New Insights into Laryngeal Articulation and Breathing Control of Trumpeters: Biomedical Signals and Auditory Perception ^†

Abstract

The activation of the musculature of the larynx of six professional trumpeters during performance was analysed using audio, electroglottography (EGG), oxygen saturation, and heart rate signals. Two university trumpet teachers listened to the audio recordings, to evaluate the participants’ laryngeal effort during performance. Statistical analysis was performed to explore if there were any correlations between parameters extracted from the EGG data and the responses to the audio stimuli by the listeners. Two hundred and fifty (250) laryngeal articulations were identified where laryngeal raising and effort was observed during trumpet performance. It was not possible to find any correlation between the EGG data and the auditory evaluation results, but both listeners could clearly hear the laryngeal effort.

1. Introduction

This paper aims to contribute new scientific evidence that will foster improvement in the teaching of trumpet playing and to enrich the information conveyed to the students. Currently, one can easily find information on various aspects of learning to play the trumpet. There are several books and videos, such as Arban’s Complete Conservatory Method for Trumpet [1] or The Breathing Gym [2] by tuba players Sam Pilafian and Patrick Sheridan, who explain the position of the mouthpiece, how to breathe, and many other technical aspects that aid students and teachers in their search for a highly effective performance. However, “the function of the orofacial and pharyngeal musculature for sound generation in brass instruments is insufficiently investigated” [3] (p. 1171). It is therefore important to gather information in this area to respond to problems students constantly face, regarding breathing, air emission, and study strategies.

When a person exhales forcefully with a blocked mouth and nose, changes in intrathoracic and intra-abdominal pressure take place that dramatically affect venous return, cardiac output, blood pressure, and heart rate. This phenomenon sometimes occurs in our daily lives, like when we pick up something that is very heavy. This forced expiratory effort creates an internal pressure in our body that can result in a Valsalva manoeuvre.

The Valsalva manoeuvre comprises four phases [4]. During phase I, heart rate and oxygen saturation decrease because of increased aortic pressure. In phase II, the opposite occurs; heart rate and oxygen saturation increase and aortic pressure decreases. In phase III, the participant begins to breathe normally, and aortic pressure decreases rapidly, because the external compression on the aorta is removed, and heart rate increases rapidly. Finally, during phase IV, an increase in aortic pressure occurs, which in turn decreases the heart rate.

In order to play a wind instrument, the glottis has to be open for the air to circulate, but Valsalva manoeuvres have been previously observed [5] (pp. 98–99) in wind instrument players, whether they are amateurs, students, or professionals.

According to Steenstrup [5] (pp. 98–99), when analysing the different wind instruments, the ones that can most easily trigger the Valsalva manoeuvre are those that require greater pressure in air exhalation, namely the trumpet and the oboe. High expiratory pressure is required in trumpet playing, especially in the upper register. It is also quite common in this register to produce sounds that resemble those that are made when lifting heavy objects.

Steenstrup [5] (p. 103) lists several situations that can activate the Valsalva manoeuvre, starting with the fact that it can also be caused by air pressure being blocked by the tongue, or by stop consonants that do not allow air to escape from the body. This can happen if the participant articulates a sound using only the consonant [t] and not a consonant-vowel (CV) syllable, contracting the abdominal region.

According to Schuman [6] (pp. 23–25), “a strategy for good articulation is to stop thinking about the mechanics and tongue movement and just think of articulation as a syllable that we use when we speak”. Steenstrup [5] (p. 101) also states that during articulation “the use of the tongue in wind instruments should not be different from that which is used when speaking or singing”.

Traditionally, the syllable [to] is used to define the articulation, as can be seen in several methods used in teaching how to play the trumpet, but Steenstrup [5] (p. 101) suggests that the syllable [tɔ] be used, where the vowel is more open. This way, the stop release occurs earlier (in a gesture of anticipatory coarticulation) which allows the air to circulate naturally. Also, the first language of performer, tongue position, and vocal tract resonances have been shown to influence the sound produced by trumpets and trombones [7].

