Changes in voice during a day in normal voices withoutvocal loadingMarjanna Artkoski, Johanna Tommila and Anne-Maria Laukkanen

From the Department of Speech Communication and Voice Research, University of Tampere, Tampere, Finland

Log Phon Vocol 2002; 27: 118–123

Various changes in vocal parameters have been observed during the working days of voice professionals, e.g. teachers. Thepresent study investigated vocal parameters during daytime when voice was not used. Eleven females and ten males (meanages 30 and 40 years) volunteered as subjects. They recorded the same reading task in the forenoon and in the afternoonwith an interval of 6–8 hours. Between the recordings voice use was avoided. Six females repeated the test in six days withthe intervals from one week to three months. Fundamental frequency (F0) and sound pressure level (SPL) were measured.Voice quality was studied by calculating the alpha ratio (SPL 1000–5000 Hz:SPL 50–1000 Hz). In most cases F0 was higherin the afternoon, although the change was on the average small (in most cases within 5 Hz) and statistically non-signi� cant.Alpha ratio was signi� cantly higher for the females and lower for the males in the afternoon. Similar changes have beenreported after vocal loading. The results suggest that studies on the effects of vocal loading should also address changes invocal parameters when no vocal loading is present. Gender difference in the vocal changes warrants further study.

Key words: diurnal rhythm, psycho-physiological activity, F0, SPL, alpha ratio, voice quality.

Anne -Maria Laukkanen, Department of Speech Communication and Voice Research, University of Tampere, FIN-33014Finland. Tel. »358-3-2157249. Fax »358-3-2156063. E-Mail: anne -maria.laukkanen@uta.�

ORIGINAL ARTICLE

INTRODUCTION

Various investigations have reported a rise in funda-mental frequency (F0) and sound pressure level (SPL)after prolonged voice use, i.e. vocal loading (3, 9, 12,20–22, 25, 28). Typically, in vocally loading tasks themechanical stress imposed on the vocal organ hasbeen increased by requiring not only prolonged voiceuse but also louder and more high-pitched phonationthan that used in normal conversation. After a vo-cally loading task the voice source or the spectrum ofthe voice has been found to change in a way thatsuggests a move towards a more hyperfunctionalvoice production in females (16, 18, 21, 28). In males,a similar but less systematic change has been ob-served after prolonged loud reading (16, 28). How-ever, an opposite change towards a morehypofunctional voice production has also been re-ported in male actors after a vocally loading theatreperformance (18). Apparently, vocal loading affectsvocal parameters. On the other hand, there may beother reasons for changes in vocal parameters duringa long-term loading task—like a teacher’s wholeworking day.

The behavior of a human being is known to beaffected by various biological rhythms (4). Theserhythms regulate e.g. the psycho-physiologica l state.

Rhythms occurring at intervals of approximately 24hours are called circadian rhythms (e.g. sleep-awakerhythm, changes in body temperature and heartbeat).It can be supposed that biological rhythms wouldalso cause changes in vocal parameters. Bouhuys etal. (2) investigated the relations between mood andvocal parameters in normal conditions and in thecontext of sleep deprivation. According to their re-sults changes in F0 followed a clear circadian pattern:F0 tended to rise from 7 a.m. up to 1 p.m., where-after it decreased until about 4 p.m. After that itstarted to rise again till 7 p.m., while it remainedsteady between 7 and 10 p.m. Furthermore, therewere differences in the timing of changes in the F0pattern depending on the sleep rhythm, which wasdisturbed in the test by total sleep deprivation. Thecircadian pattern in F0 bears a resemblance to theresults on circadian changes in body temperature andin the reaction time (4) for example. Differences inthese rhythms between personality types (introvertsvs. extroverts) have also been reported (4).

Cooper and Yanagihara (5) found three differentpatterns of F0 change in the basal pitch (lowestcomfortable sustainable pitch) of their adult subjectsduring a day: F0 either rose or fell throughout theday or rose till noon and fell towards late afternoon.

© 2002 Taylor & Francis. ISSN 1401-5439 Log Phon Vocol 27

Log

oped

Pho

niat

r V

ocol

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f M

elbo

urne

on

11/0

6/14

For

pers

onal

use

onl

y.

Changes in voice during a day 119

According to Garrett and Healey (8) mean F0 inspeech rose steadily for males throughout the day,while different patterns were found for females.

