Introduction

Humans are the only beings capable of producing voice. Through articulated sounds, we express our thoughts and feelings and transmit our wishes. Since speech is the most important means of communication and expression, any vocal disorder may have severe implications in the social and professional life of the person.

The phonation system consists of the respiratory system, the larynx (vibration source) and the vocal tract (the resonating system comprising the pharynx, the mouth and nose). The respiratory airflow, when going through the opening and closing vocal fold cycle, produces a vibration that will resonate through the vocal tract.

The diagnosis of phonation disorders depends on a multidisciplinary approach and for a precise diagnosis to be made, we should carry out detailed anamnesis, local-regional examination, vocal behavior assessment and psychodynamic analysis.

Larynx and vocal tract may be the center of benign and malignant lesions, in addition to the neurological disorders that affect the mobility and sensitivity of these organs.

Vocal assessment is a reference for vocal therapy, since it investigates the progress of the treatment, in addition to being extremely important to understand the physiology of vocal organs. The assessment was initially made in a subjective fashion, through perceptive acoustic analysis, but lack of consensus among experienced examiners, even with the use of different vocal alterations scales, required a more objective approach in which voice could be analyzed through the use of devices capable of measuring vocal acoustic parameters.

The theory of vocal production is based on the source-filter linear theory by Gunnar Fant (1970)8, in which the source is the laryngeal vibration and the filter (system of selective transmission or resonating system) is the vocal tract. The larynx is a transducer of aerodynamic energy (airflow) in acoustic energy, through the cycles of opening and closure of the vocal folds.

The resonance of the resonating tube depends on its length and transversal diameter (Kent, 1992)14. Formants are the reference standard of vocal tract resonance, in which the highest acoustic energies are concentrated. Through the formants, we may recognize the segmented characteristics of speech. The three first formants are the main determinants of the phonetic quality of a vowel (Camargo, 20003; Kent, 199214).

The energy source and the resonator are independent; vocal fold vibration frequency does not affect the properties of the resonator.

Objective vocal assessment is necessary to study the different segments involved in the process of phonation, such as pulmonary airflow, laryngeal activity and resonance.

The history of acoustic vocal analysis started in this century, from analog methods up to the modern processes of digitalization. In 1920, the oscillogram was developed; it was an amplitude graph compared to time using continuous sign. In 1940, Henrici analyzer with Fourier's analysis was used for the first time and in 1950, spectrography was introduced, which is the analysis of the complex wave in its components of frequency by the use of filters, showing the concentration of speech energy in time. In 1970, a new era started with the introduction of digital technology (Kent, 1992)14.

Since the 50's, there has been a progressive increase in the number of studies focused on vocal objective analysis, together with the advent of theories that explained vocal production and the development of a number of voice laboratories.

There are many instruments for analysis, called dedicated or based if they have their own sound card, or non-dedicated, if the software is coupled to a regular computer sound card (Camargo, 2000)3. Among the different vocal laboratories commercially available in the United States, we may list CSpeech developed by Paul Milenkovich, Computerized Speech Laboratory by Kay Elemetrics and SoundScope (GW Instruments), (Bielamowicz, 1996)2.

Voice samples necessary for the analysis depend on the purpose of the investigation. Sustained vowels may be used in cases of pathologies and to standardize new measures and parameters (Horii, 1979)12. Connected speech checks the impact of vocal alterations in situations of oral communication. Vowels in position of sentence stress may detect difficulties in laryngeal control, since they require more laryngeal activity (Camargo, 2000)3. Initial and final portions of sustained production may reveal higher levels of perturbation and instability (De Krom, 1995)6.

Other technical details must be observed, such as digital recording, repetition of analysis and type of microphone used. The distance from the microphone to the mouth to record the voice should be standardized for the software and once it is done, it should always be maintained. The study by Titze & Winholtz (1993)26 demonstrated superiority of the condensed microphone over the dynamic type.

Sound wave, in this case voice, has three physical qualities: vocal fundamental frequency (hertz - Hz), related to frequency of vibration of the vocal folds, vocal intensity or amplitude (decibel - dB), that depends on sub-glottic pressure and the energy transported by the sound wave, and quality, which is determined by the harmonic combination of the sound, based on the characteristics of the sound source that produced it.

