Application Research of Music Education to Improve the Mental Health of College Students under the Background of 5G

Abstract

Fifth-generation (5G) communication technology, with its high speed, low latency, and wide coverage, will open up new possibilities for the extension of online collaborative learning as the society improves with the times. The 5G era’s intelligent information environment has begun to influence our way of life and work, and it has now spread to include our learning habits. In today’s world, with the rapid advancement of computer and communication technologies, a big number of high-tech subjects in our traditional curriculum have had a considerable influence. This study proposes a method of integrating soft computing techniques to the advancement of music education to enhance the mental health of college students in the conditions of 5G in order to help music education stay updated with the progress trends of the periods in this modern generation. Conventional control techniques and new accurate mathematical model control approaches both fail to accurately assess students’ mental health. To enhance the assessment impact of mental health of the students, this study uses an intelligent fuzzy system as the regulation center and offers an assessment method based on an improved fuzzy neural network to assess the impact of music instruction in boosting mental health of the students. When compared to standard ways, the system developed in this work has a better effect, according to the empirical research findings.

This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

1. Introduction

One of the most significant and vital forms of amusement in human lives is music. Music evolves into a vehicle for expressing human emotions and feelings. The large music industry of today is growing and becoming well known thanks to the Internet and streaming devices and applications [1]. Because of the natural relationship between music and emotions, it is an excellent vehicle for working with them. The impact and effects of music on emotion have been explored throughout history and from diverse perspectives. Teachers and researchers in education are concerned with the development of emotional skills, which improve health and improve cognitive behavior. Music is an excellent vehicle for conveying emotions, and music education must play a role in human emotional development [2].

The utilization of music in a therapeutic setting to aid in the rehabilitation of mental health is known as music therapy. In response to the question “What is therapy?” Brown and Pedder believe that therapy is simply a conversation in which people in distress are listened to and played with to assist them know and solve their problems. It is clear that, as enjoyable as it may be, music education requires a greater active participation of patients in order to use the power of music to help students recover their mental health and treat mental illnesses [3].

Music, as an artistic discipline, is immensely significant in basic and secondary schools, as well as higher education. It is the major means of popularizing art in primary and secondary schools, as well as the only method to practice it. In order to attain an aesthetic educational system, it is also a mandatory course for colleges. Music education differs from traditional schooling in various areas. It stresses educators’ emotional experiences as well as teachers’ and students’ emotional exchanges and perceptions. Music education, in general, is a high-quality educational program that incorporates psychological and musical principles. Conventional music education, on the contrary, only scratches the surface of the program and neglects emotional experiences in the classroom, failing to attain the goal of the aesthetic education. As a result, it is critical and valuable to investigate the architecture of the emotional schooling system in musical education in a timely and efficient manner [4].

Schools and other organizations are increasingly paying attention to students’ mental health. Furthermore, many schools and universities require newcomers to take a mental health education course as part of their overall curriculum. Yet there are certain challenges with mental health education right now. Teachers, for example, are unable to care for each and every kid in a large class. Secondly, the substance of this subject is not truly connected to students’ psychological requirements, and there is no structured and highly operable course assessment methodology in place. There are also variances in the professional abilities and personal qualities of teachers [5].

Hence, in this paper, we propose soft computing techniques for teaching music to improve the mental health of college students under the background of 5G. The further portion of the article is structured as follows.

Section 2 explains the literature works associated to this paper. Section 3 explains the flow of the proposed work. Section 4 analyzes the behavior of the suggested methodology and compares it with the traditional techniques. Finally, Section 5 concludes the overall idea of the paper.

2. Related Works

Holochwost et al. [6] investigated whether music instruction was linked to enhanced educational attainment and executive function efficiency. 265 school-aged children were selected by lottery to be involved in an after-school program, which offered individual and large-ensemble orchestral instrument training. Academic attainment was measured using normalized test scores and grades in English language arts from student’s educational details, while executive functions (EFs) were examined utilizing a computerized battery of general EF programs. Students in the musical training program scored maximum on normalized examinations, got better marks in English language arts, and math, and performed better on certain EFs and short-term memory tasks.

Dobos et al. [7] examined the socio-demographic and psychological aspects that contribute to musical performance anxiety (MPA). Musicians who were learning or had accomplished their musical training made up the sample. Music-associated questions were answered by study participants. Females had higher MPA and social anxiety than males, but there was no significant difference in perfectionism. MPA has been connected to social anxiety and perfectionism.