The present study investigated changes in F0, SPLand acoustic voice quality (frequency distribution ofsound energy) in the speech of adult subjects beforeand after a normal 6–8-hour working day when allkinds of vocal and physical loading was avoided. Thefollowing research questions were posed: (1) Do theacoustic parameters change signi� cantly during a dayeven when the voice is not used? (2) Are there differ-ences in these changes between the genders?

MATERIAL AND METHODS

Subjects and tasks

Eleven females (age 23–56 years, mean 30 years) andten males (age 30–55 years, mean 40 years) volun-teered as subjects. The subjects had no known pathol-ogy of voice or hearing. One of the females and oneof the males was a smoker. All of the females andtwo of the males had some vocal education. Since thestudy aimed to address pure physiological changes invoice, the absence or presence of vocal education wasnot regarded as essential. Many of the subjects had aprofession that does not require a lot of voice use.Those who were voice professionals (teachers) wereable to rest their voices on the test days.

The same text (133 words, no:s:-sounds included)was read aloud once or twice in the morning (before10 a.m.) and once or twice in the afternoon (between2 and 5 p.m). A text with no:s:-sounds was used inorder to be able to draw conclusions from changes invoice quality on the basis of alpha ratio. In thepresence of fricatives with a strong noise at a widefrequency band this would not be possible. Subjectswere instructed to read the text neutrally (avoidingthe expression of strong emotions). Two males read atext twice in succession both in the morning and inthe afternoon and took the test twice, at intervals ofthree months. Six of the female subjects read the texttwice in succession and repeated the test six times onsix different days. Typically, the two � rst recordings

were made on consecutive days and the third 1–4weeks later. In total, 77 pairs of reading samples wereobtained from the female subjects and 14 from themales. The subjects were instructed to keep silent asmuch as possible and also to avoid physical loadingon the test day. If a subject was not able to meetthese demands on a particular test day he:she wasrecorded again on another day. The aim of includingboth successive repetitions of the task during thesame day and a re-recording of the test at differentintervals was to gather evidence of the systematicityof the results obtained.

Recordings

All reading samples were recorded in a well-dampedstudio using a DAT-recorder (TASCAM DA-30MKII DAT tape deck) and a measuring microphone(Bruel & Kjaer 4156). The mouth-to-microphone dis-tance was 40 cm. The signals were high-pass � lteredat a frequency of 50 Hz (Bruel & Kjaer 2120 fre-quency analyzer) to eliminate noise components inthe room related to illumination, ventilation etc.

Acoustic analyses

SPL and F0 analyses were carried out with a speechanalysis system called Intelligent Speech Analyser(ISA), developed by Raimo Toivonen, M.Sc. Thealpha ratio was calculated by dividing SPL at therange 1000–5000 Hz with SPL at the range 50–1000Hz (7). The alpha ratio illustrates the frequencydistribution of sound energy and thus, the acousticvoice quality. Typically, the harmonics over 1 kHzare relatively stronger in hyperfunctional voices com-pared to normal voices, while the opposite is true forhypofunctional voices ((13), see Fig. 1). This isre� ected in the alpha ratio. As an increase in SPL ingeneral leads to the strengthening of higher harmon-ics in relation to lower ones (e.g. (6)), an increase inSPL also tends to imply an increase in alpha ratio.No clear relation between F0 and alpha ratio hasbeen reported.

Fig. 1. LTAS of text reading (oneminute duration, pauses andvoiceless sounds excluded) for afemale subject reading in twoways. Thick line:hypofunctionally ; thin line:hyperfunctionally . Alpha ratio(SPL b:SPL a) in the former 0.80,in the latter 0.92.

Log Phon Vocol 27

Log

oped

Pho

niat

r V

ocol

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f M

elbo

urne

on

11/0

6/14

For

pers

onal

use

onl

y.

M. Artkoski et al.120

Fig. 2. Distribution (in percentage) ofchanges in parameter values for F0 (a), SPL(b) and alpha ratio (c) during a daytime.Values obtained in text reading in theafternoon are compared to those obtained inthe forenoon.

The acoustic results were statistically analysed us-ing a paired Student’s t-test and a Spearmancorrelation.