The main acoustic parameters currently used are the following:

A. Fundamental frequency measure (F0-Hz) - it corresponds to the number of glottic cycles per second. According to Behlau and Pontes (1995)1, normal values are 80 - 150Hz for men, 150 to 250Hz for women and above 250Hz for children.

B. Perturbation measures - it refers to how much a period of vibration differs from another that comes after it, concerning frequency and amplitude, representing the level of vibration stability; it depends on the control over the phonation system. Jitter represents the frequency periodicity variation and shimmer represents the amplitude periodicity variation.

C. Noise measures - they evaluate noise in different ranges of frequency. Noise is an aperiodical acoustic signal, originated from the overlapping of various vibration movements with different frequencies, which are not interrelated.

The first studies involving vocal acoustic analysis were made by Saito, Kato & Teranishi (1958)25 who analyzed the fundamental frequency of Japanese voice. Von Leden et al. (1958)27 conducted the first important study about fundamental frequency definition and periodicity parameters (jitter and shimmer). Lieberman (1961)17 studied pitch perturbation, a quick variation of fundamental periodicity of connected speech. Risberg (1962)21 evaluated fundamental frequency variation on connected speech of Swedish and English language. Lieberman (1963)18 studied 32 subjects between the ages of 24 and 71 years and evaluated the periodicity of fundamental frequency in a IBM computer and pitch perturbation in connected speech reflected alterations in the glottic shape and periodicity. Finally, Wendahl (1966)28 correlated jitter and vocal harshness.

Rontal (1975)23 advocated the visual quality of spectrograms, which were easier to understand than the exact quantification of acoustic parameters; nevertheless, these parameters have been widely used and studied throughout the history of acoustic analysis aiming at defining the acoustic correlations with vocal alterations (Wendahl, 196628; Koike, 197715; Hammarberg et al., 198010; Yumoto & Gould, 198230; Eskenazi et al., 19907; De Krom, 19956; Hillenbrand, 199611; Omori et al. 199720; Michaelis et al., 199819). There is interdependence between the parameters, which are sensitive to different acoustic properties, hindering its interpretation and resulting in contradictory outcomes reported, as reported in the literature.

In summary, the parameters of periodicity variation are more related to hoarseness and noise parameters are related to breathiness and glottic chink. In the vocal study of patients with voice pathology, there is no simple comparison of acoustic data and normal range criteria, it is necessary to define new physiological, acoustic and auditory correlations (Camargo, 2000)3.

Many studies were dedicated to standardizing speakers of Brazilian Portuguese. Behlau et al. (1995)1 defined values of fundamental frequency, Jitter and Shimmer for groups of 30 men, 30 women and 30 children who spoke Portuguese from Brazil. Castro & Pegoraro-Krook (1993)4 studied fundamental frequency of 150 male subjects who spoke Brazilian Portuguese and demonstrated that f0 of high vowels /i/ and /u/ were greater that of the lower vowel /a/. Castro, Pegoraro-Krook (1994)5 studied 140 female speakers of Brazilian Portuguese and mean values for speech fundamental frequency during reading aloud and number counting were, respectively, 213.9 Hz and 200.1Hz. Rodrigues et al. (1994)22 evaluated harmonic-to-noise ratio (HNR) of sustained vowel /a/ for 40 subjects (20 men and 20 women) and found values of 8.63 for male and 10.17 for female subjects.

In his doctorate thesis, José Francisco de Góis Filho9 demonstrated the feasibility of a system to record voice close to the producing source, that is, the vocal folds, studying 10 male patients submitted to supracricoid laryngectomy and 10 men with normal voice. A miniaturized hearing aid microphone was adapted to be placed in the biopsy channel of a flexible laryngoscope. Thus, the microphone recorded the voice of the glottic source, without the modifications suffered by the vocal tract when we use a regular external microphone.

The purpose of the present study was to analyze the functioning and voice type obtained by this recording method with a microphone at the pharyngeal position, comparing it to external measures, through acoustic vocal analysis of 50 normal subjects (25 men and 25 women).

In the literature, there were no similar studies that recorded voice and analyzed it from the glottic source. Therefore, we believe that the present study may add new data to the understanding of physiology and vocal pathologies.