Students who are beginning to study the trumpet, whose lips do not vibrate when attempting to play notes in the higher register for which they are not yet ready, generate a resistance which can, potentially, stimulate the Valsalva manoeuvre. This movement may also be brought on by psychological factors, with hesitation/stress being a possible cause. Often, the trumpeter anticipates that a high note or passage is going to be difficult and, consequently, this stress can lead to the use of exaggerated force, which can trigger the Valsalva manoeuvre. In order to avoid this problem, Steenstrup [5] (p. 104) proposes that the student breathe rhythmically before the first attack of the phrase. In this way, there is no stop between inhaling and blowing, which will make it more difficult for the Valsalva manoeuvre to be activated.

Finally, Steenstrup [5] (p. 100) also points out that sometimes teachers can stimulate the Valsalva manoeuvre by using expressions such as “pull out your stomach to support the note”. This expression can lead the student to contract the diaphragm and abdominal muscles, believing that this will harness the necessary respiratory effort to complete the task. However, the action of these muscles can easily create intra-abdominal pressure that can trigger the Valsalva manoeuvre.

The acoustics of musical instruments, and our current understanding of the physics of how brass instruments produce sound, still raise interesting questions amongst acousticians, such as [8] how can a quarter-wavelength pipe have resonances that include overtones? These questions can only be answered judiciously if we capture the acoustic signal produced by these instruments with measurement microphones, in non-reverberant environments, given the dynamic sound sources hence generated [9].

Pulse oximetry is widely used in various areas of healthcare, sports and research [10,11,12]. It is a non-invasive spectrophotometric method that allows the continuous recording of peripheral oxygen saturation (SpO2—percentage of arterial haemoglobin in the oxyhaemoglobin configuration) and heart rate (HR), at the level of the extremities such as the fingertip or earlobe. This assessment is based on the following physical principles: the absorption of light by haemoglobin varies according to its SpO2; and the signal generated by arterial blood in systole is independent of the signal generated by arterial blood in diastole.

Electroglottography (EGG) is a non-invasive and simple means of measuring laryngeal articulation [13,14]. It was used in this study to monitor vocal fold activity and indicate elevation or lowering of the larynx. This technique is easy to apply to participants, only entailing the placement of a pair of electrodes externally on the neck, in the alignment of the thyroid lamina. A high frequency current flows between the electrodes, and the electrical impedance of the contact area of the vocal folds is measured.

EGG allows for the detection of low frequency signals that are related to the slow movement of the larynx, as can be observed in Figure 1. It has long been known [15] (p. 183) that the EGG waveform reflects “temporal aspects of laryngeal movement during swallowing and that the EGG has potential as a behavioural modification technique in swallowing therapy”.

Sorin et al. [16] conducted a study to test the feasibility of using EGG to record and measure aspects of swallowing, as shown in Figure 2. Their study, based on the EGG waveform during swallowing for six healthy people, two people with dysphagia, and two people with Parkinson’s disease, revealed an increase in impedance that was synchronised with “increases in laryngeal height” [16] (p. 232). It has also been shown that “expiration generally occurs before and after deglutitive apnea” [17] (p. 131).

This could be explained by the movement of tracheal air to a higher position; Sorin et al. [16] (p. 234) concluded that given “the position of the electrodes”, the variations in larynx height were the major factor determining the duration of the different phases and the amplitude of the EGG waveform.

We have focused this introduction on pedagogical issues related to teaching trumpet playing and briefly reviewed studies based on biomedical signals that could reveal new insights into the mechanisms of laryngeal articulations during wind instrument performance. Various other biomedical instrumentations (e.g., magnetic resonance imaging or electromyography) have been used to study musical performance, including surface electromyographic measures of facial muscle activity patterns of trumpet [18] and clarinet [19] players.