RESULTS

The values of F0 and SPL were either higher, loweror the same in the afternoon. In most cases, however,

F0 was higher in the afternoon (46:77 cases in thefemales, 9:14 cases in the males). The difference wasrelatively small, within 5 Hz in most cases (see Fig.2a), and statistically non-signi� cant. The averagechange expressed in semitones was 0.14 for the malesand 0.15 for the females (Table 1). For the femalesSPL was in 37 cases higher, in 32 cases lower and in8 cases the same in the afternoon. For the males SPL

Log Phon Vocol 27

Log

oped

Pho

niat

r V

ocol

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f M

elbo

urne

on

11/0

6/14

For

pers

onal

use

onl

y.

Changes in voice during a day 121

Table 1. Values of fundamental frequency (F0),sound pressure level (SPL) and alpha ratio (SPL1000 Hz–5000 Hz:SPL 50 Hz–1000 Hz) in textreading recorded in the forenoon and in the after-noon. Sign.¾signi� cance (paired Student’s t-test)

F0 (Hz) SPL (dB) Alpha ratio

FEMALES (N 11)Forenoon

177.8 67.4 0.85Mean3.6 0.0022.8SD

131–242range 62–76 0.78–0.96

Afternoon67.5 0.87179.5Mean3.7 0.04SD 23.961–78 0.80–1.10128–234range

DifferenceMean »1.7* »0.1 »0.02

¼21–»35 ¼7–»7 ¼0.06–»0.25rangeNS p 0.0009NSSign.

MALES (N 10)Forenoon

65.3Mean 0.86101.75.5 0.05SD 10.557–75 0.79–0.9482–119range

AfternoonMean 102.4 64.7 0.85

4.9 0.078.8SD87–120range 56–73 0.76–0.97

Difference¼0.6 ¼0.02»0.7**Mean

range ¼13–»7 ¼4–»4 ¼0.07–»0.03NS p 0.059NSSign.

*¾0.15 semitones, **¾0.14 semitones, derived from theformula 39.86½log(F0:16.35).

patterns were found in the changes of F0 and SPL.This is in line with earlier observations (1, 5, 8).These different patterns may re� ect the actual circa-dian rhythm of each subject. On the other hand,changes in vocal parameters may naturally be re-lated to emotional state (14, 15, 23, 24, 31) andsituational anxiety (19) for example. These kinds offactors may play a larger role when text reading isused than when the basal pitch is measured as in thestudy by Cooper and Yanagihara. However, textreading was used in the present study in order togain results that are comparable with earlier resultsobtained from text reading before and after a vo-cally loading task. Furthermore, we � nd measure-ment of basal pitch to be a non-reliable methodespecially for untrained subjects whose performancein various vocal tests may vary randomly.

Although in the present study no statistically sig-ni� cant change (¾suf� ciently homogenuous patternof change) was found in F0 and SPL, in most casesF0 was higher in the afternoon. This is also theresult reported in studies on working-day relatedchanges in vocal parameters of voice professionalssuch as teachers. Furthermore, the alpha ratio of thefemales was higher in the afternoon. A similar ob-servation has been made for female teachers after aworking day (21). Additionally, alpha ratio has beenreported to be higher after speci� c vocal loading(theatre performance or a day-long vocal loadingtask) for females, while a somewhat less systematicor even an opposite change has been found formales (16, 18, 21). Higher F0, SPL and alpha ratiosuggest a higher psycho-physiological activity level.Although the changes observed in the present studywere relatively small, the amount of changes in theparameter values does not seem to differentiateloading effects from daytime (or task repetition) ef-fects in general. For example, according to Garrettand Healey (8) F0 was in the afternoon 0.4 semi-tones higher for the females and 1.3 semitoneshigher for the males when no vocal loading waspresent. The circadian changes in F0 reported byBouhuys et al. (2) were about 0.4–1 semitones. Onthe other hand, F0 increase related to (female)teacher’s working day was about 0.7 semitones inthe study by Rantala et al. (22). Thus, it seems thatin order to distinguish pure vocal-loading relatedchanges from possible daytime or repetition relatedchanges in vocal parameters, repeated recordingsshould be made at the same time of the day ondifferent days with and without vocal loading.

Gender differences observed in the vocal changesare particularly interesting. They are also dif� cult to

was slightly lower in 8 cases, higher in 5 and re-mained the same in one case. In most cases the SPLdifference was within 5 dB (Fig. 2b), and it wasstatistically non-signi� cant.

Alpha ratio (Fig. 2c) increased signi� cantly in thefemales (p¾0.0009) while it decreased in the males(p¾0.059). There was a moderate correlation be-tween alpha ratio and SPL for the females (r 0.63,p¾0.000) but no correlation for the males(r ¼0.24, NS).