Plumb and Stickley [8] published a qualitative analysis that looked into the advantages of joining a community choir in the United Kingdom. The Institute of Mental Health funds the choir that allures people who utilize mental health applications. To increase mental health and welfare, the choir is guided by social inclusion values. The data were subjected to a thematic analysis procedure after ten members were interviewed. Members mentioned public and health benefits, personal accomplishments, and satisfaction.

Sheppard and Broughton [9] wanted to know how music and dance engagement links to important public determinants of healthcare and how it might be used as a tool for population well-being, health promotion, and prevention for persons in various social groups who do not have preexisting health conditions.

Silverman [10] figured out how music therapists in mental health settings form therapeutic alliances with adults. Eight music therapists who worked with people in health settings were interviewed in semistructured interviews by the investigator. Because therapeutic alliance is such an important part of treatment and is linked to outcomes, music therapy professionals can use emerging themes and subthemes to direct their interactions with adults in mental health settings.

Saarikallio [11] established a theoretical framework for music-dependent social-emotional competence (SEC), which combines general SEC research with articles on musical psychology, musical training, musical therapies, and music for health and well-being. Music-dependent social-emotional competence (MuSEC) is defined by the suggested access-awareness-agency (AAA) design as an interaction of bodily approach, deliberate alertness, and sensation of agencies. Such three elements are characterized as the key competences, which music facilitates in specific; competencies underpin and explain subsequent competence in a variety of performances ranging from affective self-control to public communication.

Henderson et al. [12] highlighted the potential positive health and welfare effects of interactive music acts for culturally and linguistically diverse people, particularly those in fragile or “at-risk” migrant groups.

Hense et al. [13] conducted a practice-based investigation on the effectiveness of introducing the Healthy-Unhealthy Uses of Music Scale (HUMS) items into brief musical therapeutic interventions with infants receiving mental health services. The findings indicated that there was no substantial variation in good and unhealthy music utilization between diagnoses. The sessions provided a variety of benefits to the young people, which are regarded in association to musical therapy practices.

McCaffrey [14] sought to understand the viewpoints of adult service users who have received musical therapy in Ireland’s statutory mental health applications. This sought to provide comprehensive explanations of practices in order to supplement conventional knowledge of what music therapy sessions shall offer. Findings on meaningful occupation, challenge, reciprocity, and frustration contribute to a better understanding of what music therapy may give to mental healthcare users.

Stegemann et al. [15] explored the evidence on the efficacy of musical therapies and alternate music-dependent therapies in pediatric healthcare. Because musical therapy is a tailored intervention that is often provided in a person-centered manner, it is often simple to incorporate into therapeutic practices. It is crucial to emphasize, however, that in order to completely realize the capability of music therapy, specialized education and medical education, as well as a thorough choice of intervention approaches tailored to the client’s requirements, are required.

2.1. Problem Statement

Musical education has been shown to physically develop the regions of the brain known to be involved in processing language and reasoning and can literally connect the brain’s pathways in particular ways, in addition to supporting mental development. Hence, we propose soft computing techniques for teaching music to improve the mental health of college students under the background of 5G.

3. Proposed Work

The introduction of 5G will usher in a qualitative shift in music education. 5G will bring significant improvements to current network technologies in terms of bandwidth, service reliability, and device density compared to current network technologies. More specifically, we will concentrate our efforts on music education, a subject in which having a big bandwidth is critical for exchanging high-quality multimedia streams, and bidirectional communication latency should be kept to a minimum, in the millisecond range. Figure 1 shows the schematic representation of the proposed methodology.

[figure(s) omitted; refer to PDF]

3.1. Dataset Description

There are 1,458 music compositions in the ISMIR 2004 dataset, divided across training (50 percent) and testing (50 percent). It consists of six diverse styles of music: classical, electronic, jazz, punk, pop, and world. The training and test sets are preset, and the dispersion is not equal. Since there are no data regarding the performing artist of every musical piece, the artist filter could not be used with this database. Table 1 shows the count of samples per class in the ISMIR 2004 dataset [16].

Table 1

Dataset class and count of samples.

