The systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [11]. Articles search of the five databases, e.g., Scopus, ScienceDirect, Web of Knowledge, CINAHL, and PubMed, was done systematically in October 2019. The articles that been included in this study have been published from year 1980 to 2019. Table 1 provides full electronic search strategies with a string of keywords.

The inclusion criteria for studies in the review were specified in terms of participants, interventions, comparisons, outcomes, and study design (PICOS) as follows [11]. Studies were included in the systematic review if they (1) involved individuals (children and adults) with hearing loss fitted with hearing aids or cochlear implant unilaterally or bilaterally, (2) included participants who underwent music training as rehabilitation, (3) compared preand post-rehabilitation effect or repeated measures (experiments with additional purposes), (4) incorporated outcome measure(s) related to speech perception/intelligibility, auditory perception, musical perception, or communication improvement, and (5) integrated study design of randomized controlled trials, non-randomized controlled trials, cohort studies, and repeated measures (experiments with additional purposes) to report the results of pre- and post-training.

Titles and abstracts of the articles were screened according to the selection criteria and identified for preliminary articles as inclusion. Additional information was identified manually by two independent authors (NFAS and WH) who also checked any relevant articles that may not have been returned by the initial database search.

Two authors independently extracted data from each study following PICOS criteria [11]. Both children and adult participants who had hearing loss and have been fitted with either hearing aids or cochlear implants were included as the participants. In the intervention, all studies included the music training sorted by the stimuli used, frequency, duration of the training, and study settings of music training. All outcome measures related to musical perception after conducting the music training were collected. For comparison, we used pre- and post-training results to study the effectiveness of the music training. No restrictions were specified in terms of the duration of rehabilitation and follow-up. Finally, primary outcomes included one or more of the following: (1) detection of music perception including pitch, rhythm, frequency discrimination, and melody; (2) improvement in music perception including pitch, rhythm, frequency discrimination, and melody; (3) improvement in melodic contour identification that indicates tone recognition and speech perception; (4) improvement in the perception of recognition of musical instruments. Secondary outcomes comprised the improvement in musical performance, i.e., singing and music appreciation.

The methodological quality and risk of bias for the included studies were assessed using the standardized Physiotherapy Evidence Database (PEDro) scale. This scale’s items assessed the (1) specific eligibility criteria, (2) randomization of the allocation of the subjects, (3) concealed allocation of subjects, (4) pretherapeutic interventions baseline, (5) blinding of all subjects, (6) blinding of therapists, (7) blinding of assessors who measured key outcomes of the study, (8) measure of at least one key outcome form 85% of the recruited subjects, (9) intention to treat analysis (mentions that all subjects received treatment or control conditions as allocated, (10) statistical comparison between groups for at least one key outcome, and (11) point measure for size of treatment effect and variability measure for at least one key outcome.

One point was added to the studies only if a criterion was clearly stated on literal reading. The point would not be added if the criteria were missing or not clearly stated. For criteria 4 and 7 through 11, key outcomes provided the primary measure of the effectiveness or lack of the effectiveness of the therapy. Based on the suggestion of Moseley et al. [12], studies scoring 9–10 on the PEDro scale were considered methodologically to be of “excellent” quality. Scores ranging from 6 to 8 were considered “good” quality, while studies scoring 4 or 5 were of “fair” quality, and studies scoring below 4 were considered “poor” quality.

Data analysis was run using Comprehensive Meta-Analysis ver. 3 (Biostat Inc., Englewood, NJ, USA). Meta-analysis was performed using all studies that contained musical perception data having pre- and post-music training, appropriate outcome measures, and intervention. Means and standard deviation of preand post-training and also their correlation were used to calculate effect size of the music training.

Since most studies employed different measures outcome for musical perception, the effect sizes were calculated as standardized mean difference (SMD) in which it is necessary to standardize the result of all the studies to uniform scale before they can be combined; thus SMD expresses the size of the intervention effect in each study relative to the variability observed in that study [13]. All effect sizes were pooled using a random-effects model with 95% confidence intervals (CIs) [14]. A mixed-effects Q test for between-subgroup analysis of variance was used to compare the effects of four subgroups [14]. The funnel plot and Egger’s regression asymmetry test were used to assess publication bias [15].

Cochrane’s Q and I² values were calculated to test for homogeneity of variance across the studies. The Q value represented the total amount of variance among the set of studies. I² was calculated using the formula, I² =100%·(Q–df)/Q. According to Higgins and Altman [13], it provided a precise and easily interpreted measure of heterogeneity. I² values of 25%, 50%, and 75% represented low, medium, and high heterogeneity, respectively. A significant Q value indicated that the data were heterogeneous.

