FormalPara Core Messages

  • High-speed videos give details about the development of the vocal folds.

  • Voice Range Profiles can be measured before and after training or treatment of pathological cases, e.g., neoplasms or genetic disorders. The measurement can make pupils conscious of their voices.

  • The mean fundamental frequency development (F0) and the total semitone range as well as the semitone range during continuous speech are interesting biological parameters in child development. They can be the basis for mathematical models, telecommunication, biomarkers, genetics, brain research, etc.

  • Knowledge of the pediatric and hormonal dependency of voice development is necessary for pathology as well as during education and training of voices.

5.1 High-Speed Videos (HSVs)

High-speed videos are valuable to see details during normal childhood and especially pubertal development. HSV examples in boys show in some cases what can be interpreted as vocal fold modification of two adult and two child registers in boys. Two markings of contact maxima of the vocal folds are seen in Fig. 4.6 during the pubertal period. They are seen in Fig. 4.10 but weaker in boys in the postpubertal period corresponding to Voice Range Profiles (Fig. 4.21f–i).

From the beginning to the end of the pubertal period, the vocal folds seem to be mostly matte in both girls and boys; this may be another characteristic of the pubertal period. The finding is also present in the postpubertal period. Normal vocal folds are shiny and slim; this was found in some prepubertal and postpubertal children. Regularity of the surfaces and margins is a sign of normality; the irregularities could be secondary to development but also to difficulties to find/regulate the tones during speech. This is also the case for mucus found in many pubertal high-speed videos.

Especially in girls, insufficiency in the rear part of the glottis is found. Further research on glottal gaps is necessary. A method based on deep learning has been suggested [1]. One theory is that the rear glottal gap is due to the longitudinal growth of the vocal folds in discordance with the growth of the laryngeal skeleton; another is uneven growth from childhood to adulthood of registers since the upper register always has a rear distance. The same arguments could be the reason for the thickness of the vocal folds found in some cases.

It is noted that already in the prepubertal period which begins with the adrenarche, the vocal folds can be slightly irregular with glottal gaps. As shown in Figs. 4.30 and 4.31 for girls, E1 (estrone) changes from 57 pmol/L to 123 pmol/L and pubic hair stage from 1–4 to 4–6 in the pre- to postpubertal group and for boys total testosterone changes from 0.54 nmol/L to 18.9 nmol/L and testis volume from 2.3 mL to 20 mL. There is a large pediatric activity also with Tanner periods 1–2 versus 5 corresponding to changes in the semitone range, the lowest tones, and the Voice Range Profiles [2].

It was not the intention to study HSV in younger children—for many reasons, one of which was a possible lack of cooperation. But based on a few HSVs, the small larynxes have a less developed skeleton and the mucosa is thicker than in older children. An HSV of a 5-year-old boy is given to the larynx skeleton (Fig. 5.1).

Fig. 5.1
A magnified view of the larynx skeleton of a 5-year-old boy represents a muscular passage with a major gap between the pair of lip-like flaps of the vocal fold.

5 Years, F0 = 327, loudness (dB(A)) = 93

Full size image

Probably, it is impossible to use HSV alone to evaluate a child to predict voice change in puberty. A combination with Voice Range Profiles is suggested. High-speed videos are better than videostroboscopy due to the correct reproduction of the vocal fold movements. Voice as a biomarker of puberty is an interesting aspect; it covers a short period of up to mostly 8 months of vocal pubertal changes [2], and HSV cannot be used alone for this purpose.

Based on the HSV with pictured oscillometry (kymography) of the middle of the vocal folds, irregularity of fluctuations was not a main finding in the middle of the vocal folds. But we did give up making kymography at other places in the glottis, with the oscillations often being different but regular for that other place in the glottis. Therefore, overall quantitative measures of irregularity of the glottal movements during puberty cannot be used. The pattern can only be described depending on the position of the marker in the glottis [3, 4]. A future perspective is the development of supplementary devices for high-speed films [1]. The high-speed video setups will most likely include more pixels, and a new study might refine the results and understanding of the pubertal vocal fold changes.

Boys and girls from an ordinary school and a high school with an entrance test of minimal musicality in central Copenhagen were used for the study—because a kind of standard reference for voice development is needed. The puberty phenomena dominate the vocal fold appearance over the specific minimum musicality demanded. The results are therefore usable also without specific inclusion tests.

