High cm Speaking voice Hz a db(a) ns max. Hz a ns SImax. db(a) ns max. st ns SI max. db(a) ns /PR % ns Slope SC 10dB/st MPT sec ns X mean; a 5% trimmed mean; SD standard deviation; 95%CI of X±2SD the 95% confidence intervals of normally distributed parameters reference interval limits (normal range) (2.5% 97.5% percentiles of reference interval were calculated for not normally distributed parameters); ns not significant; significance level p<0.05. fessionals speakers groups (p<0.05); mean slope of speaking curve for 10db/st. was steeper in professional speakers, significantly for females (p<0.05). Coefficient of sound also showed significant difference between professional and non-professional speakers for both male and female subjects. Mean values of this parameter was statistically significantly higher for professional speakers: 91.8±1.1% vs. 76.6±10.6%, males; 90.9±4.7% vs. 77.7± 4.9%, females (p<0.05). Means of aerodynamic parameter MPT differed significantly only for male subjects. Professional healthy versus professional dysphonic speakers. Summary of data comparison between professional healthy and professional dysphonic speakers voices is shown in Table 4. For male subjects only 4 of 8 VRP parameters: pitch range, high frequency limit, total area and area of high frequencies showed significant difference between groups mean values of these parameters were significantly lower in dysphonic professional speakers group (p<0.05). Female subjects demonstrated significant differences in more parameters: seven of eight VRP parameters (except low frequency limit), three of 8 speaking voice parameters (speaking fundamental frequency, maximum speaking intensity, slope of speaking curve) and MPT. Means of VRP parameters pitch range, high frequency limit, VRP areas, maximum voice intensity and intensity range, as well as speaking voice parameter maximum speaking intensity, and MPT were significantly lower in dysphonic professional speakers group compared to healthy professional females; then, minimum voice intensity, fundamental frequency higher, slope of speaking curve longer and more flat (p<0.05). Sensitivity and specificity of VDD was 13.0% (sensitivity) for dysphonic professional male speakers, 16.3% for females, and 100% (specific-

7 42 Table 4. Results of quantitative voice parameters for healthy and dysphonic professional speakers groups Male Female Parameters Dysphonic Healthy p- Dysphonic Healthy p- professionals professionals value professionals professionals value (n=23) (n=21) (n=80) (n=59) X±SD Me X±SD Me X±SD Me X±SD Me Voice range profile PR st 34.1± ± ± ± LFL st No. a 27.3± ± ns 38.0± ± ns HFL st No. a 60.4± ± ± ± Min.VI db(a) 51.3± ± ns 50.3± ± Max.VI B(A) 96.0± ± ns 95.1± ± Max. min.ir db(a) 44.7± ± ns 45.6± ± A Total cm ± ± ± ± A High cm 2 4.6± ± ± ± Speaking voice Hz 117.6± ± ns 216.1± ± db(a) 61.5± ± ns 60.6± ± ns -max. Hz a 261.9± ± ns 404.4± ± ns SImax. db(a) 91.8± ± ns 88.9± ± max. st 13.9± ± ns 10.7± ± ns SI max. db(a) 30.3± ± ns 28.4± ± ns /PR % 18.9± ± ns 20.9± ± ns Slope SC 10dB/st 4.7± ± ns 3.9± ± MPT sec. 21.7± ± ns 18.1± ± X mean; a 5% trimmed mean; Me median; SD standard deviation; ns not significant; significance level p<0.05. ity) for healthy professional male and female speakers. The same tendency was stated, in assessment separate VDD system parameters: sensitivity for professionals ranged from 8.7% (max. min. IR) to 26.1% (A High ), specificity from 95.2% to 100%. Coefficient of sound did not differ significantly between healthy and dysphonic professionals for both males and females (Fig 2). Mean of coefficient of sound was 91.8±1.1% for healthy and 84.9±7.05% for dysphonic professionally voice trained males as well as 90.9±4.7% for healthy and 89.7± 6.5% for dysphonic professionally voice trained females. Means of MPT showed significant difference between groups only for female subjects: 21.3±6.2 sec. vs. 18.1±6.1 sec. (p<0.05). Binary logistic regression using stepwise variable selection revealed that one VRP parameter area in high frequencies was sufficient to discriminate between healthy and dysphonic professional speakers for male subjects; combination of three quantitative parameters: VRP high frequency limit, maximum voice intensity and speaking voice parameter slope of speaking curve for 10db/st were more sensitive in female subjects. Model classification sensitivity for males was 82.6%, specificity 81.0%, overall accuracy 81.8%, for females 77.2%, 72.9%, and 75.4%, respectively (cut value of probability=0.5) (Table 5). Models are in good fitness (chi-square of omnibus test = 0.000). The predicted probability for voice damage (VD) in professional male speakers could be defined with the equation: P VD. =1/1+e In professional female s speakers: e P VD. =1/1+e ( AHigh ) ( HFL max. VI slope SC ) Discussion To quantify the voice quality and to ascribe healthy voice in term of quantitative criteria is important for