The results of the smokers did not differ fromthose of the other subjects.

DISCUSSION

According to the results, the values of the vocalparameters measured were in most cases different inthe afternoon compared to the morning. Different

Log Phon Vocol 27

Log

oped

Pho

niat

r V

ocol

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f M

elbo

urne

on

11/0

6/14

For

pers

onal

use

onl

y.

M. Artkoski et al.122

explain. As far as normal speech (de� ned as theopposite to shouting or singing) is concerned, a de-crease in alpha ratio suggests a decrease in glottalclosing speed. This in turn, is regulated by the ac-tivity of the adductory muscles, by the subglotticpressure, and by the biomechanical properties ofthe vocal folds. The male subjects in the presentstudy were somewhat older than the females and allexcept two of the males were vocally untrained,while all female subjects had received vocal train-ing. However, this does not seem to explain thedifferences, since similar observations have beenmade in other studies where the subjects of bothgenders have been more homogenous regarding age,vocal training, profession etc (16, 18, 21, 28). Onthe other hand, differences between the genders areknown to exist in the laryngeal structure and in thevocal fold tissue (10, 11, 17, 26, 27, 30). It seemsmost likely that the male and female vocal organsmay also react differently to vocal loading, changesin stress hormone level etc. The vocal fold tissue isknown to contain hormone receptors (17). They, inturn, may regulate the function of the tissue cellsby modulating gene expression (11, 30). Clinical� ndings suggest that there is some gender-speci� cityin the reactions of the vocal fold tissue e.g. to vari-ous chemicals. We might speculate that the possiblecircadian rhythm related increase in stress hormonelevel in the late afternoon caused an increase in theglottal closing speed for females, while the oppositechange took place in male subjects. Vilkman andManninen (29) also reported that when exposed tosome combinations of various stress-evoking factors(dry bulb temperature, stable broadband noise,stochastic and sinusoidal whole body vibration)their male subjects (unfortunately , no females wereincluded in the study) showed a decrease in spectralenergy above 1000 Hz (which corresponds to low-ered alpha ratio). Further investigation on thistopic is warranted.

CONCLUSIONS

A tendency toward a slightly higher mean F0 wasfound in both males and females in text readingsrecorded in the afternoon compared to readingsrecorded in the morning. The alpha ratio was sig-ni� cantly higher for females and lower for males inthe afternoon. Since similar changes have been re-ported after vocal loading, it seems advisable thatstudies on the effects of vocal loading should alsoaddress changes in vocal parameters when no vocalloading is present. Gender difference in the vocalchanges warrants further study.

ACKNOWLEDGEMENTS

The authors warmly thank the subjects for partici-pation of the study. Mrs Virginia Mattila, M.A., isthanked for language correction of the manuscript.

REFERENCES

1. Austin MD, Leeper HA. Basal pitch and frequencylevel variation in male and female children: A prelimi-nary investigation. J of Comm Disorders 1975; 8: 309–16.

2. Bouhuys AL, Schutte HK, Beersma DGM, NieboerGLJ. Relations between depressed mood and vocalparameters before, during and after sleep deprivation: acircadian rhythm study. Journal of Affective Disorders1990; 19: 249–58.

3. Buekers R. Voice performances in relation to demandsand capacity. Development of a quantitative phono-metric study of the speaking voice. Doctoral disserta-tion, University of Maastricht, 1998.

4. Colquhoun WP (ed.) Biological rhythms and humanperformance. London: Academic Press, 1971.

5. Cooper M, Yanagihara N. A study of the basal pitchlevel variations found in the normal speaking voices ofmales and females. J of Comm Disorders 1971; 3:261 –6.

6. Fant G. Acoustic theory of speech production.Mouton, The Hague, 1970.

7. Frøkjær-Jensen B, Prytz S. Registration of voice qual-ity. Bruel and Kjær Technical Review 1976; 3: 3–17.

8. Garrett KL, Healey CE. An acoustic analysis of � uctu-ations in the voices of normal adult speakers acrossthree time of day. J Acoust Soc Am 1987; 82 (1):50–62.

9. Gelfer MP, Andrews ML, Schmidt CP. Effects ofprolonged loud reading on selected measures of vocalfunction in trained and untrained singers. J Voice 1991;5: 158–67.