5.2 Voice Range Profiles

Voice Range Profiles had bigger dynamics in the older pupils in both sexes after the pubertal register shift, with a falling of the lowest tones in both sexes. The method discussed in chapter 2 was presented by the European Union of Phoniatricians [5]. A standardized background template was introduced with tones of a piano as abscissa and dB(A) as ordinate. 30 cm distance from a microphone was given. The tone ranges from the lowest to the highest could be measured by the test person from the lowest to the highest intensity with an intensity meter or computer equipment. The background noise should be less than 40–50 dB(A). A survey of setups in the literature was made by Rychel et al. as shown in chapter 2 (Fig. 2.5) [6]. There are several points for discussion.

In our case, we used our own constructed phonetograph calibrated with Brüel & Kjaer, a sound level meter, and a 1000 Hz tone and compared it with equipment from colleagues. With the equipment, mean and range calculations were made. The variations were large as illustrated in the drawings. The average Voice Range Profiles were on the drawings marked until the mean of the highest and lowest tone. The standard deviations of the highest and lowest tones were given.

Since the Voice Range Profile is a physiological parameter, we did not ask for a “warm-up.” “Warm-up” is difficult to standardize, but it can be done. Probably, the measurement should then be done several times on the test person to ensure maximum “warm-up.” There seem to be too many biases. We preferred just to make the examination—but secured each tone to be held for a minimum of 2 s, in the same school room for all in the afternoon after school lessons had finished.

There are and have been many equipment for Voice Range Profiles, e.g., Wevosys and Xion. We controlled the frequency and intensity of our phonetograph systematically with equipment from Brüel & Kjaer. The apparatus gave a tone with a sound that was constructed to give a tone resembling a tone from a piano and was made so that it could only measure a response variation of Hz in that semitone region—not in tones nearby. We developed software to draw lines between the measured standard tones of C-E-G-A-C and included the lowest and highest single-end tone. Average Voice Range Profiles and semitone ranges could then be made. Standard deviations could be made for the lowest and highest tone, and the software in the equipment calculated the mean and standard deviations hereof. The mean yearly Voice Range Profiles and ranges can be used for comparison with other parameters (Figs. 4.24 and 4.23).

The Voice Range Profiles usable in singing were given on the abscissa in some of the figures (Figs. 4.20 and 4.21). These measures are much more variable than the physiological ones, being dependent on many phenomena (talent, training, category, country). Singing category analyses and calculations were made separately, and no significant differences were found before nor after the pubertal pediatric/hormonal changes (Figs. 4.43 and 4.44). Also, here average Voice Range Profiles and ranges were given as well as standard deviations for the lowest and highest tones.

The Voice Range Profiles were standardized in the material of the girls and boys in the presented stratified study. This means that a detailed statistical comparison with other pubertal parameters in the same normal test persons could be made. The total semitone range, the lowest tone, and the average Voice Range Profile area were used. The total semitone range and the lowest tone were analyzed based on the traditional chromatic tone scale of 12 semitones. The lowest tone could also be given in Hz. As for the calculation of the area—the engineer of the phonetograph wanted to use the diatonic scale of 7 semitones (C-D-E-F-G-H). Conversion to chromatic scale can be made. In this study, this was not considered needed since the main purpose of the measure was to use the results to describe how voice changes are related to other parameters of puberty.

Voice Range Profiles in the eighth school year, where the boys mostly get pubertal voice changes, were made of three boys whom all had child’s voice ranges at the beginning of the school year. Boy number three had a totally changed Voice Range Profile in December, and in boys one and two, a total change happened at the first and last measure, respectively. In this period, no statistically specified relationship to serum testosterone could be found (Figs. 4.24 and 4.25). The point of the old versus the new register change as presented in the figures should be noted.

Before the use of Voice Range Profiles as a method for simultaneous registration of the tonal and dynamic range of voice, the development of the voice was mostly described by the F0 and total tonal range. Already at an early stage in the history of phoniatrics, investigations of the tonal range for normal school children were carried out [7]. A summary of the results of research into children’s voices was proposed at the Conference of Logopedics and Phoniatrics in 1936 and subsequently performed by Weiss [8]. This summary covers a period of 4.000 years and shows that people concerned themselves almost exclusively with boys’ and eunuchs’ voices. The average age for the change of voice was 14.5 years; the fundamental frequency in continuous speech (F0) for boys dropped by about an octave and for girls by about 1/3 octave. Frank and Sparber and Wendler et al. arrived at comparable results [9, 10]. Blatt discussed the topic of voice training during puberty [11].