8 Quantitative analysis of professionally trained versus untrained voices A B Fig 2. Diagrams of coefficient of sound A male subjects, B female subjects; 1.0=100%; A: * p<0.05 healthy non-professional speakers group vs. healthy professional speakers group; B: * p<0.05 healthy non-professional speakers group vs. healthy and dysphonic professional speakers groups. Data are expressed as minimum maximum values, 25 75% quartiles and median. both patients and otorhinolaryngologists. The first step for quantification of voice quality is to determinate normal range of voice parameters. It is important to assess normal ranges separately for healthy trained and untrained voices, because of voice training positive effect on voice possibilities. With the nowadays statistical definitions it is not enough to use mean ± 2 SD for reference intervals of estimated parameter. In this study normal range of reference intervals was defined finding 95% confidence intervals for lower and upper limits of firstly defined RI as X±2 SD for normally distributed parameters or 2.5 and 97.5 percentiles of observations values for not normally distributed parameters. This study confirmed previously raised hypothesis that healthy male subjects have larger pitch range, intensity range, maximum voice intensity, VRP areas, lower frequency limit, fundamental frequency and more flat slope of speaking curve, than female. Normative ranges of voice range profile parameters as well as speaking voice stated in

9 44 Table 5. Classification table for discrimination of healthy professional speakers versus dysphonic professional speakers groups according to selected quantitative parameters Professional speakers Predicted group membership Healthy Dysphonic Percentage correct Male Healthy professionals (n=21) Dysphonic professionals (n=23) Overall percentage 81.8 Female Healthy professionals (n=59) Dysphonic professionals (n=79) Overall percentage 75.4 Selected variables for males VRP area of high frequency; for females combination of VRP high frequency limit, maximum voice intensity and slope of speaking curve (10 db/st). The cut value of probability is 0.5 our study for trained and untrained male and female subjects are in agreement with the reported normative data in literature (9, 13, 14, 18, 19). However, MPT showed to be longer as stated in the D.S. Lundy et al. study of singing students (20). It may be reported that authors tested only singing students, who were asymptomatic, but during laryngeal examination, some VC pathology was found. Another problem of quantitative voice assessment for healthy subjects is to choose the most important methods to investigate voice training effect. Voice area and combined spectral voice area are proposed as the best methods (1 3, 5). This study supports this opinion. Our data showed that the most statistically significant differences while comparing healthy trained versus untrained voices were found for VRP parameters and coefficient of sound for both genders. VRP parameters as expressed frequency characteristics especially in region of high frequencies high frequency limit, area in high frequencies, also pitch range were the most important. Mean of coefficient of sound defined in this study was 91.8±1.1% for professionally voice trained males and 90.9±4.7% for professionally voice trained females versus 76.6±10.6% and 77.7±4.9% for untrained voice subjects (p<0.05). Similar results were reported by M. Bütner et al. (5): CS for voice professional males were %, non-professionals 78 79%; for females 85 87% and 65 75%, respectively as well as other s authors works (2, 21). Speaking voice parameters and aerodynamic MPT were not so relevant in this field. The same tendency was seen in comparison of professional healthy and dysphonic speakers. Most VRP parameters and only few speaking voice parameters showed statistically significant difference between groups. This study showed early stated multidimensional VDD witch detection of limiting values was based on data of untrained healthy voice subjects and untrained dysphonic voice patients suffering only from organic dysphonia is not relevant for separation trained healthy and dysphonic voices as well as in cases with subtle VC pathology. To select parameters, which could detect differences in mild organic and functional pathology, is very important in clinical practice. Logistic regression analysis revealed only one VRP parameter area of high frequencies to be sufficient to discriminate between healthy and dysphonic professional male speakers voices. Parameter is sensitive and specific enough (up to 80% with cut off value P=0.5). This parameter is originally calculated and cannot be compared with the other data. The value of this parameter was confirmed in prior author s investigations (1, 11). In female subjects, logistic regression method revealed three parameters VRP high frequency limit, maximum voice intensity and slope of speaking curve 10 db/st to be the most relevant in discrimination analysis (overall accuracy of discrimination 75.4%). High frequency limit as one of the most sensitive parameter for changes in voice quality was included in well-known calculation of dysphonia severity index by F. L. Wuyts et al. (2000) and voice range profile index by L. Heylen et al. (1998) (7, 9). Slope of speaking curve is in close relation with the regulation of subglottal pressure and is important for functional