Ten studies involving 186 participants met the PICOS inclusion criteria. The participants consisted of both adult and pediatric patients with hearing loss who had been fitted with hearing aids and cochlear implants. The age of the 101 adult participants who had consistently used hearing aids before post-lingual cochlear implantation [16-19] ranged from 18 to 88 years old, and the age of the 85 children ranged from 1 to 15 years old [20-25]. Among the participants, 89 were female and 76 were male; however, the sex of 21 participants in two studies by Fuller et al. [16] and Yucel et al. [23] was not specified.

The music training programs used in these studies consisted of various kinds of stimuli and musical programs. Fu et al. [22] and Yucel et al. [23] used musical tones as stimuli, while the training stimuli in two other studies consisted of musical instrument tones at low, mid, and high frequencies [17,24]. Three studies directly measured cochlear implant users’ melodic perception using a melodic contour identification task [16,18,25], and three other studies conducted self-developed music training to enhance perceptual detection and to evaluate its improvement [19-21].

Most studies had a training period of 5 weeks to 3 months [16-18,22,25], while two studies [20,21] utilized training that lasted 3 months to 11 months, which were classified as an intermediate duration, and other two studies conducted training with a long duration (more than 12 months) [23,24]. Five studies were conducted in home settings [17,18,21-23] and the other studies implemented the training in rehabilitation centers [10,16,19,24,25]. The characteristics of the 10 studies are summarized in Table 2.

The outcomes of each study are also summarized in Table 2. In the pooled analysis (Fig. 2A), participants’ musical perception was significantly higher after music training (SMD=2.092, 95% CI, 1.333–2.850, P<0.001). Although a funnel plot showed that the data were asymmetrical (Fig. 2B), the Egger regression test detected no publication bias in the studies (intercept=2.313; standard error, 1.376; P=0.1312). Due to high heterogeneity (I²=86.57), we conducted a subgroup analysis.

Four subgroup analyses were conducted to investigate the effects of age, the hearing device used, the participants’ musical experience before the training, and music training duration on improvements in musical perception. Table 3 presents the effect sizes for subgroups, 95% CIs, and heterogeneity.

Table 4 shows the quality scores for each question of the PEDro. Two authors independently analyzed the scores and then verified them on the official PEDro website. The overall mean PEDro score was 6.5. The quality of the articles ranged from fair to good, with the quality of two studies being assessed as fair [18,24] and the quality of the remaining studies classified as good [16,17,19-23,25] due to a lack of information about the random allocation of subjects, the blinding process, and whether the key outcome were analyzed by “intention-to-treat” or not. None of the studies provided information about the blinding of the therapists involved in therapy sessions (i.e., whether they were unable to distinguish between the treatments applied to different groups) or about the subject concealment.

Music therapy can address several objectives of auditory training. Previous studies have suggested that music training can improve perception, localization, differentiation, and recognition of sound and attention towards the sound [5]. Furthermore, appropriate musical input is more effectively heard and assimilated than speech; thus, it is more likely to stimulate a natural motivation to use residual hearing [6]. The present study evaluated the efficacy of music training on musical perception among individuals with hearing loss who had been fitted with either hearing aids, cochlear implants, or both. The synthesis of data from these 10 studies suggested that significant improvements in musical perception were achieved in these individuals. Significant differences emerged according to age (between adults and children), the type of hearing device used, and the duration of training. However, nonsignificant differences in terms of improvements in musical perception were observed between participants with musical experience and those with no musical experience before starting music training.

A significantly different effect of music training was found between children (below 18 years of age) and adults (18 years of age and older). The effect size for children was significantly larger than that for adults, suggesting that music training may provide greater benefits for children than for adults. From the perspective of neuroplasticity, it is logical that pediatric users of hearing aids and cochlear implants may benefit more from music training than adult users [6]. Chronological age has long been linked to neuroplasticity, with greater neuroplasticity associated with younger age and/or immaturity [26]. The capacity for synaptic plasticity, with consequences for learning and memory, is not constant throughout the lifespan and typically declines with age at variable rates [27]. It peaks relatively soon after birth, with some research indicating that infants’ brain plasticity is about two times higher than that of adults. The effects of music training in children might therefore be stronger due to their greater brain plasticity. A clear differentiation of the effects of musical therapy across different age groups would help in the development and implementation of programs according to patients’ needs. In addition to age, however, it would have been desirable to consider in this study whether individuals were affected by pre-lingual or post-lingual deafness. In previous studies, significantly different therapeutic effects have been found for these two groups [26]. However, due to limitations in the data that could be extracted, it was not possible to clearly subcategorize the participants into pre-lingual and post-lingual deafness groups. Future research comparing differences between individuals with pre-lingual and post-lingual deafness through high-quality randomized controlled trials is needed to confirm the therapeutic effectiveness of aural rehabilitation in these two groups.