Komiyama et al. performed an analysis of Voice Range Profiles during puberty [12]. They did not, however, make any comparisons with other pubertal phenomena and fixed the lower measurement limit for intensity at 60 dB(A). In our investigations, the intensity of the voice during soft singing was significantly lower than 60 dB(A), and thus the measurements are not comparable.

Meuser and Nieschlag showed that the type and category of voice for adult men (tenor, baritone, bass) are related to the serum testosterone level [13]. Large and Iwata found differences between the formants, which depended on the voice type of adults [14]. We also believe that a distinction between the types of voice must be made if an exact artistic appraisal of the development of the voice during the time of puberty is to be achieved. We did not find hormonal related voice categories in childhood in this study. This aspect could in the future possibly also be considered in investigations of the pathology of the voice. Pedersen et al. made a follow-up on voice disorders [15].

Klingholz et al. carried out Voice Range Profiles on members of the Tölzer boys choir; in addition, Konzelmann et al. investigated the Voice Range Profiles of choirboys [16, 17]. A summary of the literature can be found in the thesis of Bühring [18]. Behrendt followed the development of the falsetto register of the boys of the Thomanerchor school until adulthood but did not relate the phenomena to other parameters [19]. Hacki used the shouting voice measurements in Voice Range Profiles and electroglottography for studying dysfunctions [20, 21]. As referred to in chapter 2, tone ranges were measured in a big German population study during childhood and adolescence in 2021 (Figs. 11 and 12) [22]. It was concluded that two octaves (24 semitones) were the average during childhood and adolescence.

Details of the development of Voice Range Profiles have now been presented in a stratified randomized study and statistically compared with the fundamental frequency in running speech (mean F0), as dependent on pediatric and hormonal development in puberty, in one (the same) population.

5.3 The Speaking Voice

The development of voice in childhood and adolescence is depending on several parameters in a connection, where the mean fundamental frequency (F0) is one out of at least the given ones: the lowest measurable semitone, the semitone range in continuous speech, the total semitone range, and the Voice Range Profile. Results have been presented of the voice parameters combined with the pediatric and hormonal parameters that are usable in further studies on voice development also in pathology (e.g., Figs. 4.26, 4.27, 4.30, 4.31, 4.32, 4.33 and 4.34). A detailed overview of the relations between mean F0, tonal semitone range during speech, total semitone range, and height is given in boys in Fig. 4.28a. In Fig. 4.28b, the relationship between mean F0 and serum testosterone is given. All these details can vary differently in pathology.

As presented in Figs. 4.24 and 4.25, the register shift is abrupt in boys during artistic singing. It is related to serum testosterone. As shown in Fig. 4.29, it changes from 627 Hz when serum testosterone is under 1 nmol/L to 321 Hz when over 10 nmol/L. The register shifts in girls were not in focus but can also be seen, as illustrated in Fig. 4.19d. The definition of an artistic singing Voice Range Profile is made by the pupils themselves, in dialog with the teacher and examiner if they seldomly were in doubt.

Summaries of the scientific work which relates to the fundamental frequency of the speaking voice in children have been made by Baken and Schultz–Coulon et al. [23, 24]. Among others, Fairbanks et al., Michel et al., Hollien and Malcik, Hollien and Shipp, Hollien, Hollien et al., Fitch and Holbrook, McGlone and McGlone, and Coleman et al. have studied the development of the fundamental frequency of the speaking voice in children, without however also investigating the tonal range of the speaking voice [25,26,27,28,29,30,31,32,33].

Vuorenkoski et al. have compared the average fundamental frequency of the speaking voice with hormonal levels in children with endocrinological diseases [34]. Bastian and Unger investigated the fundamental frequency of the speaking voice in the different stages of puberty [35]. Harries et al. used laryngographic measurements on boys and found a good correlation between the sudden drop in frequency seen between Tanner stages 3 and 4 [36]. Lundy et al. used the singing power ratio as an objective means of quantifying the singers’ formant; the values were not significantly different between the sung and spoken samples in young singing students [37].

In the literature, there are not many studies in which the process of bodily maturation in connection with hormonal development has been related to the important secondary sexual characteristic that the voice constitutes. Barlow and Howard used the closed quotient with electrolaryngographic measurements on 127 children with measurable effects on training [38, 39]. Amir et al. and Amir and Biron-Shental showed that it is a good idea to make supplemental sex hormone evaluations in different medical vocal conditions [40, 41]. They also showed that oral contraceptives might stabilize the voice. Cheyne et al. suggest normative values for electroglottography [42].