10 Quantitative analysis of professionally trained versus untrained voices 45 diagnosis in professional speakers: normal and good voices raise less than suspect and pathological voices (16, 18, 19). Conclusions 1. Healthy professional speakers demonstrate expanded vocal abilities in comparison to healthy nonprofessional speakers. Quantitative voice range profile parameters pitch range, high frequency limit, area of high frequencies as well as coefficient of sound are superior to speaking voice and aerodynamic parameters to show vocal training and might be used as criteria for voice education. 2. The most sensitive parameters out of 18 in discrimination between healthy and pathological professional speakers voices selected by logistic regression analysis for males was VRP area of high frequencies (overall discrimination accuracy 81.8%), for females VRP high frequency limit, maximum voice intensity and slope of speaking curve (overall model discrimination accuracy 75.4%). 3. Quantitative voice assessment with selected parameters may be useful for evaluation of voice education for healthy professional speakers as well as for detection of vocal dysfunction and assessing rehabilitation effect in dysphonic professionals. Kiekybinë profesionaliai lavintø ir nelavintø balsø analizë Kauno medicinos universiteto klinikø Ausø, nosies, gerklës ligø klinika Raktaþodþiai: kiekybinis balso vertinimas, balso laukas, kalbos parametrai, skambëjimo koeficientas, balso lavinimas. Santrauka. Darbo tikslas nustatyti sveiko lavinto balso parametrø normà ir palyginti su sveiko nelavinto balso bei patologiðko lavinto balso asmenø duomenimis; nustatyti jautriausius parametrus, galinèius atskirti sveikà lavintà balsà nuo patologiðko. Iðtirti 186 sveiko ir 103 patologiðko balso asmenys. Sveiko lavinto balso asmenø buvo 80 (21 vyras, 59 moterys), nelavinto 106 (36 vyrai, 70 moterø). Patologiðko lavinto balso asmenø grupæ sudarë 103 ligoniai (23 vyrai, 80 moterø), turintys funkcinæ ir organinæ gerklø patologijà. Analizuoti 8 balso lauko, 8 kalbos kiekybiniai parametrai, taip pat skambëjimo koeficientas, aerodinaminis parametras maksimalus fonacijos laikas ir jungtinis parametras balso paþeidimo laipsnis. Gautais duomenimis, informatyviausi kiekybiniai balso parametrai, vertinant balso lavinimo átakà, buvo balso lauko. Sveiko lavinto balso asmenims nustatyti statistiðkai reikðmingai padidëjæ tonø diapazono, aukðtø daþniø ribos, balso lauko aukðtø daþniø ploto ir skambëjimo koeficiento vidurkiai palyginti su sveiko nelavinto balso asmenø duomenimis. Logistinë paþingsninë regresinë analizë atrinko jautriausius parametrus, atskirianèius iðlavinto sveiko balso asmenis nuo patologiðko. Tarp vyrø jautriausias parametras buvo balso lauko aukðtø daþniø plotas (bendrasis teisingai klasifikuotø asmenø procentas 81,8), tarp moterø balso lauko aukðtø daþniø riba, maksimalus balso intensyvumas ir kalbos kreivës nuolydis (bendrasis teisingai klasifikuotø asmenø procentas 75,4). Kiekybinis balso vertinimas gali bûti naudingas nustatant balso iðlavinimà, balso paþeidimà bei sekant balso reabilitacijos efektyvumà. Adresas susiraðinëjimui: N. Ðiupðinskienë, KMUK ANG ligø klinika, Eiveniø 2, 3007 Kaunas El. paðtas: Norai_S@hotmail.com References 1. Siupsinskiene N, Uloza V. Quantitative assessment of trained and untrained voices of children. Communication and its disorders: a science in progress. In: Proceedings 24 th World Congress IALP. Ph. Dejonckere, H.F.M. Peters, editors Nijmegen University Press; p Mürbe D, Sundberg J, Iwarsson J, Pabst F, Hofmann G. Longitudinal study of solo singer education effects on maximum SPL and level in the singers formant range. Log Phon Vocol 1999;24: Brown WS, Howard BR, Sapienza ChM. Perceptual and acoustic study of professionally trained versus untrained voices. J Voice 2000;14(3): Lundy DS, Roy S, Casiano RR, Xue JW, Evans J. Acoustic analysis of the singing and speaking voice in singing students. J Voice 2000;14(4): Bütner M, Seidner P, Eichhorst P. Der Klangkoeffizient ein beachtenswerter Parameter bei der Messung von Singstimmprofilen. Sprache Stimme Gehör 1991;15:135-8.

