INTRODUCTION

Voice is developed under the influence of variations of estrogen, progesterone and testosterone. The deepest hormonal effects are physiological and occur during puberty, period in which there are changes to laryngeal dimensions, which lead to reduction of fundamental frequency (f0) of male voices in approximately one octave and of female voice in some notes, owing to the action of estrogen associated with progesterone.

Androgen - secreted by adrenal cortex and also by the internal lining of the ovary in women, if present in high doses, can develop male characteristics in female voices.

Some hormonal interrelations are the key to help us understand the foundations of menstrual cycle, being that the focus of the cyclic nature of these events is the circulating level of estrogen (Ling, 1986).

During the menstrual cycle, on average 28 days, there is variation in hormone levels resulting in cycle phases. Before ovulation there is a small but significant increase of basal progesterone production, which facilitates the positive estrogen feedback function and increases amplitude of luteinizing hormone (LH) and stimulating follicle hormone (FSH) peaks induced by estrogen (the top peak takes place in the middle of the cycle). Before ovulation, LH starts to secrete progesterone, which is maintained during the luteum stage. Luteum body lasts 14 days and if there is no pregnancy, it decreases its development, reducing the level of progesterone and starting menstruation (Hung, 1993). The decline of pre-menstrual estrogen and progesterone levels modify the vocal fold structure (Perelló, 1973).

During pre-menstrual period, there may be vocal alterations but few women realize that. These changes are normally noticed only four to five days before the menstrual flow and, after it, there is relieve of symptoms within 24 to 48 hours, restoring the normal control women have over their voices (Quinteiro, 1989).

According to Perelló (1973), there are reports of cases of recurrent aphony in each cycle and artists that were hoarse during the menstrual cycle and recovered the voice after the end of the period. The author described the case of a singer with vocal nodules, being that during menstrual cycle, nodules increased in size and prevented her from singing; he also reported 36% out of 1,500 workers in an industry that presented aphony some days before the menstrual cycle. Symptoms were more frequent between 25 and 35 years of age and few women searched for medical help, since they knew from experience that they would be "cured" once they had their menstrual cycle. He also reported that it is possible to find pre-menstrual dysphonia in patients who came to medical visit for dysphonia of other origin, especially by vocal nodules. They reported that some days before menstrual cycle, they were more dysphonic.

Even though authors agree that physiological hormone variation causes dysphonia by increasing fluid retention, there is no agreement concerning etiopathogenesis. Opinions can be classified into four groups, according to Perelló (1973):

1. Retention of sodium ion: the huge fluid retention is produced by retention of sodium ion. The edema influences the nervous system, originating all symptoms.

2. Hormone disorder: in most cases (61%) there is normal hormone formula, in 27% of the cases there is hyperestrogen levels, in 4% there is global deficit, and in 8%, hyperandrogen levels.

3. Water retention is associated with increase in anti-diuretic hormone by overfunctioning of posterior lobe of hypophysis.

4. Sensitization of progesterone.

In addition to fluid retention, other changes can be detected. Laryngeal mucosa cytology suffers cyclic variations. Thus, the percentage of eosinophilic horny cells and cells with kariopinotic nucleus suffers a large increase in moments of ovulation and menstrual cycle. Thus, there are similarities between behavior of laryngeal and vaginal mucosa as a result of estrogen variation.

Laryngeal hemorrhage and tracheal disorders are also reported in association with dysphonia. During the pre-menstrual cycle there is drying of vocal folds, increase in level of acidity (sometimes increased by the presence of esophageal reflux), reduction of laryngeal muscle tonicity, vocal fold edema, and dilation of microvarices, factors that comprise the pre-menstrual vocal syndrome. Estrogen and progesterone act synergistically on the muscle-mucosa complex (Abitbol, 1999). The same author reported that in some patients the influence of estrogen-progesterone modified the structure of the laryngeal mucosa before ovulation, and vocal tone could be slightly affected by the presence of mucus in the vocal folds. This increase in mucus production, even though very substantial, normally does not affect spoken or sung voice. Progesterone has anti-proliferative effect on the mucosa and accelerated desquamation. The hormone reduces and even inhibits capillary permeability, causing tissue congestion. The congestion is very apparent in the thorax, below the abdomen, and in pelvic tissues, as well as in the vocal folds, which can cause dysphonia in the pre-menstrual period. Progesterone increases viscosity and level of acidity of secretions in granular cells, but it reduces the volume and causes relative dryness.

