Muscle activity and aerodynamic voice changes at different body postures: a pilot study

1. Introduction

Respiratory and extrinsic laryngeal muscles are affected by postural changes. The literature provides several relationships between posture and voice, including laryngeal muscle tension, glottic configurations, voice production, and breathing support for singing and speech.^1–5 For example, it appears that specific body postures and movements induce positive voice changes in both individuals with dysphonia and healthy voice speakers due to the activation of muscle groups that favor greater respiratory control, more economical voice emission, better pitch control, and optimization of vocal resonance.^5–7 Further, voice teachers and clinicians often use postural adjustments as a strategy to promote optimal and healthy voice production and voice rehabilitation,^5,8–10 as an upright posture,¹¹ modified upright posture,^12–14 leaning Tower of Pisa posture,^6,15 or over unstable surfaces.^16,17 Nevertheless, the effect of posture on voice production has been inconsistently reported in previous studies. Moreover, no previous studies were found that use instrumental biomechanical analysis of muscle activations during phonation and posture change.

To address these gaps in our understanding, the primary purpose of this pilot study was to compare the magnitude of electromyographical activation of muscles involved in phonation-breathing functions and their changes due to four standardized body posture. The secondary purpose was to investigate which body posture produces greater changes in aerodynamic parameters, vocal pitch, and loudness in experienced singers. We hypothesize that the most demanding postures (Tower of Pisa posture and the posture over an unstable surface) will generate higher levels of muscular activity, determining changes in the aerodynamic variables of the voice.

To investigate the effects of posturing on voice production and muscle activation of phonation respiratory muscles, this pilot study will investigate whether commonly defined body postures (or postural changes) would generate measurable greater or lesser activation of external laryngeal and respiratory muscles. Testing these body postures will allow for more likely translation into voice training and therapy routines. This is pivotal to understanding the effects of body posture techniques during voice training and therapy and could lead to improved voice production.

2. Methods

2.1. Participants

Eight participants (four males and four females, with sex as a biological factor; 21–35 years) were recruited for this cross-sectional study. Participants had at least five years of vocal training and singing voice. Exclusion criteria included participants diagnosed with vocal pathologies, chronic or acute airway disease, respiratory pathology, and any physical condition that interfered with postural changes. This study was approved by the ethics committee of the Faculty of Medicine at Pontificia Universidad Católica de Chile; each participant signed informed consent before participating in this study.

2.2. sEMG and kinematic collection

Before testing procedures, the height and weight of each participant were measured, and the age and sex as biological factors were recorded. To start the experiment, participants were prepared for the kinematic recording of each postural task to be performed. Since each posture used was performed after an oral explanation, motion capture of kinematic variables was carried out to assess the main joint angles variation between postures. According to the Plug-In Gait full-body marking protocol, the participants put on a black full-body suit for the subsequent installation of 35 reflective markers on specific anatomical points.¹⁸ For cinematic recording, a three-dimensional motion capture system with eight infrared cameras (Vicon ©, Oxford) was utilized. After this, participants’ skin was shaved and cleaned with alcohol at 90% to place sEMG recording electrodes according to the Non-Invasive Assessment of Muscles (SENIAM) protocol.¹⁹

To measure the muscular activity involved in the phonation-breathing process and postural tasks, sEMG electrodes were placed in the following seven locations: [1] for external laryngeal muscles, suprahyoid group (SH) in the submandibular region, to obtain signs mainly from mylohyoid and digastric muscles considered as mouth floor, toward anterior belly of digastric muscle;²⁰ infrahyoid group (IH), bilateral to the larynx, 1 cm from the thyroid notch.²⁰ For cervical postural muscles [2] scalene (S), an electrode was placed 2–3 cm caudal to the mastoid process,²¹ sternocleidomastoid (SCM), an electrode was placed 2 cm below mastoid process, towards the medial end of clavicula,²¹ upper trapezius (UT), at 50% on the line from the acromion to the spine on vertebra C7, in the direction of the line between the acromion and the spine on vertebra C7.¹⁹ For respiratory muscle [3] rectus abdominis (RA), electrode 2 cm from the midline of the abdomen and 3 cm above the umbilicus.²² Finally, for thorax postural muscle [4] lumbar multifidus (LM), with electrode aligned with a line from the tip of posterior superior iliac spine to the interspace between L1 and L2 interspace at the level of L5 spinous process (i.e. about 2 – 3 cm from the midline).¹⁹