It is possible that calculations based on new mathematical models can reveal unknown aspects of hormonal regulation of the voice [43,44,45,46]. This would also be interesting for the quantitative differentiation between physiological and pathological voice development [47,48,49]. For voice research, the employment of technologies and the interpretation of the measurement results from a biological point of view are of the greatest importance.

5.4 Puberty Stages and Hormonal Status Analysis

Tables have been made with the traditional division in prepubertal, pubertal, and postpubertal results with the dependent voice parameters as a supplement. Statistical differences in girls were found between the groups for E1 (estrone), E1 sulfate, DHEAS, and androstenedione. This corresponded to a significant difference between the groups for the semitone range in continuous speech, the lowest tone, and the Voice Range Profile area (Fig. 4.30). In boys, the yearly changes were given in the androgens and voice parameters (Fig. 4.31).

The question of hormone-related variations in categories of singing was answered in Figs. 4.32 and 4.33. There was not a significant hormonal or pediatric difference of categories (soprano versus altos, tenors versus basses) at this stage of life.

Laryngologists are asked for the prediction of pubertal voice changes in girls and boys—for many reasons, with one being related to the aspect of child soloists. The mean F0 was used to find predictive results in both sexes.

Predictive calculations of the mean F0 are described in Figs. 4.34 and 4.35. In girls, the expansion of the semitone range in continuous speech (from 3.7 semitones prepubertal to 4.2 pubertal and 5.2 postpubertal) had an overall predictive value together with E1SO4 (estrone sulfate) of P < 0.05. A division in pre-menarche and post-menarche changed the picture. Pre-menarche E1SO4 was still significant, but also height and pubic hair stage were significant. After menarche, semitone range in continuous speech was a predictive factor, together with age and time after menarche (Fig. 4.34). The results were based on logarithmic calculations. Interestingly, there was a linear correlation of SHBG with menarche, r = 0.93. SHBG is predicting mean F0 change in boys: a boy in Tanner stages 2–4, with a mean F0 of 210 Hz and SHBG of 91 nmol, is in puberty based on logarithmic calculations (Fig. 4.35).

Puberty is defined as the period during which the ability to reproduce is attained. In practice work, it is related to the development of secondary sexual characteristics. The normal development of humans during puberty is a very complex process. Howard et al. have produced a survey article that is partly based on an investigation by Tanner and Whitehouse [2, 50]. The development of the voice is described as “the breaking of the voice” at the age of about 14.5 years and the definite attainment of an adult voice about a year later. The body size of Danish children was reported by Andersen and later by Roed et al. and Hertel et al., and it matches our measurements [51,52,53].

In the book edited by Brook, it is highlighted that knowledge of the development of the heart and lungs is limited, and the development of these organ systems until now has only been related to body size and to the development of secondary sexual characteristics [54]. Similar remarks apply to the pediatric literature on voice development. Hägg and Taranger characterize the voice as childish, pubertal, or adult. Karlberg and Taranger describe the breaking of the voice in relation to the stage of puberty at an age of 14.5 years. Heinemann’s work is concerned with abnormal processes in the development of the voice during puberty [55,56,57]. Kahane analyzed the development of the thyroid cartilage in relation to body size [58]. Potassium metabolism increases in close relationship to the level of sexual hormones and depends more on the stage of puberty than on age [59]. Hirano et al. measured the growth of the vocal cords during the time of puberty [60, 61].

Normal endocrinological development is controlled by the gonadotropin-releasing hormone from the hypothalamus. Through the influence of this decapeptide, LH and FSH are released from the frontal lobes of the hypophysis. They regulate the growth of the testes and the ovaries. Sex hormones are produced by these organs. Our methods of measurement have been described by Lykkesfeldt et al.; they were carried out at the Danish Statens Seum Institut and are comparable to Binder [62, 63]. The measurements are also comparable to those of other authors [64]. A review of SHBG has been given [65].

With the method by means of which one can perform hormonal analysis on saliva, possibilities are open for investigating the close relationship between hormonal changes and voice [66]. New insights into the relationships between cerebral regulation and development of the voice in physiological and pathological cases will also make it possible in the future to explore the phenomenon of the change of voice from a neurophysiological point of view [67,68,69,70]. One further perspective of this is that we may expect to discover a new understanding of the psychology of music [71,72,73,74].

Niedzielska et al. compared the change of voice with pathological activation of the gonads in male puberty [75]. Abitbol et al. found that the harmonics are hormonally dependent in female puberty [76]. Breteque and Sanchez analyzed the deepening of the speaking voice in boys and showed the individual nature of the related change of the singing voice [77]. Charpy underlines the concept that voice breaking does exist in adolescent females [78]. Chernobelsky shows that electroglottograms are highly effective in training vocal registers in deaf children also [79]. Chan documented electroglottography improvements in the voices of training kindergarten teachers [80].