According to Abitbol (1999), vocal dynamics seen by telelaryngoscopy in the menstrual period shows: congested vocal folds, posterior chink, microvarices on the surface of both vocal folds, edema on the posterior third of the vocal folds and in the area of cricoarytenoid muscle, less flexible epithelium, with increased amplitude of vibration and asymmetrical vibration seen under stroboscopic light. Clinical signs of vocal pre-menstrual syndrome are: vocal fatigue, difficulty to produce high notes and sometimes the whole register - low frequencies are rarely affected, loss of vocal power, loss of some high harmonics, and hoarser voice. In addition to these signs, there are other associated symptoms, such as increased nervousness, irritability, pelvic pain, sensation of swelling and asthenia.

A study conducted by Jean Brux (1986) demonstrated correlation between spots in the cervical area and in the vocal folds.

Davis (1992) noticed that vocal changes in pre-menstrual period have been associated with the cases related to problems of vocal pitch, volume, quality and flexibility.

Silverman and Zimmer (1978) studied the presence of pre-menstrual hoarseness and ovulation. They believed that hoarseness could be associated with lowering of f0 and they did not observe any statistically significant difference between f0 in ovulation (on average 217 Hz) and pre-menstrual period (216 Hz).

Biase and Silva (2000) conducted a study of f0 of voice in younger women with pre-menstrual syndrome. Measures of f0 were made two to three days before menstrual cycle, on the second and third days of the cycle and about the 14th day of the cycle. They acoustically analyzed sustained emission of three vowels (/a/, /i/, /u/) concerning hoarseness and values of f0, since the former could be associated with f0 lowering. In this study, the f0 values obtained varied very little, and did not indicate statistically significant difference in determining parameters of f0 in the three studied periods.

Voice professional users can be particularly affected by vocal pre-menstrual syndrome. To speech and voice therapists, whose main work tool is the voice, it is necessary to have complete knowledge of the operation of vocal system and realize when there are vocal variations, so that they can also provide guidance to other professional voice users.

The purpose of the present study was to check the existence of differences in vocal pattern in women in the ovulation period compared to the first menstrual cycle day, using auditory perceptual, acoustic, spectrogram and acoustic parameters; if there were differences, we wanted to find out whether they were perceived by women.

MATERIAL AND METHOD

The study emphasized the occurrence and level of vocal quality affection in 30 undergraduate students of Speech Therapy and Audiology, aged 18 to 25 years, mean age of 20 years, with regular menstrual cycles, non-smokers and without use of oral contraception. The study was approved by the Research Ethics Committee, Federal University of Sao Paulo - EPM (nº505/01).

The subjects were submitted to vocal recording on the first and second days of menstruation. The equipment used was: Marantz recorder - model PMD 221, microphone Audio-Technica - model MB 4000C and cassette tape Sony. The recording was conduced in a silent room with 15cm distance from the microphone to the mouth, for emission of sustained vowel /e/ at habitual conditions, counting of numbers (1 to 20) and singing "Happy Birthday". After 13 days from onset of menstrual cycle (approximately during ovulation), we conducted a new recording for comparison purposes.

After recording, voices were transferred to a Macintosh PC system 7.5, using software SoundScope developed by GW Instruments for spectrogram analysis. To conduct the assessment, we used time of sustained voice for one second, frequency range of 0.0 kHz to 5.4 kHz in the spectrogram and amplitude of 2.0 V in the oscilloscope, with pass filter of 59Hz - narrow band.

To some voices, it was necessary to modify the gain in the software so that the intensity would be within the pre-defined limit of 2.0V. The modification aimed at standardizing intensity in the spectrograms, allowing the comparative analysis between harmonics and noise.

We analyzed in the study:

 Parameters related to sound wave:

- Computer analysis of f0: corresponding to number of vocal fold vibration cycles every second;

- Range of harmonics: frequency in which the highest harmonic is well defined and without interruptions;

- Definition of harmonics: analysis of quality of harmonic definition;

- Jitter (J): indicates the variability of fundamental frequency in the short run;

- Shimmer (S): indicates the variability of sound wave amplitude in the short run, and it is a measurement of phonation stability;

- Harmonic-to-noise ratio (HNR): relates the harmonic component to the noise component of sound wave.

 Presence of noise in the region of harmonics and above them: amount of noise among the harmonics.