For further sEMG amplitude comparison, all muscles’ maximum isometric voluntary contraction (MIVC) was recorded and used for signal normalization. The MIVC was obtained using manual resistance generated by one of the researchers. For SH and IH, participants were asked to “open the jaw” as strongly as possible. For SCM and S, the participants were asked to tilt their heads to the dominant side as strongly as possible in a sitting position. For the UT, participants were asked to raise both shoulders as forcefully as possible in a sitting position. For the RA, participants laid in a supine position on a table with flexed knees and hips and were asked to perform a maximal trunk flexion. For LMs, participants were laid in a prone position and asked to perform a maximal trunk extension. Each participant rested one minute between each maximal test. For all sEMG records, a 16-channel Surface EMG system (Delsys Trigno Wireless System ©, Boston, USA) was used, with a capture frequency of 2000 Hz. Once the preparation of each participant was ready, they were instructed on how to perform two vocal tasks during the four different postures selected in this study.

2.3. Vocal Tasks and Aerodynamic Variables.

To ensure stability and control of vocal variables, auditory feedback was provided through a digital piano and sonometer. Visual feedback was also provided through the same sonometer located 2 meters away from each participantś head. The vocal tasks utilized were two isolated voice utterances for each posture: [1] /pa:/ as a staccato (five times, at a tempo of 30 BPM given by a digital metronome), and [2] /a:/ as a sustained vowel for 10 seconds. By performing two different vocal tasks, it was possible to compare which of the two phonatory mechanisms generated a better sEMG burst signal and obtain the aerodynamic measures during the staccato task. The output was produced at high intensity and performed at A3 (220 Hz), and A4 (440 Hz) notes by males and females. The sonometer displayed an intensity between 80–100 dB SPL during the voicing. During the performance of the /pa:/ task in each posture, subglottic pressure (SGP), mean flow rate (MFR), laryngeal resistance (LR), vocal loudness (dB SPL), and fundamental frequency (fo) were also collected using the Aeroview System 1.7.0 version (Glottal Enterprises Inc ©, USA). From the fo values, semitones were estimated during each emission at each posture (ref=220 Hz, where 1 ST = 1/12 octave). This metric was calculated using a MATLAB script.²³

2.4. Body postures

Due to their presence in the literature and their use in both voice therapy and training, we selected the following postures, which will be described below (Figure 1): The upright posture,¹¹ modified upright posture,^12–14 leaning Tower of Pisa posture.^6,15 With the aim to evaluate muscle activity during the use of an unstable surface, a balance ball or BOSU^® ball was incorporated. These tools have been used in voice pedagogy and have been related to increased core muscle activation and postural alignment during voice tasks.^16,17

Figure 1.

Representation of each posture utilized in the study. P1= Upright; P2= Modified upright; P3= Leaning “Tower of Pisa”; P4= Upright and standing posture on an unstable surface (BOSU^® ball).

Upright posture (P1): Normal and aligned posture; on a sagittal plane, a vertical alignment through the mastoid process, acromion vertex, greater trochanter, head of the fibula, and anterior to the lateral malleolus.¹¹

Modified upright posture (P2): The modified upright; on a sagittal plane, chin parallel to the floor; pelvic region tilted forward; body weight resting on the metatarsals, and slightly bent knees. On a transverse plane, shoulders rotated posteriorly towards an aligned position. On a frontal plane, feet shoulder-width apart.^12–14

Leaning Tower of Pisa posture (P3): The Leaning “Tower of Pisa” posture; on a sagittal plane, the body is subjected to a state of imbalance by leaning 10° forward to increase proprioception and decrease phonatory effort.^6,15

Upright posture over an unstable surface (P4): The upright posture on an unstable surface. Same as P1 but on a balance ball or BOSU^®.^16,17

For the participants to perform the requested postures, verbal, physical and visual feedback was given to the participants during the pre-sampling practice by a kinesiologist. Participants were allowed to practice each vocal task and posture until they were comfortable with the procedure and could perform each of the abovementioned tasks.

To detect the occurrence of each vocal task, an accelerometry signal from the same sensors of the Surface EMG system (Delsys Trigno Wireless System, Boston, USA), which included a triaxial accelerometer sensor (16-bit, 147 Hz), was used. The suprahyoid sEMG sensor was selected for the detection of the accelerometry signal (Figure 3).