Wiskirska-Wonica et al. studied the delay of voice break in adolescent boys [81]. Van Lierde et al. found no statistically significant difference for females, using the dysphonia severity index (DSI) between resonance parameters in the menstrual cycle in 24 healthy young professional voice users [82]. There are other changes of sounds during childhood: Harmonics-to-noise ratio was examined in 9–18-year-olds with no significant changes noticed in females. A transition in harmonics-to-noise ratio was seen in males at the age of 14–15 years [83]. Wide intersubject variation was found in a study of female adolescents in an exploratory study using LTAS and inverse filtering [84]. These measures are related to the development of the lips and jaw. The authors used 3D motion analysis of children with typical speech development compared with children with sound and speech disorders [85].

The fundamental frequency (F0) of voice is naturally an interesting biological parameter not only in childhood puberty, which is the limit of this study, but also during menopause. Truuverk and Pedersen investigated the Voice Range Profile of the speaking voice and its relationship to androgen and estrogen in amateur female choir singers in the World Festival Choir [86]. A connection was found between high estradiol and a larger area in the Voice Range Profile for the speaking voice. Russell et al. analyzed the tonal range of the speaking voice in adult women and obtained similar results [87].

5.5 Further Results from the Statistical Analysis

Further statistical calculations were used to find out how tight the connections were between voice development and pediatric/hormonal development.

In Fig. 4.36, in girls, all voice parameters, mean F0, total semitone range, semitone range in continuous speech, and lowest tone were shown, related to the pediatric/hormonal parameters. Mean F0 in continuous speech, lowest semitone, time after menarche, height, and E2 (estradiol) were not related to the Voice Range Profile area. In boys, the total semitone range, mean F0 in continuous speech (F0), semitone range in continuous speech, and lowest frequency were all significantly related to age and the Voice Range Profile area (Fig. 4.37).

Figure 4.38 illustrates the development of the F0 in continuous speech (in Hz as abscissa) and the relations to Tanner pubic hair stage, testis volumes, serum testosterone, and sex hormone-binding globulin (SHBG).

Figure 4.39 shows in girls the connection between voice parameters and pediatric/hormonal measurements: F0 is related to the lowest tone and E1 (estrone). Nearly all voice parameters are related to age. The semitone range during speech (the F0 tone range) is related to the total semitone range, the Voice Range Profile area, and some of the pubertal as well as hormonal parameters including E1. The lowest semitone is among others also related to E1.

Illustrations were also given with the Voice Range Profile area as abscissa in semitone time dB(A) on a diatonic scale for girls in Fig. 4.40 and for boys in Fig. 4.41. The biological connections between the voice parameters and the pediatric/hormonal measures are illustrated here. In a study of four Thomaner school boys, the prepubertal sopranos and pubertal hormonal values were as in Copenhagen. The prepubertal sopranos had Voice Range Profiles of the same configuration as in Copenhagen although with less variance in the ranges (Fig. 4.42). In Figs. 4.43 and 4.44, the average voice categories in girls and boys in Copenhagen are presented with ranges.

Based on an overview in Fig. 4.45, a yearly difference between mean fundamental frequency (F0) in continuous speech in boys and girls is found after 13 years of age. This is also the case for the lowest biological tone. The semitone range in continuous speech and the average Voice Range Profiles are illustrated.

Overall, the results are usable to help youngsters understand their own voices and discuss their voice-related possibilities. This normal material is basic in a future where voice as a biomarker in pathology in many genetic syndromes deviates from normal [88].

Another use of this book will be in restoring voices in treated child neoplasms. Peripheral precocious puberty is one of the first clinical manifestations of chorion-gonadotropin-secreting intracranial tumors [89]. Deepening on the voice, genetic hyperandrogenism was found in a 13-year-old girl [90]. An 11-year-old girl presented to her oncologist with a recent voice change and increased leg hair growth due to a Sertoli-Leydig cell tumor with androgen excess [91]. Ovarian hilus cell hyperplasia in a girl with Turner’s syndrome and progressive virilization including voice was treated with gonadectomy [92]. Wendler glottoplasty of voice feminization was carried out in a young female patient with irreversible voice changes due to a treated adrenocortical adenoma [93]. Deep learning methods probably can help in future voice diagnostics and treatment [1, 94].