The usage of different hearing assistive devices is known to be linked to differences in speech recognition performance by hearing-impaired individuals. Research by Gfeller et al. [28] found that bilateral cochlear implants provided a positive impact on the recognition of music with lyrics, whereas bi-modal users who were fitted with hearing aids and cochlear implants showed better perception and enjoyment of instrumental music. This finding provides natural support for the possibility that the perception of music is likely to be meaningfully improved by combining acoustic and electric stimulation.

Regardless, the current study found that the trainees who used only cochlear implants showed greater improvements in musical perception after music training than those who used both cochlear implants and hearing aids. One possible explanation for this seemingly contradictory finding is that most trainees who were only fitted with cochlear implants in this study were young children. As discussed above, the effects of music training differed between children and adults due to neuroplasticity. Therefore, in further research, music training should be applied to carefully differentiated subgroups among participants of the same age depending on the mode and type of devices.

A nonsignificant effect of previous musical experience on trainees’ musical perception was found. However, the authors suggest that it may not be possible to draw a definitive conclusion for this subgroup because of the high level of heterogeneity that was present among the relevant studies. For example, the studies did not provide clear and specific information on how long and how often the trainee had musical experience and how intensive their experience was. Furthermore, the effects of music on the brain have been previously investigated by several researchers who compared the brain structure of musicians and non-musicians. For instance, in the study conducted by George and Coch [29], musicians demonstrated faster updating of auditory and visual working memory representations and more efficiently drew upon working memory resources to process deviant auditory stimuli than non-musicians with no musical experience.

The relationship between previous musical experience and the results of music training still need to be further explored through high-quality evidence-based studies, such as randomized controlled trials with a larger number of participants, as it remains possible that musical experience may have a significant effect on the outcomes of music training.

We analyzed the effectiveness of music training in terms of the training period. From the pooled results in this subgroup analysis, short and long training durations had a significant positive effect on improving participants’ musical perception, whereas a nonsignificant effect was observed for the intermediate duration. From the perspective of audiological practice, we suggest a long training duration (e.g., 12 months or longer) to optimize the effectiveness of rehabilitation programs for hearing-impaired individuals. The effects of various training durations on the effectiveness of music training are expected to serve as a basis for the development of further music training programs to improve the auditory perception of hearing-impaired individuals with hearing devices.

The studies included in this review varied in terms of the severity of the disease, the type and duration of intervention, and the quality of the methodology; therefore, caution is needed when interpreting our findings. The outcome measure of musical perception was used to assess the efficacy of music training among individuals with hearing loss. Participants’ listening abilities, duration of the training, and the content of training varied considerably across studies. This not only resulted in a limited ability to make direct comparisons but may also explain the inconsistent pattern of results that was observed across studies. Furthermore, because of the limited number of suitable articles that have been published, the number of participants was rather small and may not represent the entire population. Moreover, we only included studies that were published in English. In light of the small number of trials and participants and heterogeneity among studies, we still do not have enough evidence to confidently conclude whether there is a true subgroup effect for the type of device and musical experience. It is, therefore, useful to consider the plausibility of the demonstrated subgroup effects for these two groups. High heterogeneity is expected due to differences across studies, but it can nonetheless be useful to present the pooled data from comparable studies.

The findings of this systematic review regarding the effectiveness of music training in terms of improvements in musical perception can provide further directions for research in this field. We believe that the effectiveness of music training for the improvement of musical perception may be linked to the effectiveness of music training in aural rehabilitation and speech-language development because speech and music share many common properties. The auditory perception of speech and music involves the ability to distinguish between different sounds, their pitches, duration, intensities, and timbres, and their changes over time [6]. These properties enhance the listener’s ability to interpret sounds and attach meaning to them. Furthermore, these commonalities between music and speech allow music and music therapy to provide an alternative and pleasurable tool to enhance traditional auditory training techniques. Specifically, based on the studies reviewed herein, patients’ discrimination levels significantly improved through music training, and the practice that patients received in the setting of music generalized to environmental sounds.

Overall, music training can be implemented as an aural rehabilitation approach, as it effectively improves aural perception and musical perception in patients with hearing loss. In addition, it is important to consider that the effects of music training differ by the age of the trainee, the type of hearing device used, and the duration of music training. The finding of this study can serve as a reference for clinicians, patients, and health policymakers regarding the application of music training in clinical or rehabilitation programs. However, further high-quality randomized controlled trials are needed to confirm the effectiveness of aural rehabilitation in hearing-impaired patients.