10. Hirano M, Kurita S, Nakashima T. Growth, develop-ment and aging of human vocal folds. In: Bless DM,Abbs JH, editors. Vocal fold physiology. Contempo-rary research and clinical issues. San Diego: College-Hill Press, 1983. pp. 22–43.

11. Holt GR, Aufdemorte TB, Sheridan PJ. Estrogen re-ceptor in the larynx of the aged baboon (Papio cyno-cephalus). Ann Otol Rhinol Laryngol 1986; 95 (6):608 –17.

12. Kitzing P. Glottogra� sk frekvensindikering. Doctoraldissertation. University of Lund, 1979.

13. Kitzing P. LTAS criteria pertinent to the measurementof voice quality. Journal of Phon 1986; 14: 477–82.

14. Laukkanen A-M, Vilkman E, Alku P, Oksanen H.Physical variations related to stress and emotionalstate: a preliminary study. J of Phon 1996; 24: 313–35.

15. Laukkanen A-M, Vilkman E, Alku P, Oksanen H. Onthe perception of emotions in speech: the role of voicequality. Log Phon Voc 1997; 22: 157–68.

16. Lauri E-R, Alku P, Vilkman E, Sala E, Sihvo M.Effects of prolonged oral reading on time-based glottal� ow waveform parameters with special reference togender differences. Folia Phon 1997; 49: 234–46.

17. Newman SR, Butler J, Hammond EH, Gray SD. Pre-liminary report on hormone receptors in the humanvocal fold. J Voice 2000; 14 (1): 72–81.

Log Phon Vocol 27

Log

oped

Pho

niat

r V

ocol

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f M

elbo

urne

on

11/0

6/14

For

pers

onal

use

onl

y.

Changes in voice during a day 123

18. Novak A, Dlouha O, Capkova B, Vohradnik M. Voicefatigue after theatre performance in actors. Folia Phon1991; 43: 74–8.

19. Penzien DB, Hursey KG, Kotses H, Beazel HA. Theeffects of anticipatory stress on heart rate and T-waveamplitude. Biol Psychol 1982; 15: 241–8.

20. Rantala L, Lindholm P, Vilkman E. F0 change due tovoice loading under laboratory and � eld conditions. Apilot study. Log Phon Voc 1998; 23: 164–8.

21. Rantala L, Paavola L, Korkko P, Vilkman E. Working-Day effects on the Spectral Characteristics of TeachingVoice. Folia Phon Log 1998; 50: 205–11.

22. Rantala L, Vilkman E. Relationship between subjectivevoice complaints and acoustic parameters in femaleteachers’ voices. J Voice 1999; 13 (4): 484–95.

23. Scherer K. Speech and emotional states. In: Darby JK,editor. Speech evaluation in psychiatry. New York:Grune and Stratton, 1981. pp. 189–220.

24. Sedlacek K, Sychra A. Die Melodie als Faktor desemotionellen Ausdrucks. Folia Phon 1963; 15: 89–98.

25. Stemple JC, Stanley J, Lee L. Objective measures ofvoice production in normal subjects following pro-longed voice use. J Voice 1995; 9: 127–33.

26. Titze IR. Physiologic and acoustic differences betweenmale and female voices. J Acoust Soc Am 1989; 85:1699–707.

27. Vilkman E. Occupational risk factors and voice disor-ders. Log Phon Voc 1996; 21: 137–41.

28. Vilkman E, Lauri E-R, Alku P, Sala E, Sihvo M. Effectsof prolonged oral reading on F0, SPL, subglottal pres-sure and amplitude characteristics of glottal � ow wave-forms. J Voice 1999; 13: 303–15.

29. Vilkman E, Manninen O. Changes in prosodic featuresof speech due to environmental factors. Speech Commu-nication 1986; 5: 331–45.

30. Virolainen E, Tuohimaa P, Aitasalo K, Kytta J, Van-haranta-Hiltunen R. Steroid hormone receptors in la-ryngeal carcinoma. Otolaryngol Head Neck Surg 1986;94 (4): 512–7.

31. Williams CE, Stevens KN. Emotions and speech: Someacoustic correlates. J Acoust Soc Am 1972; 52: 1238–50.

Log Phon Vocol 27

Log

oped

Pho

niat

r V

ocol

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f M

elbo

urne

on

11/0

6/14

For

pers

onal

use

onl

y.