Material and Method

In the period between August 1999 and May 2000, in the Service of Head and Neck Surgery at Instituto do Câncer "Arnaldo Vieira de Carvalho" (ICAVC - SP), we examined 50 people, 12 men aged 20 to 58 years (mean age of 32.96 years) and 25 women aged 25 to 40 years (mean age of 36.68 years), with history of no vocal or auditory problems. The research project was approved by the Research Ethics Committee (CAPPesq) at Hospital das Clínicas and Faculdade de Medicina, Universidade de São Paulo.

All subjects were submitted to oroscopy, laryngoscopy with laryngeal telescope at 70º, rhinopharyngoscopy with flexible fiberlaryngoscope, neck palpation and voice assessment (auditory perceptive and acoustic assessments).

Auditory perceptive vocal assessment was performed by two experienced speech pathologists that collected data about vocal quality, resonance, pitch and loudness, which were all normal.

Acoustic assessment was conducted with the voice laboratory Computerized Speech Lab, Model 4305B by Kay Elemetrics, using a dynamic external standard microphone and internal microphone specially designed for the study, similarly to the one used in hearing aids.

The specifications of the microphones were:

A - External microphone: Shure Model SM48S, dynamic, frequency response from 55 to 14,000 Hz, impedance 270 ohms, output level (at 1kHz) of 77.5 dB (0.13mV), noise of 32 dB SLP equivalent by milioersted (60Hz) and 672 g weight;
B - Special microphone: Knowles Model EM 3046, electreto condensed, frequency response from 100 to 10,000 Hz, impedance of 4400 ohms, output level (at 1 kHz) of 97.0 dB (0.4mV), noise of 31 dB SLP equivalent (at 1KHz) and 0.08 g weight.

The special microphone was recovered with silicone to prevent from the entry of saliva into the system.

Voice was collected through three methods, using sustained vowels /a, i, u/:
1. External standard microphone from Computerized Speech Lab (CSL) - patients were standing, arms along the body, microphone 15cm far from the mouth;
2. Special microphone in the external position (MIE) - patients were standing, arms along the body, microphone 2cm far from the mouth;
3. Special microphone on the internal position (MII) - the microphone was placed in the biopsy channel of the fiberlaryngoscope and recorded the voice 1.5 cm above the glottis.

Patients were instructed to breathe in deeply before each emission and to sustain vowels /a/, /i/ and /u/ as natural and long as possible. We selected the most representative emissions for frequency, intensity and quality, or in other words, the most stable portions, discarding the beginning and the end of each emission. Acoustic parameters were collected within 3 seconds.

Vocal data were stored in the computer, recorded in CD and analyzed with the software Multi-Dimensional Voice Program (MDVP) - Model 4305 by Kay Elemetrics. This program calculates 32 acoustic parameters and compares them to a large database of voices, plotting them in graphs that enable quick identification of normal and abnormal parameters.

Among the 32 parameters, we selected 12 that represented voices acoustically and satisfactorily to be compared with the other recording methods.

The selected parameters were:

A. F0 - fundamental frequency (Hz);
B. PPQ(%) - Pitch Period Perturbation Quotient;
C. APQ(%) - Amplitude Perturbation Quotient;
D. NHR (Noise-to-Harmonic Ratio), VTI (Voice Turbulence Index) and SPI (Soft Phonation Index).

Statistical comparison was made between the same parameters recorded in three different methods for the same vowel. The statistical test used was analysis of variance (ANOVA) and the level of significance (F) was 5%. If F was greater than 0.05, there was statistically significant difference. The coefficient of variation of up to 30% showed homogeneity of the sample.

Results

The tables show below present the comparative results between the acoustic parameters of the 3 sustained vowels recorded through different methods:

· CSL - microphone from Computerized Speech Lab in an external standardized position for the Voice Laboratory of Kay Elemetrics;
· MII - special microphone in the internal position;
· MIE - special microphone in the external position.
We presented mean, standard deviation, coefficient of variation, level of F significance and final comparison.