Figure 3.

sEMG activity detection based on acceleration data during sustained vowel /a:/ and staccato voice tasks.

2.5. sEMG data analysis

As is common for sEMG analysis, for each sEMG signal, a 20 Hz digital 4^th order Butterworth high-pass filter was used to suppress postural and voice noise from motion artifacts.^24–26 A digital bandstop filter (220 Hz for males, 440 Hz for females) was used to suppress the noise effect from each vocal over sEMG signals. Finally, a maximal root means square (RMS) calculation with an analysis window of 500 ms was used for synchronizing the two vocal task signals with their recorded accelerometry data. The sEMG activity was expressed as a percentage of MIVC.

2.6. Kinematics data analysis

Kinematic data during each postural task was obtained using a digital reconstruction of the hip and ankle joints using the Vicon© Nexus software version 1.8.5. The signal trajectory of each marker was filtered using a 4th order Butterworth low-pass filter, with a cut-off frequency of 6 Hz. Joint rotation centers were reconstructed based on the whole-body Plug-In Gait marker model. Once the body reconstruction was finalized, the hip and ankle’s kinematic information (flexion/extension angular was exported for further analysis.

2.7. Statistical Analysis

Descriptive data are presented as mean and standard deviation. A one-way repeated measures ANOVA was used to determine differences in the mean of hip and ankle angles between postures. After this, the Bonferroni post hoc rank test was performed to determine which means differed from the P1 position. A non-parametric Kruskal-Wallis test was used to compare sEMG activity of phonatory muscles and aerodynamic voice variables (including fo and loudness) between postures. The effect size between group differences for aerodynamic voice variables are reported as Eta squared (η²) effect size and interpreted as small (<0.01), medium (0.01–0.14), and large (>0.14). Statistical significance was set at p<0.05. Statistical analyses were performed with STATA 15.0.

3. Results

3.1. Participant demographics

Eight participants (four males and four females, with sex as a biological factor) with a mean (SD) age of 26.12 (5.35) years old completed the study. Mean and SD of height and weight were 165.12 (6.74) meters and 63.75 (10.71) kilograms, respectively. No abnormal movements were noticed during participation.

3.2. Reliability of postural changes

A one-way ANOVA was used to detect angle joint differences between postures and confirm measurement acuity. This analysis revealed that hip (p<0.031) and ankle (p<0.01) joint angles differ between postures. The Bonferroni post hoc rank test results show significantly greater hip flexion in the modified upright posture compared to the upright posture (p<0.01). The degree of ankle dorsiflexion was greater in the modified upright posture (p<0.01) and the upright posture over an unstable surface (p=0.004) when compared to the upright posture. The mean and SD of joint angles between postures are presented in Table 1.

Table 1.

Mean and standard deviation of hip and ankle angles in the four postures during the sustained and staccato vocal task.

	Posture
Joint degree	Uprigh mean (SD)	Mod. Upright mean (SD)	Leaning mean (SD)	Unstable surface mean (SD)
Hip	4.49 (6.26)	20.74 (24.05)*	9.68 (8.71)	12.47 (5.12)
Ankle	8.19 (6.57)	23.33 (6.05)*	8.45 (5.65)	16.98 (9.44)*

^(*)Bonferroni test shows a significant difference in angles (p<0.05) between the posture and the corresponding Upright position.

3.3. Sustained vocal task

Mean and SD of the sEMG activity of phonatory muscles in the four postures during the sustained vocal task are presented in Table 2. While there were some differences in muscle activity across postures during the sustained vocal task, these differences were not significant. There is a trend for the LM muscle to increase its activity when switching from an upright to a modified upright and leaning position, but there is no statistically significant difference.

Table 2.

Mean and standard deviation of sEMG activity percentage (%) of MVIC of phonatory muscles in the four postures during the sustained vocal task.