Class	Sample
Classical	639.9
Electronic	228.8
Jazz	52.0
Punk	90.1
Pop	202.9
World	244.0
Total	1,457.7

3.2. Frame Generation

The input music dataset is split into frames in this stage. The input dataset is raw in nature, and hence it is fed into the preprocessing stage for the removal of the noises in the dataset.

3.3. Preprocessing Using Normalization

The entering music dataset is unprocessed and may contain errors and noise. It has been cleaned and preprocessed to remove redundant and duplicate instances, as well as missing data. Because the datasets for the music education system are so large, sample size reduction techniques must be employed. Because this database contains so many features, feature extraction technologies are required to filter out the ones that are not significant. The database may be normalized during the preprocessing phase. In the first phase of the normalizing process, the $z$-score, which is given by equation (1), is obtained.$$\begin{array}{}\text{(1)}& Z=\frac{R-\alpha }{\omega },\end{array}$$where $\alpha$ denotes the mean of the data and $\omega$ indicates the standard deviation. Z is expressed as$$\begin{array}{}\text{(2)}& Z=\frac{R-\overline{R}}{SD},\end{array}$$where $\overline{R}$ denotes the mean of the sample and $SD$ denotes the standard deviation of the samples.

The random sample is in the form of$$\begin{array}{}\text{(3)}& Z_k={\beta }_0+{\beta }_1{R}_k+{\epsilon }_k,\end{array}$$where ${\epsilon }_k$ denotes the error that is dependent on ${\omega }^2$.

Following that, the errors must not rely on each other, as provided below.$$\begin{array}{}\text{(4)}& r_k\sim \sqrt{W}\frac{r}{\sqrt{r^2+w-1}},\end{array}$$where $r$ denotes a random variable.

Thereafter, the standard deviation is utilized to normalize the movements of the variable.

The below expression is utilized to estimate the moment scale deviation.$$\begin{array}{}\text{(5)}& MS=\frac{{\lambda }^{ms}}{{\varnothing }^{ms}},\end{array}$$where $ms$ denotes the moment scale.$$\begin{array}{}\text{(6)}& {\lambda }^{ms}=ER-\alpha \hat{M}S,\end{array}$$where $R$ indicates a random variable and $E$ denotes the expected value.$$\begin{array}{c}\text{(7)}& {\varnothing }^{ms}=ER-\alpha \hat{M}\hat{S}\frac{1}{2},r_w=\frac{ms}{\overline{R}},\end{array}$$where $r_w$ represents the coefficient of the variance.

The feature scaling procedure will be terminated by setting all of the variables to 0 or 1. The unison-based normalizing approach is the name for this procedure. The normalized equation would look like this:$$\begin{array}{}\text{(8)}& R^{\prime }=\frac{r-r_{\min }}{r\_\max -r\_\min }.\end{array}$$

The data could be kept after they have been normalized, and the range and irregularity of the data could be maintained. This phase’s purpose is to minimize or eliminate data delays. The normalized data could then be used as an input in subsequent steps.

3.4. Feature Extraction

This article employs multiple feature extraction techniques to extricate the timbre characteristics of the bottom music properties and the melody characteristics of the middle musical properties from the actual audio signals and later shall be comprised of such characteristics. The training set describes how to utilize the training classification method to enhance the categorization technique’s efficiency [17].

3.5. Mel Cepstral Coefficient

It is a mathematical model that simulates and adapts to the characteristics of human hearing. It features good anti-noise capabilities and a high detecting accuracy. It has become a commonly utilized characteristic factor in modern voice signal studies. To turn the time-specific signal into the frequency-specific signal, firstly construct the audio signal, do frame processing and windowing on it, and then employ the Fourier transform:$$\begin{array}{}\text{(9)}& xk={\displaystyle \sum }_{K=0}^{N-1}xn{e}^{-j2\pi nk/N},0\le k\le N-1.\end{array}$$

x(n), n = 0, 1, J, in which J denotes the length of the signal, represents the input signal, and x(n) represents the signal input strength at n. The number of points processed by the discrete Fourier transform is represented by N.