11 46 6. Barrichelo VMO, Heuer RJ, Dean CM, Sataloff RT. Comparison of singer s formant, speaking ring, and LTA spectrum among classical singers and untrained normal speakers. J Voice 2001;15(3): Haylen L, Wuyts FL, Mertens F, De Bodt M, Pattyn J, Croux C, Van de Heyning PH. Evaluation of the vocal performance of children using a voice range profile index. J Speech Lang Hear Research 1998;41: Piccirillo JF. Multivariate analysis of objective vocal function. Ann Otol Rhinol Laryngol 1998;107: Wuyts FL, De Bodt MS, Molenberghs G, Remacle M, Heylen L, Millet B, Van Lierde K, Raes J, Van de Heyning PH. The dysphonia severity index: an objective measure of vocal quality based on a multiparameter approach. J Speech Lang Hearing Research 2000;43: Dejonckere PH. Clinical implementation of a multidimensional basic protocol for assessing functional results of voice therapy. A preliminary study. Rev Laryngol Otol Rhinol 2000;121(5): Ðiupðinskienë N. Uþkimusiø pacientø, kuriems átariamas gerklø ir ryklës refliuksas, balso charakteristikø kitimas atliekant diagnostiná omeprazolio testà. (Subjective and objective voice assessment of reflux related dysphonia according empiric trial with omeprazole.) Lietuvos bendrosios praktikos gydytojas 2002;6: Doskov D, Ivanov T, Boyanov B. Comparative analysis of singer s high formant in different type of singing voices. Folia Phoniatr Logop 1995;47: Eckley CA, Sataloff RT, Hawkshaw M, Spiegel JR, Mandel S. Voice range in superior laryngeal nerve paresis and paralysis. J Voice 1998;12(3): Neuschaefer-Rube Ch, Ðram F, Klajman S. Three-dimensional phonetographic assessment of voice performance in professional and non-professional speakers. Folia Phoniatr Logop 1997;49: Pedersen MF, Frokjaer-Jensen B, Pabst F, Shutte HK, Hacki T, Hansen HL. Standardizing voice range profile measurements (Phonetography/ Stimmfeldmessung). In: Proceedings of the 23 rd World Congress IALP. M.N.Kotby, Cairo, editors; p Buekers R. Pitch adjustment in speaking levels. In: Proceedings of the 23 rd World Congress IALP. Ed. M.N.Kotby, Cairo; p Bland M. An introduction to medical statistics. 3 edition. Oxford University Press; Åkerlund L, Gramming P. Average loudness level, mean fundamental frequency, and subglottal pressure: comparison: comparison between female singers and nonsingers. J Voice 1994;8: Awan SN. Superimposition of speaking voice characteristics and phonetograms in untrained and trained vocal groups. J voice 1993;7(1): Lundy DS, Roy S, Casiano RR, Evans J, Sullivan PA, Xue JW. Relationship between aerodynamic measures of glottal efficiency and stroboscopic findings in asymptomatic students J Voice 2000;14(2): Sundberg J. Level and center frequency of the singer s formant. J Voice 2001;15: Received 2 December 2002, 13 December 2002 Straipsnis gautas , priimtas