In this study, the hypothesis that the laryngeal musculature is activated during trumpet performance, resulting in a Valsalva manoeuvre that can perceived by musicians, was tested in six professional trumpet players using audio signals, EGG, SpO2 and HR.

2. Methods

Audio, EGG, and pulse oximeter data of all the participants during performance were recorded during the first stage of this study. Each participant had to play a predefined excerpt six times, and for all repetitions, low-frequency EGG signals were observed throughout the performance. These signals had a specific typology, quite different from that observed in the laryngeal movement during swallowing (shown in Figure 2). A pulse oximeter was also used to calculate SpO2 and HR levels during the participants’ performance. A study of auditory perception of laryngeal effort was also performed, and statistical analysis was used to calculate the correlation of the perception test results with the audio and EGG data.

2.1. Trumpet Performers

Six trumpeters (five male and one female) were recruited, with a mean age of 22.3 ± 2.4 years and a mean of trumpet playing experience of 11.0 ± 3.2 years (see

Table 1. Demographic characterisation of participants: trumpet performers.

ID	Sex	Age (Years)	Years of Experience
DM	Male	21	13
DO	Male	20	8
FM	Male	23	13
IR	Female	24	14
JR	Male	20	12
LC	Male	26	6

). None of the participants had a high body mass index, were (or had been) smokers, or had any relevant (to the objectives of the study) pathologies.

2.2. Listeners

Two male expert listeners (LG and JS), aged 39 and 45 years, were also recruited. They were orchestra trumpeters and had more than 30 years of experience as performers, having obtained several individual national and international awards.

2.3. Choice of Musical Excerpt

The study was based on the interpretation of a short excerpt: the beginning of Mahler’s Fifth Symphony with the Funeral March, carried out by one instrument only, the trumpet. This excerpt was chosen because certain specific aspects of trumpet performance are present, namely, dynamics, articulation, tuning, register, and resistance. Another factor considered was the participants’ familiarity with the musical passage, because of its wide use in orchestral work and music education.

2.4. Biomedical Signals Data Collection

Participants were standing 30 cm in front of an MKH20-P48 omnidirectional condenser microphone (Sennheiser, Wedemark, Germany) connected to a Scarlett 6i6 audio interface (Focusrite, London, UK) using a Gold Edition XLR Microphone cable (Mogami, Los Angeles, CA, USA). An EGG signal was also collected with an EG2-PCX2 electroglottograph (Glottal Enterprises, Syracuse, NY, USA) connected to the second channel of the audio interface, with two 35 mm diameter EL-2 electrodes (Glottal Enterprises, Syracuse, NY, USA). The electrodes were placed externally in the alignment of the thyroid lamina and their positioning was adjusted to obtain the EGG signal with the highest possible peak-to-peak amplitude, when observing the waveform in real time with Soundcard Scope 1.46.

The recordings were made with Adobe Audition 3.0, at a sampling rate of 48,000 Hz, with 16 bits per sample, using the Focusrite USB 2.0 ASIO Driver Audio Driver 1.8. The data were recorded in stereo format .wav (Windows PCM) without compression.

SpO2 and HR [10,11,12] were also collected with a Pulsox-300i pulse oximeter (Konica Minolta, Tokyo, Japan) at a sampling rate of 1 Hz. The pulse oximeter was switched on for 1 min prior to the beginning of the musical performance, to establish a baseline of SpO2 and HR values. The DS-5 Version 2.00 (090608) 06/2008 program (Konica Minolta Sensing, Tokyo, Japan; Software by Stowood v.2.00 (090608)) was used to extract the pulse oximeter data to an ASCII file.

The participants produced six repetitions of the excerpt with a 30 s pause between each repetition. At the end of the repetitions of the musical piece, the participants sustained the vowel [a] for about 2 s, to establish a reference in the audio and EGG signals regarding amplitude and time. In Figure 3 we can observe the audio signals, EGG, SpO2, and HR of the first three repetitions of participant DO.