It is necessary to compute and then transfer the energy spectrum. The transfer method is implemented using a set of mel scale triangle filters. The center frequency, indicated as $fm,$ is an important characteristic of this type of filter. The group’s output energy is calculated and expressed in logarithm for each triangle filter, and then$$\begin{array}{}\text{(10)}& Sm=ln{\displaystyle \sum }_{k=1}^{N-1}{xk}^2{H}_{m^k},0\le m\le M-1.\end{array}$$

The second step is to apply a discrete cosine transform (DCT) to compute the mel cepstrum coefficient parameters:$$\begin{array}{}\text{(11)}& Cn={\displaystyle \sum }_{m=0}^{N-1}Sm\cos n\pi m-\frac{0.5}{m},n=\mathrm{1,2}\dots ,L.\end{array}$$

3.6. Pitch Frequency

An audio sequence is an audio signal that consists of several tones. The tone variation represents the composer’s feelings while writing this piece of music, and the tonal frequency determines the tone. Because the voice is represented by the pitch frequency, it is a critical signal processing parameter. The short-term reliability of the speaker signal must be taken into account while extracting pitch frequencies. Autocorrelation function (ACF), average amplitude difference (AMDF), and peak reduction are now the most used methods. The autocorrelation function identification technique is suggested to retrieve the pitch frequencies in this study due to the reliability and regularity of the pitch signal. The speech signal $sm$’s short-term autocorrelation function $R_{n}k$ is defined as$$\begin{array}{}\text{(12)}& R_{n}k={\displaystyle \sum }_{m=0}^{N-k-1}{S}_{n}m{S}_{n}m+k,\end{array}$$where $N$ denotes the length of the window included by the audio signals and $S_{n}m$ represents the segmented window audio signal interrupted by audio signal $sm$ through a window with window length $N$ and is termed as$$\begin{array}{}\text{(13)}& S_{n}m=smwn-m.\end{array}$$

The ACF of the basic component of the audio sample shall possess apparent peaks, whereas the high frequency tones will be less noticeable. As a result, checking for a clear peak can help determine whether it is a fundamental tone or a higher frequency tone, and looking for the distance between peaks can help determine pitch frequencies.

3.7. Resonance Peak

Formant is also known as resonance frequency. As a consequence of the auditory signal’s resonating mechanism, the energy stored in specific sound channels is increased. The auditory tracts could be conceived of as an audio tube having evenly spread vibrations, with the audio mechanism being the resonant of the vibrations in various places. The vocal tract’s structure is frequently related to the shape of the formant. The shape of the formant will alter as the framework of the auditory tract varies. Distinct emotions relate to diverse channel structures for a segment of speech stream. As a result, the formant frequency can be employed as a significant element in the identification of speech signal emotion.

3.8. Frequency Band Energy Distribution

The energy distribution of a segment of an auditory signal that includes details like the strength and frequencies of the auditory signal is referred to as band energy distribution. It is linked to music’s sweetness and emotions. The pleasantness and psychological features of an audio signal could be determined in the music world by analyzing the frequency band’s energy distribution features. Assume that there is a musical portion of length M that comprises the sound features of diverse devices and human vocals. Now we are looking for one of the subbands in the frequency domain of the musical portion, ranging from $a$ to $a+N$ that includes energy. Firstly, the actual time-domain musical signal $ft$ is transformed to a frequency-domain signal using the Fourier transform $Ft$.$$\begin{array}{}\text{(14)}& Ft={{\displaystyle \int }}^{\text{}}ft{e}^{-jwt}\text{dt}.\end{array}$$

The band energy $E$ is expressed as$$\begin{array}{}\text{(15)}& E=\frac{1}{N}{\displaystyle \sum }_a^{a+N}{Ft}^2.\end{array}$$

3.9. Classification Using Support Vector Machine (SVM)

The SVM [18], a directed organization method that determines the extreme boundary splitting two groups of information, was utilized to classify the music data. If the information is not directly distinguishable in the feature space during this time, it can be put into an upper dimensional space using the Mercer kernel approach. In fact, the internal outcomes of the information concentrated in this higher-dimensional space are crucial, and the projection can only be understood if such an interior item can be worked out clearly.

The set of potential classifier jobs in this kernel space is made up of bias direct configurations of essential preparation events. To strengthen the boundary between classifier boundary and training orders, the SVM training algorithm selects these weights and support vectors. Because training instances are used primarily for characterization, using complete tracks as samples helps to solve the problem of track taxonomy. Once we have an incomplete set of points in various dimensions, SVM can be employed in direct or indirect ways. SVM has the ability to find the linear separation that should exist. Hence, it tends to be true. SVM works well with outliers because it only uses the most closely linked points to find a valid separation.