Muscular group	Muscle	Postures
		Upright mean (SD)	Mod. Upright mean (SD)	Leaning mean (SD)	Unstable surface mean (SD)	p-value
External laryngeal muscles	SH	18.22 (9.9)	20.87 (13.94)	26.35 (24.29)	23.83 (18.56)	0.930
	IH	10.21 (3.21)	9.76 (3.72)	9.06 (3.89)	9.19 (2.45)	0.864
Cervical postural muscles	SCM	4.20 (2.60)	4.60 (3.78)	3.92 (3.09)	4.31 (2.97)	0.894
	S	3.25 (2.48)	2.89 (2.06)	2.86 (2.05)	2.92 (1.49)	0.988
	UT	3.55 (2.28)	2.53 (2.28)	2.92 (2.10)	3.79 (3.05)	0.585
Thorax postural muscle	LM	4.88 (5.73)	9.49 (6.53)	9.56 (6.62)	6.91 (7.46)	0.206
Respiratory muscle	RA	6.27 (3.13)	4.46 (3.97)	4.72 (3.95)	6.26 (4.55)	0.516

SH=suprahyoid, IH=infrahyoid, SCM=sternocleidomastoid, S=scalene, UT=upper trapezium, RA=rectus abdominis, and LM= lumbar multifidus. Maximum voluntary isometric contraction (MVIC)

3.4. Staccato vocal task

Mean, and SD of the sEMG activity of phonatory muscles in the four postures are presented in Table 3. No significant differences in muscle activity and voice variables were detected across postures during the staccato vocal task. As with the sustained vocal task, there is also a trend for the LM muscle when switching from an upright to a modified upright position and leaning position; however, it is not statistically significant.

Table 3.

Mean and standard deviation of sEMG activity percentage (%) of MVIC of phonatory muscles in the four postures during the staccato vocal task.

Muscular group	Muscle	Posture
		Upright mean (SD)	Mod. Upright mean (SD)	Leaning mean (SD)	Unstable surface mean (SD)	p-value
External laryngeal muscles	SH	14.91 (9.49)	18.11 (13.13)	23.93 (22.26)	19.51 (16.77)	0.981
	IH	9.05 (3.22)	8.80 (3.55)	8.67 (3.79)	8.79 (4.53)	0.941
Cervical postural muscles	SCM	3.31 (2.54)	3.78 (2.73)	3.22 (2.20)	3.30 (2.28)	0.962
	S	2.95 (2.13)	2.72 (1.88)	3.01 (2.17)	2.90 (1.70)	0.998
	UT	3.51 (2.27)	2.48 (2.16)	2.97 (2.05)	2.74 (2.25)	0.511
Thorax postural muscle	RA	5.65 (3.55)	3.90 (3.98)	4.00 (3.45)	4.69 (3.67)	0.458
Respiratory muscle	LM	5.25 (7.14)	8.87 (5.86)	9.90 (6.42)	5.61 (5.29)	0.156

SH=suprahyoid, IH=infrahyoid, SCM=sternocleidomastoid, S=scalene, UT=upper trapezium, RA=rectus abdominis, and LM= lumbar multifidus. Maximum voluntary isometric contraction (MVIC)

Mean, and SD of voice production variables are presented in Table 4. While there were some trending differences in fundamental vocal frequency, they were not statistically different. However, after multiple comparisons analysis, a significant difference was detected in semitones between modified upright and unstable surface postures. The three aerodynamic variables and vocal loudness were no statistically significant differences.

Table 4.

Mean and standard deviation (SD) of voice production variables in the four postures during the repeated vocal task.

Aerodynamic and voice variables	Posture
	Upright (mean, SD)	Mod. Upright (mean, SD)	Leaning (mean, SD)	Unstable surface (mean, SD)	p-value	Eta squared (η2) effect size
SGP (cm H2O)	15.00 (1.82)	14.55 (2.39)	14.34 (2.97)	14.91 (2.44)	0.958	0.013
MFR (mL/sec)	309.91 (147.67)	345.40 (142.77)	287.46 (126.64)	308.94 (169.7)	0.848	0.022
LR (cm H2O/mL/sec)	0.09 (0.10)	0.05 (0.03)	0.22 (0.33)	0.11 (0.133)	0.668	0.110
SPL (dB)	66.81 (3.45)	67.20 (3.94)	67.06 (3.68)	67.71 (2.76)	0.818	0.010
fo (Hz)	220.83 (5.46)	221.43 (4.16)	220.83 (3.00)	210.70 (11.18)	0.1952	0.333
Semitones	−12.00 (0.54)	−11.90 (0.51)	−11.91 (0.32)	−12.81 (0.91) *	0.010	0.288

SGP= subglottic pressure; MFR= mean flow rate; LR=laryngeal resistance; SPL= sound pressure level (dB); fo = vocal fundamental frequency (Hz) during the performance of the staccato (/pa:/ task) in each posture.