Discussion

What kind of transformation does sound go through from the glottic source to the vocal tract? The comparison between sound recorded at the level of the vocal folds and the final voice recorded externally may answer this question. Based on this premise, we needed a device to measure independently the energy of the glottic source. In the literature, we did not find any device of the kind. There were some indirect methods of assessment of glottic vibration, such as glottography with inverse filtering and electroglottography (Camargo, 2000)3.

Professor Pedro Luiz Mangabeira Albernaz thought about using a hearing aid microphone and José Francisco de Góis Filho got from Audibel, by Philips, a prototype that was placed on the tip of the fiberlaryngoscope, in the biopsy channel, which was initially tested by Góis Filho (2000)9 in 10 normal men and in 10 men submitted to partial supracricoid laryngectomy at the Service of Head and Neck, Instituto do Câncer "Arnaldo Vieira de Carvalho", demonstrating the feasibility of the method.

In the current study, the special microphone was analyzed by comparing its results to the sound recorded by two other situations (internally, at 1.5cm above the glottis and externally, 2cm from the mouth) and the sound recorded externally 15 cm from the patient's mouth, using a regular microphone from the voice laboratory Computerized Speech Lab by Kay Elemetrics.

The analysis of results required multidimensional attention, from the physiology of vocal organs, to speech acoustics, functioning of the voice laboratory and of the special microphone.

In our study, the special microphone was a condensed microphone and the external standard microphone of the Computerized Speech Lab was dynamic. We performed recordings with the special microphone both internally and externally because we wanted to compare the performance of the special microphone with that of the Computerized Speech Lab. In addition, we recovered the special microphone with silicone to prevent saliva from entering into the system.

The voice laboratory conducts acoustic analysis through a process of recording and quantification of objective data from vocal signal, using its digital representation. The speech energy extends as far as 10kHz and 60dB and significant variations may occur in a 10ms-timeframe.

Using the Multi Dimensional Voice Program for acoustic analysis we analyzed the parameters of frequency, amplitude, variation of periodicity and noise, through which we tried to check the type of sound recorded by the special microphone.

All comparisons of results were statistically significant (significance level of 5%) through the analysis of variance (ANOVA). In order to make the description easier, we adopted the following abbreviations to indicate the microphones and its positions, as follows:

· CSL - external regular Computerized Speech Lab microphone
· MIE - special microphone in external position;
· MII - special microphone in internal position.

A - Measures of frequency

Fundamental frequency (F0) of sustained vowels /a/, /i/ /u/ recorded by CSL was respectively, 130.19Hz, 151.11Hz and 156.12Hz, close to the normal ranges of 80-150 Hz for men, according to Behlau & Pontes (1995)1. F0 of sustained vowels /a/, /i/ and /u/, recorded by CSL in women were respectively, 222.96Hz, 254.69Hz, 257.58Hz, equally close to the normal range of 150 to 250Hz reported for women by Behlau & Pontes (1995)1. The study by Castro & Pegoraro-Krook (1993)4 demonstrated that fundamental frequency of vowels /i/ and /u/ (136.2 and 140Hz, respectively) were greater than for vowel /a/ (128.3Hz) in 150 male subjects speakers of Brazilian Portuguese. The same was observed by the present study: F0 of 151.11Hz for vowel /i/ and F0 of 156.12Hz for vowel /u/, and F0 of 130.19Hz for vowel /a/, but with greater f0.

CSL and MIE presented comparable values of f0 in all vowels in men and women, showing that as to fundamental frequency, microphones recorded sounds comparable to external recordings.

MII only presented f0 comparable to CSL and MIE for vowels /a/ and /i/ in men; all the other vowels (/u/ in males and all the vowels in females), the f0 recorded by MII were statistically lower. If CSL and MIE had comparable f0, MII should have recorded comparable f0, because theoretically fundamental frequency depends exclusively on vibration of vocal folds.

The phonation produced with the flexible fiberlaryngoscope may be altered because it is disturbing for the patient. The reflex to a foreign body introduced in the pharynx may be produced by mass and length increase with the respective reduction of tension of the vocal folds. This fact would explain lower f0 found in MII compared to CSL for higher vowels.

B - Pitch Period Perturbation Quotient (PPQ)

All values of PPQ are within the normal range, that is, below 0.84%, except for vowel /i/ of MII and vowel /u/ of MIE.