Three speech and voice therapists, specialists in voice for at least five years, conducted perceptual acoustic analysis. It is known that even though subjectivity is critical and terminology is imprecise, traditional vocal quality assessments in clinical routine prevail over acoustic analysis. It may provide only impressions (hoarse, breathy, compressed voice), which has been the preference in international clinical practice, or involve different scales and indexes to lead to less subjectivity and more reliable determination of the deviations found (Behlau et al., 2001). In our study, we analyzed: type of voice, resonance, loudness (appropriate, reduced or increases), pitch (appropriate, low or high), vocal stability (stable or unstable), and voice breaks (present or absent).

Moreover, we applied a questionnaire in which the undergraduates answered objective questions to assess their level of awareness about the occurrence of vocal changes in pre-menstrual period up to two days after the beginning of the menstrual cycle (Annex 1).

We adopted significance level of 5% to apply statistical tests. T Student's test for paired data was used to assess the differences between menstrual cycle and ovulation conditions.

The software used to conduct the statistical analyzes was SPSS (Statistical Package for Social Sciences), version 10.0.





RESULTS

Results are shown in Tables 1 to 3 and in Graphs 1 to 7.





Based on perceptual auditory analysis, we observed that voices during the menstrual period had mild to moderate hoarse, breathy or hoarse-breathy quality in most cases. During ovulation, there was a tendency to reduce hoarseness and/or breathiness, that is, some of the voices that were classified as hoarse-breathy had lost only breathiness and had reduced hoarseness, or vice-versa, which justified the percentage increase of hoarse and breathy voices during the ovulation period.

It is important to highlight that 23.3% of the analyzed voices presented worsening of vocal quality in the post-menstrual period, being that there was more breathiness and/or hoarseness during ovulation. However, in general there was improvement in vocal quality in the ovulation period (Graph 1).

As to voice breaks, we could observe that during menstrual period there were few voice breaks, amounting to only 3.3% of the sample, which was not observed in ovulation. The difference between the two periods was not statistically significant (p=0.317) (Graph 2).

Concerning emission stability, we observed that during menstrual cycle voices were usually unstable (70%); during ovulation, there was further vocal stability (63.3%), being that the difference between both periods proved to be not statistically significant (p=0.021) (Graph 3).

As to pitch assessment, we could observe that there was no statistically significant difference between the two studied periods (p=0.801), being that 80% of the analyzed voices were classified as having appropriate pitch in both periods (Graph 4).

Loudness proved to be appropriate in most of the voices in both periods, even though we had observed reduced loudness in ovulation in a greater percentage than in the menstrual cycle. However, there was no statistically significant difference between the two periods (Graph 5).

As to resonance, we observed that in both periods, most of the voices were classified as having laryngeal-pharyngeal resonance, and there was no significant different between menstrual cycle period and ovulation (Graph 6).

In the analysis of harmonics, we obtained the following results: as to values of frequencies of superior harmonics, shown in Table 1, voices in the menstrual period presented lower mean ranges than voices in the ovulation period, being that there was statistically significant difference.

In the analysis of harmonics traits, 73% of the recorded voices presented worse quality of definition during the menstrual cycle (spectrograms 1 and 2).

As to intensity of noise between harmonics, 73% of the voices during the menstrual period presented greater amount of noise. Finally, the amount of noise in the region above the harmonic range was observed as more marked in 73% of the voices during the menstrual period.

In Table 2, we can find the results of the computed analysis of f0. We can observe that 53.3% of the analyzed voices showed they were lower in the menstrual period; however, this difference was not statistically significant (p=0.778).

We observed changes in magnitude of perturbation frequency (jitter), even though the difference was not statistically significant (p=0.441). There was variation in relative measures of shimmer and values of harmonic-to-noise ratio (HNR), being that the difference was statistically significant for both (shimmer and HNR) (shimmer - p=0.020 and HNR - p=0.045) (Table 3).

As to responses to questionnaire, we could observe that only 27% of the undergraduates observed changes in their voice in the pre-menstrual period up to 2 days from the beginning of the period (Graph 7). We can state that this is a representative but not significant percentage. The undergraduates referred the following complaints: tiredness during speech, vocal breaks, hoarse voice, vocal tremor, dry throat and difficulty to reach high notes when singing. On average, they referred these symptoms for three days, involving pre-menstrual period and the menstrual period per se.

It is important to highlight that out of 27% of the undergraduates that referred vocal complaints during the menstrual period, only 13% had actual vocal changes during the period.





DISCUSSION

Upon reviewing the literature, we could observe that there are few studies about voice during the menstrual cycle, being that the conducted studies analyzed the behavior of vocal folds by telelaryngoscopy during menstrual cycle and ovulation, and not vocal quality, harmonic behavior and vocal parameters (Abitbol, 1999).