The recordings took place in an ABS-AUD.45.1 cabin, produced by Absorsor, Portugal, with sound reduction of 45 dB, located at University of Aveiro’s Speech, Language and Hearing Laboratory (Aveiro, Portugal).

After storage of the raw recordings, Adobe Audition 3.0 was used to segment the files, and Praat version 6.0.05 and Matlab 8.5.0.197613 (R2015a) were used for annotation and analysis of audio and EGG signals. The audio, EGG, SpO2, and HR signals were later synchronised for analysis with Matlab.

2.5. Audio and EGG Signals Annotation and Analysis

The beginning (LA1) and end (LA2) of the laryngeal articulations (see Figure 4) were annotated with Praat version 6.0.05 and saved together with the audio and EGG signals in a binary format. The criteria used to annotate the beginning and the end of the LA1 and LA2 intervals were as follows: beginning of LA1—the instant at which the EGG signal intercepts the x-axis (amplitude = 0) after a negative deflection resulting from the adduction of the vocal folds; end of LA1—moment at which the EGG signal returns to intercept the x-axis; beginning of LA2—instant at which the amplitude of the EGG signal initiates a noticeable increase after an interval in which its mean amplitude is very close to 0, resulting from the adduction of the vocal folds; end of LA2—instant at which the EGG signal intersects the x-axis again before a negative deflection.

Figure 4 shows the audio signals, EGG and the annotation, in the Praat Editor, of the laryngeal articulations (LA1 and LA2) of one of the files.

The annotations were later exported in ASCII format and read by specific scripts in the Matlab 8.5.0.197613 (R2015a) environment where the following parameters were extracted: Duration of laryngeal articulation LA1 (LA1dur); duration of the laryngeal articulation LA2 (LA2dur); duration of the interval between the end of the laryngeal articulation LA1 and the beginning of the laryngeal articulation LA2 (LA1LA2dur); and peak-magnitude-to-RMS ratio (crest factor) of the EGG signal, calculated for the intervals LA1 (LA1peak2rms) and LA2 (LA2peak2rms) with peak2rms function of Signal Processing Toolbox Version 7.0 (R2015a) (MathWorks, Natick). The peak2rms function calculates the ratio between the maximum value and the root-mean-square (RMS) value of an array. The resulting value allows one to analyse how extreme the peak of the waveform is in the range used for its calculation.

The parameter values were finally exported to an Excel 2013 spreadsheet and IBM SPSS Statistics version 22, for statistical analysis.

2.6. Auditory Perception Experiments

The individual repetitions of the six participants (36 items) were segmented with Adobe Audition 3.0 and saved as mono files (48,000 Hz, 16 bit, uncompressed .wav Windows PCM, with a mean duration of 188 s).

The resulting 36 files were then randomised using an online tool [20] and mounted on a single mono file (48,000 Hz, 16 bit, uncompressed, Windows PCM .wav) for posterior listening.

The listeners were seated comfortably at a table in a quiet room, using HD 380 pro headphones with closed dynamic transducers (Sennheiser, Wedemark, Germany), connected to the internal sound card of a laptop computer and the Windows Media Player 12.0.7601.19148 program, to listen to the 36 samples.

A Visual Analogue Scale (VAS), shown in Figure 5, with a straight 100 mm line, was used for the listeners of the perception test to mark with a cross the degree of effort of the participants, the left end of the scale representing (minimum) “Sem esforço”/“with no effort” and the right end (maximum) marked with the words “Esforço Extremo”/“extreme effort”.

A statistical study was carried out to determine if there was a relationship between the values registered on the VAS, and the values of the LA1dur, LA2dur, La1LA2dur, LA1peak2rms, and LA2peak2rms variables. IBM SPSS Statistics version 22 was used to calculate the Pearson correlation coefficient and a significance level of 5% was considered.

3. Results

This section provides a description of experimental work based on biomedical signals and auditory perception results, and their interpretation.