The wide variability of MIE makes it difficult to compare it, both for MII and CSL. As we will see later for noise parameters, owing to the great sensitivity of the special microphone, when placed externally, it recorded too much noise, which might have influenced variability of PPQ.
In all vowels, PPQ of the sound extracted for CSL was lower than for MII, except for vowel /u/ in males. Greater PPQ in MII can be also explained by difficulty to sustain a periodical vibration because of the fiberlaryngoscope in the pharynx.

C - Amplitude Perturbation Quotient (APQ)

All vowels, under all types of recordings, presented results of APQ within the normal ranges, that is, below 3.07.

There was also great variability of results of MIE. But the results of APQ recorded by CSL were all greater than for MII. To prevent mucus from obstructing the sound input channel, a cover of silicone was applied to the microphone, with a lateral orifice, which could have buffered the effects of amplitude and, consequently, its variability. Another hypothesis is that the vocal tract caused perturbation as the sound passed, which implied greater APQ variability for CSL and lower APQ for MII, which collected the voice right after its production at the glottic source.

D - Noise measures (NHR, VTI, SPI)

NHR (Noise to Harmonic Ratio) is the mean ratio between the non-harmonic spectrum of energy (1,500 to 4,500Hz) and the harmonic spectrum of energy (70 to 4,500Hz). This parameter assesses noise in general, which may be related to frequency and amplitude variations, perturbed noise, sub-harmonic components and voice breaks. All results of NHR proved to be within the normal range (below 0.19). In all cases, results of NHR of the recorded sound by CSL and MII were comparable, but results of MIE were greater than the other two recordings. It demonstrates that an increase in noise recording by MIE, owing to the sensitivity of the hearing aid microphone used in the open, could have compromised the investigation of other parameters.

VTI (Voice Turbulence Index) is the mean ratio between non-harmonic high frequency energy spectrum (2,800 to 5,800Hz) and harmonic energy spectrum (70-4,500Hz), in areas in which the influence of frequency and amplitude variation, voice breaks and sub-harmonic components is minimum. It is a parameter correlated with breathiness but there are few studies in the literature.

Results of VTI were within the normal range (<0.061), except for the male /i/ vowel recorded by CSL and MIE and female vowel /i/ recorded by CSL and MIE. For vowels /a/ and /u/ both in males and females, measures of CSL and MIE were similar, but both were greater than MII. In vowel /u/, VTI was similar for the three methods. Breathiness in normally associated with loss of incomplete adduction or abduction of vocal folds. If it were correct, VTI results that were abnormal for male and female vowel /i/ recorded by CSL and MIE, could also be altered in VTI of vowels /i/ collected from MII. Therefore, non-harmonic components of vowel /i/ can be related to the vocal tract and not to the vocal folds.

Conceptually, VTI measures the noise of high frequencies, and it is related to turbulence caused by incomplete or incorrect vocal fold adduction. In such case, there may be an amplification of high-frequency noises by the vocal tract.

SPI (Soft Phonation Index) is sensitive to the structure of formants, because it is the mean ratio between low-frequency harmonic energy (70 to 1,600Hz) and high-frequency harmonic energy (1,600 to 4,500Hz). Formants are resonance of the vocal tract, that is, frequency ranges (within a harmonic constitution) that concentrate more acoustic energy. The change in shape and elasticity of the vocal tract provides varied acoustic combinations, which leads to a wide variability of speech sounds (Russo, 1999)24.

Software standardization was provided for vowel /a/, therefore we should analyze this vowel isolated. For CSL and MIE they were comparable, but for MII, SPI was greater. CSL and MIE recorded a similar structure of formants, whereas MII recorded a sound with a different structure, with greater low-frequency harmonic energy ratio.

Despite the differences that could have happened during phonation with fiberlaryngoscope, modifying the parameters of fundamental frequency and amplitude, the results of SPI showed that the structure of sound formants recorded internally is differently from the one recorded externally. In future studies, it will be necessary to define the structure of the sound formants recorded internally, through the spectrographic methods and its derivatives, so that we can minimize the maximum interference of the vocal tract in the sound. It may be achieved through the various different approaches related to recording methods, microphone and voice laboratory.