In our study, we conducted perceptual auditory analysis and could observe that voice on the menstrual cycle presented breathy-hoarse vocal quality, which Abitbol (1999) explained by the presence of edema on the posterior third of the vocal folds and in the cricoarytenoid muscle and presence of posterior chink during the period. As to vocal quality, the authors were unanimous to state that voice gets hoarse some days before the menstrual cycle, and women experience relieve of symptoms once the menstrual flow actually starts (Davis, 1989; Abitbol, 1999).

We analyzed the behavior of harmonics in the menstrual cycle and ovulation and we could observe the presence of harmonics in all spectrograms, being that in the menstrual period they were less defined, owing to presence of noise and maybe by variation of f0, with instability and low reach in spectroscopy, on average 2969.291 Hz. During the menstrual period, harmonics proved to be less defined, superior harmonics reached lower means, there was noise between the harmonics and above them, which made the spectrogram in the region - between the harmonics and above them - become gray. Abitbol (1999) reported that in the menstrual period there was loss of vocal power and loss of some high harmonics. We cannot state that necessarily there has been loss of harmonics, but rather that it was not possible to visualize them owing to presence of noise. During ovulation, in turn, the harmonics filled all or almost all the interval of the spectrogram, with voice average reach of 3862.30 Hz. We could observe that the difference between mean of harmonic range in the menstrual period and in ovulation was statistically significant (p=0.010).

It is known that vocal fold mass or length changes reflect in f0 values. It occurs, for example, in puberty vocal changes, when under the action of sexual hormones, there is uneven increase of vocal folds length and mass in children, which culminates in differences of values of f0 according to gender, in men at about 130 Hz and in women at 230 Hz.

It is known that in the menstrual period there is vocal fold edema with increase in mass, and it was expected that there would be changes to fundamental frequency in the period; however, f0 mean was higher during the menstrual cycle, but there was no statistically significant difference. Even though there was no significance between f0 means in both periods (menstrual and ovulation), we could notice that auditorily there was vocal worsening during menstrual cycle, which justifies the pitch classified as low in the period in 16.6% of the analyzed voices.

Among the studies concerning f0 in the menstrual cycle, we refer to Silverman and Zimmer (1978), who studied the presence of hoarseness in the pre-menstrual period and in ovulation and believed that hoarseness could be associated with lowering of f0. The results concerning f0 indicated that there would be no essential difference between f0 in ovulation (on average 217 Hz) and in pre-menstrual period (216 Hz), that is, the difference between them was not statistically significant.

Biase and Silva (2000) conducted a study with f0 in young women voices with pre-menstrual syndrome. In the study, the values obtained for f0 ranged very little and there was no statistically significant difference among the three studies periods.

We observed that during menstrual cycle there were vocal breaks in only 3.3% of the voices and there were no breaks during ovulation. The presence of edema in the vocal folds of the menstrual period is characterized by increase in vocal fold mass allowing greater amplitude of vibration and reducing the possibility of vocal breaks.

In the menstrual period, we observed greater instability of emission (70% of voices) with appropriate pitch and loudness and balanced resonance. We can infer that even though pitch and loudness were appropriate, instability of auditorily perceived emission during menstrual period was owed to changes in jitter and shimmer.

As to vocal parameters, Higgins and Saxman (1989) studied the variations in frequency of vocal perturbation (jitter) during the menstrual cycle and reported that in the first days of menstrual cycle jitter values were practically zero. However, in ovulation, there were changes in magnitude of jitter.

Contrarily, in our study, we expected greater rates of perturbation in the short run (jitter and shimmer) in the menstrual period. Jitter varied more during the menstrual cycle, which was expected considering that these jitter values in normal subjects can represent small variation of vocal fold mass or tension, varied distribution of mucus over them, symmetry of structures or the involved muscle or neural activity (Behlau et al., 2001).

Shimmer provided indirect perception of noise in the vocal production and the higher the amount of noise in emission, the higher the shimmer, such as in breathiness. It was also affected by presence of vocal folds mass damage, or even from diffuse edema, such as in pre-menstrual tension syndrome (Behlau, 2001). However, we know that in the menstrual cycle the amount of noise is increased between the harmonics, which increases shimmer values (increase characterized by slowing down of vocal fold vibration and spectrum noise). In our study, we observed statistically significant differences in measures of shimmer in the menstrual and ovulation periods.