3.1. Typology of Laryngeal Articulations

Several movements of the larynx were registered in all participants, which were designated as laryngeal articulations (LA). Therefore, we sought to justify these articulations by testing the hypothesis of them being part of a Valsalva manoeuvre. As a first test of this hypothesis, we asked a 24-year-old female Speech and Language Therapist (SLT), a member of our research laboratory, to perform laryngeal movements equivalent to those used during a Valsalva manoeuvre. The signals were collected using the same system and under the same conditions described in the Methods section. The resulting EGG signal can be observed in the segment shown in Figure 6.

Two laryngeal articulations (LA1 and LA2) corresponding to an elevation of the larynx (increased impedance corresponding to an increase in EGG signal amplitude) during inspiration (LA1) and expiration (LA2), as indicated with ellipses, were observed.

Figure 7 shows the audio and EGG signals of participant DO (trumpet player) at the end of a musical phrase.

Two laryngeal articulations (LA1 and LA2), which correspond to an elevation of the larynx during inspiration (LA1) and expiration (LA2), are also highlighted with ellipses in Figure 7. The two articulations are observed only in the intervals of silence, during which the participant breathes in a great amount of air to continue the interpretation of the piece. There is no audible inspiration; i.e., the amplitude of the audio signal during this interval is zero.

The EEG waveforms shown in Figure 6 and Figure 7 are quite similar, but they could not correspond to a Valsalva manoeuvre because the duration of both laryngeal articulations was less than 1 s, much less than the 10 s required for a Valsalva manoeuvre to occur [4] (p. 139). SpO2 and HR were also measured with a pulse oximeter (see

Table 2. Oxygen saturation (SpO2) and heart rate (HR).

	SpO2 (%)				HR (bmp)
ID	MIN	MAX	MEAN	STD	MIN	MAX	MEAN	STD
DM	96	100	99	1	82	126	101	9
DO	90	100	97	3	72	116	94	8
FM	95	99	97	1	67	155	88	9
IR	89	100	97	2	98	147	135	11
JR	93	99	97	1	87	146	109	9
LC	96	98	97	1	66	128	104	12

), which corroborated these results.

In Figure 8, the audio and EGG signals for participant DO can be observed for the first repetition, during the production of 3 short notes.

Two laryngeal articulations (LA3 and LA4) which correspond to an elevation of the larynx are indicated with ellipses in Figure 8. The participant probably moves his tongue to the most anterior region of the vocal tract, which results in a raised tongue and increased air volume in the oropharynx, blocking the air in the vocal tract. This can happen if only the consonant [t] is articulated and not a CV syllable, contracting the abdominal region.

Figure 9 shows the 11 laryngeal articulations of participant DO, during his first repetition of the musical passage.

On the y-axis of Figure 9, the amplitude of the EGG signal (arbitrary units) is shown, and the time in seconds from the beginning of the annotated articulations is represented on the x-axis. It is possible to observe that the slow movements of the larynx are constantly occurring throughout the performance.

3.2. Oxygen Saturation and Heart Rate

Table 2 shows the mean values (MEAN), standard deviation (STD), minimum (MIN), and maximum (MAX) values of SpO2 and HR, calculated for the interval that goes from the beginning of the first repetition to the end of the last repetition. The SpO2 values are given as a percentage (%) and the HR values in beats per minute (bpm). Raw Supplementary Materials Files with all SpO2 and HR values can be downloaded from the publisher’s website.

The SpO2 and HR values remained stable throughout the repetitions of all the participants, leading us to disregard the hypothesis that the laryngeal articulations were part of a Valsalva manoeuvre because, as previously discussed, SpO2 and HR levels have divergent values in the different phases of the Valsalva manoeuvre, as described in the Introduction of this paper.

3.3. Parametrical Analysis of Laryngeal Articulations Based on the EGG Signal

Table 3. Parametrical analysis of laryngeal articulations: durations.