4. Discussion

The purpose of this pilot study was to compare the magnitude of electromyographical activation of muscles involved in phonation-breathing functions in four common body postures; additionally, the relationship between body posture and aerodynamic parameters, vocal pitch, and loudness in experienced singers was investigated. The results of our study did not reveal significant differences in sEMG activity, aerodynamic parameters, vocal pitch, and loudness among body postures during the production of vocal tasks. However, the semitones revealed significant differences in the unstable surface when compared to the upright posture, modified upright, and leaning postures.

Previous studies have reported that postural modifications are associated with increased muscular activation of respiratory muscles, reduced activity of laryngeal and paralaryngeal muscles, and healthier and more economic voice production.^5,27 Particularly, it has been reported that an aligned body posture would positively affect voice production, as it improves breathing and airflow control for phonation.^10,28,29 Also, body alignment and head position would benefit a better laryngeal positioning and vocal tract shape impacting voice acoustics.^30,31

Otherwise, several studies have reported that misaligned posture or postural tension generates muscular compensations at respiratory and phonatory levels. On the one hand, increased muscle activity and tension negatively affect inspiratory volume by placing the respiratory muscles in a shortened position.^3,4,32,33 Conversely, muscle tension at the head and neck level also affects laryngeal muscles, potentially resulting in dysphonia.³⁴

While these studies may provide important insights from the point of view of postural alterations, none of them considered an experimental design like the present study. In fact, to the best of our knowledge, it appears that our research is the first to incorporate a combination of acoustic, aerodynamic, and sEMG measurements simultaneously to understand how different body positions affect muscular and phonatory function. For example, a study evaluated cervical and lower thoracic muscle activity using sEMG during the production of the maximum phonation time (MPT) task in subjects with dysarthria after a postural repositioning session, resulting in an earlier onset of EMG activity before voice production.³⁵ Although this study considered EMG measurement during a postural change and a vocal task, it is not comparable with our findings since the target group was subjects with dysarthria.

Using voice aerodynamic features, a recent study showed possible therapeutic benefits of a traditional Chinese exercise, Liuzijue Qigong (LQG), a posture similar to the P2 posture in our study.³⁶ This study in healthy subjects compared aerodynamic voice changes during the LQG posture versus sitting. Subjects experienced a statistically significant increase in SGP and fo compared to subjects who performed the exercise in a seated position. The authors attribute these improvements to the LQG posture itself, positing it helped to tighten and stabilize the abdominal and lower trunk muscles, thus generating greater breathing support during phonation.³⁶ However, this hypothesis cannot be confirmed because their study did not include quantitative measures of the muscle activity of these muscle groups. Further, they compared the LQG posture with a seated posture as the control posture, while we compared P2 with an upright posture as the control; thus, it is possible that the positive changes in Gong (2022) could be due to the magnitude of the postural change, which is more significant in the seated position. The above could be interpreted as the use of postural modifications that seek greater or lesser activations of the musculature involved in the breathing/phonation process and posture. In addition, the absence of significant differences is probably due to the similarity of the selected postures, considering that these were initially selected because they are commonly used in training and therapeutic contexts. Future studies select postures that increase the external torques by increasing external loads and/or lever arms, which would produce a greater effect on the forces generated by the tested muscles,^37,38 as well as interventions that use dynamic postural changes over postures maintained in the upright position.

As in all studies, this one is not without limitations, including that the number of participants in our study is small and that a second group of participants was not included to make some comparisons; nevertheless, this study could be considered as a reference, as a pilot study, to be replicated in larger groups and in other populations, considering our methodology. Further, when comparing the joint angles achieved in each posture, only the modified upright posture generated significant differences in the hip and ankle joint positions, so the absence of differences in muscle activity is possibly due to a low mechanical exertion of the postures used in the muscles evaluated. Several biomechanical studies have shown that muscles of the trunk and lumbar region increase their internal force and torque level when there are greater external demands with load manipulation^39,40 and increase of muscle activation and force generated during verticalization from pronounced squatting postures.⁴¹