We can use a nose and mouth mask to delimit the action of the resonating tube. If the characteristic of the tube is known, it will be possible to create filters that eliminate any influence of the vocal tract on the sound recorded close to the vocal fold.

All interference described above is preliminary and requires further study of voice acoustics and testings for each of the hypotheses raised here.

If everything arises from an idea, and an idea requires a principle to turn it into reality, this study is one further step towards the understanding of vocal physiology.

Conclusion

By the comparisons made between external recording of voice with a standard microphone of Computerized Speech Lab (CSL), external recording with a special microphone (MIE) and internal recording with a special microphone (MII, we concluded that:

1. the special microphone tested in an external position (MIE) demonstrated that owing to its great sensitivity, there was an increment in noise recording, which modified the result of frequency and amplitude variation parameters. As to fundamental frequency, the sound recorded was similar to that of CSL.

2. the special microphone in the internal position (MII) presented differences compared to MIE and CSL. Differences with CSL may be a result of the phonation with the fiberlaryngoscope, modifying the parameters of fundamental frequency and amplitude, but the results of Soft Phonation Index (SPI) showed that formant structure of the internally recorded sound was different from that recorded externally. In other words, the influence of the vocal tract was lower in the recording of MII, but there was a certain influence that still has to be investigated in further studies.

References

1. BEHLAU, M.; PONTES, P. Avaliação e tratamento das disfonias. São Paulo, Lovise. 312p, 1995.
2. BIELAMOWICZ, S.; KREIMAN, J.; GERRATT, B.R.; DAUER, M.S.; BERKE, G.S. Comparison of voice analysis systems for perturbation measurement. J. Speech. Hear. Res., 39:126-34, 1996.
3. CAMARGO, Z. Avaliação objetiva da voz. In: A atuação da fonoaudiologia no câncer de cabeça e pescoço. São Paulo, Lovise. p.175-94, 2000.
4. CASTRO, V.C.; PEGORARO-KROOK, M.I. Medida da frequencia Fundamental das Vogais em Indivíduos do Sexo Masculino Falantes do Português Brasileiro. Pró-Fono Revista de Atualização Científica, 15:19-22, 1993.
5. CASTRO, V.C.; PEGORARO-KROOK, M.I. Características da Freqüência fundamental da Fala de Indivíduos do Sexo Feminino Falantes do Português Brasileiro. Pró-Fono Revista de Atualização Científica, 6:5?7, 1994.
6. DE KROM, G. Some spectral correlates of pathological breathy and rough voice quality for different types of vowel fragments. J. Speech. Hear. Res., 38:794-811,1995.
7. ESKENAZI, L.; CHILDERS, A.G.; HICKS, D.M. Acoustic correlates of vocal quality. J. Speech. Hear. Res., 33:298-306, 1990.
8. FANT, G. Acoustic theory of speech production. Paris, Mouton. 1970 apud CAMARGO, Z. Avaliação objetiva da voz. In: A atuação da fonoaudiologia no câncer de cabeça e pescoço. São Paulo, Lovise, 2000. p.175-94.
9. GÓIS FILHO, J.F. Análise acústica da voz com microfones externo e na faringe. São Paulo, 2000. 107p. Tese (Doutorado) - Escola Paulista de Medicina, Universidade Federal de São Paulo.
10. HAMMARBERG, B.; FRITZELL, B.; GAUFFIN, J.; SUNDBERG, J.; WEDIN, L. Perceptual and acoustic correlates of abnormal voice qualities. Acta. Otolaryngol., 90:441-51, 1980.
11. HILLENBRAND, J.; HOUDE, R.A. Acoustic correlates of breathy vocal quality: dysphonic voices and continuous speech. J. Speech. Hear. Res., 39:311-21, 1996.