	LA1dur (ms)				LA2dur (ms)				LA1LA2dur (ms)
ID	MIN	MAX	MEAN	STD	MIN	MAX	MEAN	STD	MIN	MAX	MEAN	STD	N
DM	29	269	130	51	54	248	143	54	47	854	357	195	35
DO	59	419	138	55	88	265	159	36	6	476	178	101	61
FM	59	251	151	46	68	308	177	54	43	382	156	80	37
IR	56	200	132	33	39	332	130	70	11	305	114	75	48
JR	61	172	113	33	75	361	242	82	116	543	291	91	27
LC	99	319	192	57	43	266	143	42	86	467	188	76	42
All	29	419	144	53	39	361	160	64	6	854	201	134	250

and

Table 4. Parametrical analysis of laryngeal articulations: amplitudes.

	LA1peak2rms				LA1peak2rms
ID	MIN	MAX	MEAN	STD	MIN	MAX	MEAN	STD	N
DM	1.342	2.218	1.771	0.201	1.308	2.582	1.939	0.355	35
DO	1.438	2.471	1.833	0.191	1.434	2.462	1.995	0.210	61
FM	1.331	2.392	1.809	0.262	1.500	3.370	2.023	0.394	37
IR	1.471	3.372	1.784	0.295	1.381	3.900	1.963	0.446	48
JR	1.521	3.258	2.059	0.371	1.351	2.428	1.872	0.268	27
LC	1.443	3.287	1.955	0.346	1.406	3.271	1.931	0.453	42
All	1.331	3.372	1.856	0.292	1.308	3.900	1.961	0.366	250

show the mean (MEAN), standard error of mean (STD), minimum (MIN), and maximum (MAX) values of parameters LA1dur, LA2dur, LA1LA2dur, LA1peak2rms, and LA2peak2rms, and the number of annotated laryngeal articulations (N). The 250 annotated laryngeal articulations occurred in a short time for all participants (mean values of LA1dur = 144 ± 53 ms and LA2dur = 160 ± 64 ms), much less than the 10 s required for a Valsalva manoeuvre to occur [4] (p. 139). The value of LA1LA2dur was 201 ± 134 ms, the mean peak-magnitude-to-RMS ratio (crest factor) of the EGG signal was 1.856 ± 0.292, calculated for the interval LA1 (LA1peak2rms), and was 1.961 ± 0.366 for the interval LA2 (LA2peak2rms). Raw Supplementary Materials Files with all laryngeal articulations durations can be downloaded from the publisher’s website.

3.4. Auditory Perception of Laryngeal Effort

Table 5. Auditory perception of laryngeal effort.

	Listener LG				Listener JS
ID	MIN (%)	MAX (%)	MEAN (%)	STD (%)	MIN (%)	MAX (%)	MEAN (%)	STD (%)
DM	50	87	62	12	39	89	67	16
DO	32	87	64	17	59	95	81	14
FM	33	80	53	16	21	69	48	16
IR	33	92	62	19	46	100	72	16
JR	28	80	57	17	49	84	69	12
LC	32	63	48	11	31	62	45	10
All	35	82	58	15	41	83	64	14

presents the mean value (MEAN), standard error of the mean (STD), minimum value (MIN), and maximum value (MAX) of the perception scores for each participant (trumpet player). As a central tendency, both listeners perceived moderate laryngeal effort values but with a high dispersion for each participant.

Further analysis, considering the six repetitions of each participant, revealed a Pearson’s correlation of 0.079. The dispersion of these data, clearly shown in Figure 10, does not permit the drawing of a straight ascending or descending line through the marked points corresponding to each participant, leading us to conclude that there is no linear correlation between the answers of the two listeners.

Correlations between all measurements (LA1dur, LA2dur, LA1LA2dur, LA1peak2rms, and LA2peak2rms) and the scores assigned by the listeners (total of 250 pairs of observations for each listener) were also calculated, as shown in

Table 6. Correlation table for all laryngeal articulations (N = 250).