While our study did not detect differences in muscle activity or voice aerodynamics in isolated vocal tasks in singers, future studies are needed to confirm these results. Some dimension related to voice production and vocal quality that should be considered in further studies, as they are essential to understanding the overall impact of postural modifications, is the incorporation of acoustic voice assessment,⁴² and the report of proprioceptive sensations by the subjects during different postures.¹⁵ It would also be helpful to consider the use of self-perceived vocal effort scales since it has been shown that head position may be a variable that influences perceived vocal effort.³³ Future studies should also explore other more extreme postures or a more intense vocal protocol that may be able to induce greater muscle activation and/or aerodynamic modifications during phonation. In addition, as this was a pilot study with a small number of participants, future studies are warranted to explore our findings further. Based on the effect sizes reported in aerodynamic and voice variables in our study, the largest effect was detected in vocal fundamental frequency. Using an effect size (Eta square) of 0.33 (large) and statistical power of 80%, and a significance level of 0.05, it is estimated that future studies should include a sample size of at least 26 individuals per group (4) and 106 individuals in total.

While our findings did indicate some trending benefit, the results were not conclusive in support of the common assumptions that postural strategies (considering the four studied here) are beneficial for voice treatment and training. As mentioned, we found that the LM muscle trend towards increasing their EMG activity with a shifting posture from upright to modified upright and from upright to leaning posture. It may be because both postures increase the lever arm of body weight, changing the location of the vector to the anterior region of the body, and increasing the external torque due to the body weight. This may also positively impact vocal production as it has been reported that the deep postural and core musculature plays an important role in favoring the “breath support,” also called “appoggio.”^43–45 The “appoggio” is intended to allow the phonatory muscle components to move freely and the respiratory mechanisms to efficiently perform their essential functions without excessive strain on vocal production.¹

Further, the “appoggio,” and the control over expiratory flow and subglottic pressure by trained subjects may explain why the participants in our study could control vocal loudness independent of posture and vocal task performed. However, although they were also able to maintain a stable vocal pitch during the different vocal tasks and postures, voicing over the unstable surface (P4) generated vocal pitch variability, which was only identified by estimating the fo in semitones. The use of semitones has been used to describe measures of variation in voice, speech, and music as it is a more important measure to understand both the acoustic and physiological mechanisms associated with these activities, as well as to represent pitch perception more accurately.^46–48 This variability was probably due to the unstable surface itself and to the voluntary attempts and muscular adaptations of the singers to maintain postural control and stability according to the requirements of the experiment. Moreover, singing is a multi-tasking activity that requires coordination between aspects related to cognitive loads, stress, respiratory coordination, and articulation, among others.⁴⁹

In addition, it is possible that because the participants in our study were allowed to practice until they were comfortable with the required tasks, they could adapt naturally to each posture and vocal task, demonstrating the exceptional ability of trained singers to compensate for differences given a particular performance. This could be further evidence of the importance of using dynamic or more challenging postures in voice training.

Perhaps, the experiment performed when applied to another population, for example, with voice problems or without previous vocal training, our results would have been more striking. As mentioned above, it is probable that the previous vocal training of the participants in our study and their vocal and proprioceptive skills interfered positively with the stability and control of both muscular and some vocal variables. Previous studies have indicated that specific body postures and movements generate positive voice changes in subjects with dysphonia.^5–7

The results of this current study are a further motivation for further research into the physiology underlying postural modifications for vocal rehabilitation and training purposes. Furthermore, the results of this study provide insights that the effects on voice production, at the muscular and aerodynamic level, assigned to postural changes as a vocal therapy and training strategy might not be as large as is often claimed since we could not detect them.

Conclusions

While body position modifications, as a tool for voice therapy and vocal training, have shown several benefits, this study found no modifications in its target outcomes. The body postures selected did not generate voice aerodynamic modifications of the voice nor in the levels of activation of muscles involved in the phonation-breathing process in individuals with experience in singing voice. As a secondary finding, it was possible to evaluate that the postural modifications did not influence the participant’s control over vocal loudness but did influence the fo stability (estimated in semitones) during phonation on an unstable surface.

Modifications of body postures and their impact on voice production should be further investigated both for their potential physiological benefits at the muscular level and their effects on voice production in its broadest dimension.

Figure 2.

Example of aerodynamic data sampling of voice and sEMG during two postures; A.1= Upright posture; A.2= Standing posture on an unstable surface (A.2). B. = Three-dimensional kinematic reconstruction during posture A.1.