3.10. Improved Fuzzy Neural Network

Music is used as a carrier in this paper’s model to enhance students’ mental health by means of musical instruction. The fuzzy neural networks underpin the music education system. The fuzzy neural network model has a single output and two inputs. The model network layer has been increased to five layers to enhance control precision. The first unit is the input unit, the second unit is the membership function unit, the third unit is the regular unit, the fourth unit is the double normalization unit, and the fifth unit is the output unit. The pressure error and pressure error rate of variation are represented by the architectural design of the fuzzy neural network, respectively (Figure 2).

[figure(s) omitted; refer to PDF]

The feed of the improved fuzzy neural network architecture is$$\begin{array}{c}\text{(16)}& A_1=pt,A_2=\mathrm{\Delta }pt=pt+\mathrm{\Delta }t-\frac{pt}{\mathrm{\Delta }t},\end{array}$$where $pt$ denotes the pressure error, $\mathrm{\Delta }pt$ represents the pressure variation rate, and $u$ indicates the last outcome. The input-output correlation among the units in the neural network shall vary dramatically if each output portion is supposed to have n membership levels.

The input unit of the system is the first layer. The pressure error $pt$ and the error rate $\mathrm{\Delta }pt$ are represented by the input quantity $A_1$ and the input quantity $A_2$ accordingly. This article’s input consists primarily of audio and information about the student’s mental state recognition.

Any node in the input unit has a matching input parameter, and the input value is transmitted directly to the further unit without any estimation by the output nodes. The following are the expressions of the input and output units in the initial unit.

The input unit is$$\begin{array}{}\text{(17)}& f_i^1=f_i,i=\mathrm{1,2.}\end{array}$$

The output layer is$$\begin{array}{}\text{(18)}& x_{i{j}^1}=f_{i^1},i=\mathrm{1,2};j=\mathrm{1,2},\dots ,25.\end{array}$$

The second unit acts as the input membership of the proposed system. In this unit, every node denotes a fuzzy quantity associated to it, and the Gaussian operation is employed as its membership function ${\gamma }_i^j$, and its equation is provided below. The following are the particular expressions of the input and output units in the language unit.

The Gaussian function is$$\begin{array}{}\text{(19)}& {\gamma }_i^j=p^{-1/a_{ij}{x}_{i-c_{ij}{}^2}}.\end{array}$$

The input unit is$$\begin{array}{}\text{(20)}& f_{ij}^2=-\frac{x_{ij-C_{ij}}^i}{{\sigma }_{ij}},i=\mathrm{1,2};j=\mathrm{1,2},\dots \mathrm{..},25.\end{array}$$

The output layer is$$\begin{array}{}\text{(21)}& x_{ij}^2=p^{f_{ij}^2},i=\mathrm{1,2};j=\mathrm{1,2},\dots \mathrm{..},25.\end{array}$$

In the above equations, the width of the Gaussian operation is expressed by ${\sigma }_{ij}$, and the center value of the Gaussian operation is expressed by $C_{ij}$. The third unit is the condition unit of the proposed system that acts as the interconnection between the preconditions and the subsequent unit inference. As every node has correlating fuzzy conditions, the fitness $\phi j$ of every correlating fuzzy condition is estimated by matching the fuzzy condition that is termed as the subsequent equation. As a result, the input and output layer specific expressions in the third unit are represented as follows.

The fitness of fuzzy conditions is$$\begin{array}{c}\text{(22)}& \phi ij={\gamma }_1^{i_1}{\gamma }_2^{i_2}\dots {\gamma }_n^{i_n},n=i_n=\mathrm{1,2},\mathrm{...},25.\end{array}$$

The input unit is$$\begin{array}{}\text{(23)}& f_j^3={\gamma }_1^{i_1}{\gamma }_2^{i_2}\dots {\gamma }_n^{i_n},j=\mathrm{1,2},\dots ,25;i_n=\mathrm{1,2},\dots .,25.\end{array}$$

The output unit is$$\begin{array}{}\text{(24)}& x_j^3=f_j^2,j=\mathrm{1,2},\dots ,25.\end{array}$$

The fuzzy neural network’s fourth unit acts as a subordinate unit. The entire count of nodes in this layer is $h$, and it performs a standardized estimation of the application of numerous conditions. The application of various rules is normalized using this layer. The following is the relation between input and output units.