		JS	LG
LA1dur (s)	Pearson Correlation	−0.296	−0.116
	Sig. (2-tailed)	0.000	0.066
LA2dur (s)	Pearson Correlation	−0.064	0.076
	Sig. (2-tailed)	0.312	0.233
LA1LA2dur (s)	Pearson Correlation	0.027	0.068
	Sig. (2-tailed)	0.672	0.287
LA1peak2rms	Pearson Correlation	−0.009	−0.121
	Sig. (2-tailed)	0.890	0.055
LA2peak2rms	Pearson Correlation	0.035	0.032
	Sig. (2-tailed)	0.580	0.613

.

Again, the conditions of applicability of the Pearson correlation test were not completely met, since the observations had dependencies (by groups regarding each trumpet player). Therefore, the p-values were not accurate and must be interpreted with care. Only the LA1dur variable presents a “significant” correlation with the values assigned by listener JS, which is not observable in the case of listener LG. Considering what has been discussed so far, we could infer that there were in fact no significant correlations between any of the variables and the listeners’ evaluations. Indeed, we should add that the only apparently significant correlation is not actually so.

The data used in these analyses do not allow us to obtain well-founded conclusions about the correlations (the complete set has several dependences that affect the accuracy of the tests, and the reduced independent set is too small). Nevertheless, the analysis suggests that there are no significant correlations between the measured acoustic variables and scores given by the listeners.

4. Discussion

In the present study, an average of 42 laryngeal articulations were recorded for each participant during performance. Comparing the participants’ EGG signals with the EGG signals recorded during the execution of laryngeal movements equivalent to those performed during a Valsalva manoeuvre, the waveform of the EGG signal was identical. However, the mean values (calculated over the interval that corresponds to a laryngeal articulation) of all participants for SpO2 (97%) and HR (105 bpm) indicated that it was not a Valsalva manoeuvre.

Regarding the perception tests, the mean value attributed by the two listeners on a scale of 0 to 100 for all participants was 61%. Considering the extremes of the VAS used, we can sustain that the laryngeal effort of the participants is moderate. When calculating the correlations between the scores of the two listeners the value was 0.079, indicating that there is no correlation between the answers of the two listeners.

Even with no evident relation between the scores of the listeners, the correlations between all the measurements and the listeners’ scores (total of 250 pairs of observations by each listener) were calculated. In this case, the conditions of applicability of the Pearson correlation test are corrupted, since the observations have dependencies (by groups regarding each trumpet player). Only the variable LA1dur had a significant correlation (p = −0.296) with listener JS scores, but, considering the above, we could infer that there are in fact no significant correlations between any of the variables and listeners’ evaluations.

Air emission has been shown to be an important factor in the control of laryngeal stress. When exhaling the air to the trumpet, one must consider the narrow space of the tube of the mouthpiece and of the lead pipe through which the air passes. If the amount of air during the emission is too great, it will not be fully utilised, which will cause excess air to be wasted at the ends of the mouthpiece.

This study contributes to a deeper understanding of laryngeal activation in trumpet performance. Having control over the exact amount of air needed to perform a musical passage is essential, for the larynx to be relaxed during performance.

5. Conclusions

We could assert that the larynx is activated several times during the performance: 250 laryngeal articulations could be observed in the EGG signal, during the participants’ performances. It was possible to verify with EGG data in the present study that there is an effort in the larynx which is inherent to trumpet performance. It was not possible to find any correlation between the EGG data and the listeners’ perceptions of effort; nevertheless, we could conclude that both listeners recognised that this effort exists, in accordance with the EGG data, each being different for each participant and their own repetitions.

Despite the relevance of this study, it has some limitations, and the results should be interpreted with caution. Since we recorded classical musicians playing a classical piece, our findings cannot be interpreted in relation to all other genres of music (e.g., jazz or pop), so we believe this leaves room to explore this theme further for different styles of trumpet playing.

The reduced number of trumpet players and expert listeners recruited is a factor that also hinders the generalisation of the results. It is considered relevant, as future work, to carry out a similar study with a larger sample of participants to consolidate the conclusions obtained on a topic that currently has much to explore.