As to harmonic-to-noise ratio (HNR), which correlated the harmonic component with the noise component of the acoustic wave, we could observe in our study that in the menstrual cycle the ratio was reduced, since there was higher level of noise and the harmonics were less defined; in ovulation, we had reduced level of noise and more defined harmonics, which implies higher harmonic-to-noise ratio. We found statistically significant differences between mean harmonic-to-noise ratio of the two periods.

The undergraduates that participated in the study reported symptoms characteristics of pre-menstrual tension, such as severe cramps, muscle pain, breast swelling, mood changes, nervousness or depression. Vocal complaints referred by undergraduates were practically the same as reported by Abitbol (1999), including the sensation of dry throat.

As to level of awareness of undergraduates of Speech Pathology and Audiology about variation of voice during the menstrual cycle, we could realize that only 27% of them had vocal complaints during the menstrual cycle, being that two of them were choir singers. This percentage can be classified as representative, even though not significant. Thus, we could observe that even though they had observed vocal changes during the menstrual cycle, few of them realized the phenomenon.

CONCLUSIONS

Based on perceptual auditory assessment of menstrual cycle and ovulation, we could conclude that most of the parameters used to detect vocal modifications showed significant changes, however, women were not aware of the changes to their voices.

REFERENCES

1. Abitbol B, Abitbol J, Abitbol P. Sex hormones and the female voice. J Voice Paris 1999; 3: 424-46.
2. Behlau M, Feijó D, Madazio G, Pontes P. Avaliação de voz. In: Behlau M. Voz - o livro do especialista. Rio de Janeiro: Revinter 2001;85-245.
3. Behlau M, Madazio G, Pontes P. Disfonias organofuncionais. In: Behlau M. Voz - o livro do especialista. Rio de Janeiro: Revinter 2001;295-341.
4. Biase ND, Silva MC. Estudo da freqüência fundamental em mulheres jovens com síndrome pré-menstrual. Distúrb Comum 2000; 2: 301-11.
5. Davis CB, Davis ML. The effects of premenstrual Syndrome (PMS) on the female singer. J Voice Vicksburg 1992; 4: 337-353.
6. Higgins MB, Saxman JH. Variations in vocal frequency perturbation across the menstrual cycle. J Voice Michigan 1989; 3: 233-243.
7. Hung W. Ovarios y variantes del desarrollo sexual femenino. In: Hung W. Endocrinología pediátrica clínica. Madrid: Mosby; 1993. p.226-67.
8. Ling FW. Distúrbios da menstruação. In: Beckmann CRB. In: Ellis JW. Manual de ginecologia. Rio de Janeiro: Prentice Hall; 1986. p.257-68.
9. Miguel JAS, Perelló J. Foniatría endocrinológica. In: Miguel JAS, Perelló J. Alteraciones de la voz - audiofonología y logopedia. Barcelona: Científico-Médica: 1973; 1-43.
10. Quinteiro EA. Higiene da voz. In: Quinteiro EA. Estética da voz: uma voz para o ator. São Paulo: Summus; 1989. 70-6.
11. Sataloff RT, Embrich KA, Hoover CA. Endocrin dysfunction. In: Sataloff RT. Professional voice: the science and art of clinical care. 2 ed. San Diego: Singular; 1997. p.291-7.
12. Silverman EM, Zimmer CH. Effect of the menstrual cycle on voice quality. Arch Otolaryngol 1978; 104:7-10.
13. Vieira VP, Gonçalves MIR, Pontes P, Pontes AL. Características das vozes roucas ásperas e normais: análise acústica espectrográfica comparativa. Rev Bras Otorrinolaringol 2002; 2: 182-8.

1 Speech Therapist, Specialization in Voice, Centro de Estudos da Voz, Sao Paulo.
2 Ph.D., Joint Professor, Master Studies Program, Universidade Tuiuti do Paraná; Affiliated Professor, Department of Speech Pathology and Audiology, UNIFESP/EPM.
3 Physician, Instituto da Laringe - INLAR, Sao Paulo.
4 Faculty Professor, Department of Otorhinolaryngology, UNIFESP/EPM; Director of Instituto da Laringe - INLAR, Sao Paulo.
This study is the result of a Scientific Initiation Project affiliated to Federal University of Sao Paulo (UNIFESP/EPM), and financially sponsored by Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq).
Address correspondence to: Maria Inês Rebelo Gonçalves - Rua Rio Bonito, 1536 Sao Paulo SP 03023-000.
E-mail: lucianecarrillo@hotmail.com or mariainesgon@ig.com.br