The entire count of nodes is$$\begin{array}{}\text{(25)}& t=\frac{\propto j}{{{\displaystyle \sum }}_{i=1}^m\propto j},j=\mathrm{1,2},\dots 25.\end{array}$$

The input unit is$$\begin{array}{}\text{(26)}& f{j}^4=\frac{x{j}^3}{{{\displaystyle \sum }}_{i=1}^{m}x{i}^3},i=j=\mathrm{1,2},\dots 25.\end{array}$$

The output layer is$$\begin{array}{}\text{(27)}& x{i}^4=f{j}^3,i=\mathrm{1,2},\dots 25.\end{array}$$

The rule unit of the proposed system is the fifth unit. Each node function in the rule unit corresponds to its own node function, and every node operation contains a self-learning function that may be utilized to determine the outcome of every fuzzy rule. The following is the specific relation between input and output units.

The fuzzy conditions are$$\begin{array}{}\text{(28)}& y_{ij}={\displaystyle \sum }_{i=0}^{n}p{i}_{jl}{A}_l,j=\mathrm{1,2},\dots 25,l=\mathrm{1,2.}\end{array}$$

The input unit is$$\begin{array}{}\text{(29)}& f{i}^5=x{i}^4,i=\mathrm{1,2},\dots 25.\end{array}$$

The output layer is$$\begin{array}{}\text{(30)}& x{i}^5=hPi{A}_1+qi{A}_2+r_i,i=\mathrm{1,2},\dots 25.\end{array}$$

The output layer equation of (29) represents the initial order output of $T-S$ fuzzy control, and the corresponding linear system specifications are represented by $Pi{A}_{1}qi{A}_2{r}_i$, respectively. In this improved fuzzy network, the double normalization layer can help to remove the redundant data and increase the efficiency of the system.

4. Performance Analysis

The MATLAB simulation tool is used to analyze the entire system. Here, MATLAB version 9.0.0.341360 is used. The ideal variables for the proposed system are found using a genetic algorithm, then the optimized parameters are substituted into the proposed system, and the samples are trained.

Figure 3 depicts the output error convergence curve. Figure 4 shows the misclassification ratio of a fuzzy neural network tuned using a genetic method.

[figure(s) omitted; refer to PDF]

The output value of the proposed system enhanced by the genetic algorithm is better than the information before the enhancement, and the misclassification ratio is better, as shown in the above figure. The suggested system is modified with the optimal parameters, and the samples are trained. As a result, the fuzzy neural network improved with a genetic algorithm has greater control efficiency, rapid learning speed, and better convergence.

As demonstrated in Figure 5, the assessment of students’ mental health by the 5G methodology developed in this article is comparable to that of psychologists, implying that the design developed in this article has some influence. That is, music instruction has a positive impact on pupils’ psychological well-being. The next stage is to investigate the system’s stability. A total of 100 datasets are tested, and the number of system mistakes is tallied. The outputs are provided in Figure 6.

[figure(s) omitted; refer to PDF]

The system model developed in this research has great stability and shall suit actual requirements, as shown in Figure 6.

The behavior of the suggested system is compared with the existing methodologies in terms of accuracy, sensitivity, and precision as shown in Figure 7.

[figure(s) omitted; refer to PDF]

5. Conclusion

This paper proposed soft computing techniques for teaching music for studying students’ mental health under the background of 5G, based on existing fuzzy controller research. As a result, we suggest soft computing strategies for teaching music to promote the mental health of college students in this study, which is set against the backdrop of 5G. Furthermore, this paper uses soft computing techniques such as SVM classifiers to categorize the music dataset. In addition, this paper incorporates better fuzzy neural network regulation with music rehabilitation into a framework for assessing the mental health impact of students in the actual world. This study also builds design architecture depending on real-world requirements, enhances the system operation process algorithm, and rectifies faults. Following the construction of the design, this study designs experiments to test the model’s behavior in two areas: system reliability and system practical impacts. The statistical outcomes show that the system developed in this research has a specific impact and may be used in practice.

In this kernel space, the set of potential classifier tasks is made up of biased direct configurations of critical preparation occurrences.