12. HORII, Y. Fundamental frequency perturbation observed in sustained phonation. J. Speech. Hear. Res., 22:5-19, 1979.
13. KAY ELEMETRICS CORP. Operations Manual: Multi-Dimensional Voice Program (MDVP) Model 4305, New Jersey. 239p, 1993.
14. KENT, R.D.; READ, C. The Acoustic Analysis of Speech. San Diego, Singular Publishing Group. 238p, 1992.
15. KOIKE, Y.; TAKAHASHI, H.; CALCATERRA, T.C. Acoustic Measures for detecting laryngeal pathology. Acta. Otolaryngol., 84:105-17, 1977.
16. KREIMAN, J.; GERRATT, B.R.; KEMPSTER, G.B.; ERMAN, A.; BERKE, C.S. Perceptual evaluation of voice quality: review, tutorial, and a framework for future research. J. Speech. Hear. Res., 36:21-40. 1993.
17. LIEBERMAN, P. Perturbations in Vocal Pitch. J. Acoust. Soc. Am., 33:597?603, 1961.
18. LIEBERMAN, P. Some acoustic measures of the fundamental periodicity of normal and pathologic larynges. J. Acoust. Soc. Am., 35:344-53, 1963.
19. MICHAELIS, D.; FROHLICH, M.; STRUBE, H.W. Selection and combination of acoustic features for the description of pathologic voices. J. Acoust. Soc. Am., 103:1628-39, 1998.
20. OMORI, K.; KOJIMA, H.; KAKANI, R.; SLAVIT, D.H.; BLAUGRUND, S.M. Acoustic characteristics of rough voice: subharmonics. J. Voice, 11:40-47, 1997.
21. RISBERG, A. Statistical studies of fundamental frequency range and rate of change. In: Speech Transmission Lab. Quart. Prog. And Status Rept., Royal Institute of Technology. Stockholm, 1962. P.7-8 apud LIEBERMAN P. Some acoustic measures of the fundamental periodicity of normal and pathologic larynges. J. Acoust. Soc. Am., 35:344-53, 1963.
22. RODRIGUES, S.; BEHLAU, M.S.; PONTES, P.A.L. Proporção harmônico-ruído: valores para indivíduos adultos brasileiros. Acta AWHO, 13:112?6, 1994.
23. RONTAL, E.; RONTAL, M.; ROLNICK, M.I. Objective evaluation of vocal pathology using voice spectrography. Ann. Otol. Rhinol. Laryngol., 84:662-71, 1975.
24. RUSSO, I.C.P. Acústica e Psicoacústica aplicadas à fonoaudiologia. 2. ed. São Paulo, Lovise. 263p, 1999.
25. SAITO, S.; KATO, K.; TERANISHI, N. Statistical Properties of the Fundamental Frequencies of Japanese Speech Voices. J. Acoust. Soc. Japan, 14:111, 1958 apud LIEBERMAN P. Some acoustic measures of the fundamental periodicity of normal and pathologic larynges. J. Acoust. Soc. Am., 35:344-53, 1963.
26. TITZE, I.R.; WINHOLTZ, W.S. Effect of microphone type and placement on voice perturbation measurements. J. Speech. Hear. Res., 36:1177-90, 1993.
27. VON LEDEN, H.; TIMKE, A.; MOORE, P. Laryngeal vibrations: measurements of the glottic wave. Part I. The normal vibration cycle. Arch. Otolaryngol., 68:1-19, 1958 apud LIEBERMAN P. Some acoustic measures of the fundamental periodicity of normal and pathologic larynges. J. Acoust. Soc. Am., 35:344-53, 1963.
28. WENDAHL, R.W. Laryngeal analog synthesis of jitter and shimmer auditory parameters of harshness. Folia Phoniat., 18:98, 1966.
29. WOLFE, V.; CORNELL, R.; PALMER, C. Acoustic correlates of pathologic voice types. J. Speech. Hear. Res., 34:509-16, 1991.
30. YUMOTO, E.; GOULD, W.J. Harmonics-to-noise as an index of the degree of hoarseness. J. Acoust. Soc. Am., 71:1544-50, 1982.

---

1 Specialist in Head and Neck Surgery - Instituto do Câncer "Arnaldo Vieira de Carvalho", SP

Faculdade de Medicina da Universidade de São Paulo, SP
Rua Martinico Prado 26, cj 151 - 15o andar - Higienópolis, São Paulo - 01224-010
Tel/fax: (55 11) 3337-7813 cellular: (55 11) 9702-2645 - E-mail: tamaris@ig.com.br

Article submitted on April 25, 2001. Article accepted on June 29, 2001.