References

[1] D. S. Rahardwika, E. H. Rachmawanto, C. A. Sari, A. Susanto, I. U. W. Mulyono, E. Z. Astuti, A. Fahmi, "Effect of feature selection on the accuracy of music genre classification using SVM classifier," .

[2] E. . Á. Campayo–Muñoz, A. Cabedo–Mas, "The role of emotional skills in music education," British Journal of Music Education, vol. 34 no. 3, pp. 243-258, 2017.

[3] S. Wang, M. Agius, "The use of music therapy in the treatment of mental illness and the enhancement of societal wellbeing," Psychiatria Danubina, vol. 30 no. 7, pp. 595-600, 2018.

[4] R. Geng, "Improvement of music teaching efficiency based on learner emotional model," Educational Sciences: Theory and Practice, vol. 18 no. 6, 2018.

[5] T. Zhang, S. Liu, "Evaluation of the effect of music education on improving students’ mental health based on intelligent fuzzy system," Journal of Intelligent and Fuzzy Systems, .

[6] S. J. Holochwost, C. B. Propper, D. P. Wolf, M. T. Willoughby, K. R. Fisher, J. Kolacz, V. V. Volpe, S. R. Jaffee, "Music education, academic achievement, and executive functions," Psychology of Aesthetics, Creativity, and the Arts, vol. 11 no. 2, pp. 147-166, DOI: 10.1037/aca0000112, 2017.

[7] B. Dobos, B. F. Piko, D. T. Kenny, "Music performance anxiety and its relationship with social phobia and dimensions of perfectionism," Research Studies in Music Education, vol. 41 no. 3, pp. 310-326, DOI: 10.1177/1321103x18804295, 2019.

[8] L. Plumb, T. Stickley, "Singing to promote mental health and well-being," Mental Health Practice, vol. 20 no. 8,DOI: 10.7748/mhp.2017.e1182, 2017.

[9] A. Sheppard, M. C. Broughton, "Promoting wellbeing and health through active participation in music and dance: a systematic review," International Journal of Qualitative Studies on Health and Well-Being, vol. 15 no. 1,DOI: 10.1080/17482631.2020.1732526, 2020.

[10] M. J. Silverman, "Music therapy and therapeutic alliance in adult mental health: a qualitative investigation," Journal of Music Therapy, vol. 56 no. 1, pp. 90-116, DOI: 10.1093/jmt/thy019, 2019.

[11] S. Saarikallio, "Access-awareness-agency (AAA) model of music-based social-emotional competence (MuSEC)," Music & Science, vol. 2,DOI: 10.1177/2059204318815421, 2019.

[12] S. Henderson, M. Cain, L. Istvandity, A. Lakhani, "The role of music participation in positive health and wellbeing outcomes for migrant populations: a systematic review," Psychology of Music, vol. 45 no. 4, pp. 459-478, DOI: 10.1177/0305735616665910, 2017.

[13] C. Hense, M. J. Silverman, K. S. McFerran, "Using the healthy-unhealthy uses of music scale as a single-session music therapy intervention on an acute youth mental health inpatient unit," Music Therapy Perspectives, vol. 36 no. 2, pp. 267-276, DOI: 10.1093/mtp/miy013, 2018.

[14] T. McCaffrey, "Evaluating music therapy in adult mental health services: tuning into service user perspectives," Nordic Journal of Music Therapy, vol. 27 no. 1, pp. 28-43, DOI: 10.1080/08098131.2017.1372510, 2018.

[15] T. Stegemann, M. Geretsegger, E. Phan Quoc, H. Riedl, M. Smetana, "Music therapy and other music-based interventions in pediatric health care: an overview," Medicines, vol. 6 no. 1,DOI: 10.3390/medicines6010025, 2019.

[16] Y. M. G. Costa, L. S. Oliveira, C. N. Silla, "An evaluation of convolutional neural networks for music classification using spectrograms," Applied Soft Computing, vol. 52, pp. 28-38, DOI: 10.1016/j.asoc.2016.12.024, 2017.

[17] K. Zhang, "Music style classification algorithm based on music feature extraction and deep neural network," Wireless Communications and Mobile Computing, vol. 2021,DOI: 10.1155/2021/9298654, 2021.

[18] M. A. Ali, Z. A. Siddiqui, "Automatic music genres classification using machine learning," International Journal of Advanced Computer Science and Applications, vol. 8 no. 8, pp. 337-344, 2017.