Eighteen patients with spinal cord injury, including complete spinal cord injury and incomplete spinal cord injury, who was hospitalized in CRRC from January 2021 to December 2021, were recruited. Inclusion criteria: (1) classified as spinal cord injury class A and class B by the American spinal cord injury association (ASIA) (American Spinal Cord Injury Association [ASIA]); (2) The course of the disease should have been at least 3 months (inclusive); (3) Aged between 18 and 75; (4) Moderate or above respiratory dysfunction and phonation disorder after cervical spinal cord injury, voice volume in normal conversation is lower than 40 decibels; (5) no tracheotomy or tracheotomy has healed; (6) Can tolerate seat training for more than 15 min at a time without postural hypotension; (7) No previous music learning experience; (8) native Mandarin Chinese; (9) Patients and their families were informed and consented to this study. Exclusion criteria: (1) a history of severe speech disorder, a history of mental disorder, or a history of severe respiratory disease before injury; (2) Severe cognitive dysfunction, MMSE score < 17 (illiteracy) or < 20 (primary school); (3) tracheotomy, vocal cord damage, or posterior damage; (4) epilepsy, malignant arrhythmia, or other serious physical diseases. Criteria for withdrawal and termination: The study can be promptly terminated if the patient’s condition changes or discharged or voluntarily quit the study. The general information of patients is as follows (Table 1).

Since this study was a randomized controlled trial, n = Z2⋅σ2/d2 was calculated according to the sample size formula, where n was the minimum sample size. Z is the confidence interval, usually 90%; Sigma is the standard deviation; D is the sampling error range; usually, 0.5 and the minimum sample size was thus calculated. Therefore, this study met the requirement of a minimum sample size (n ≈ 18) and was divided into two groups. Computer-generated sequences (Excel 2013, US, Washington, Seattle, Microsoft Office) randomly divided patients into two groups: vocal intonation therapy group (n = 9) and respiratory phonation training group (n = 9). This study was conducted from January 2021 in CRRC.

After gaining the Scientific Research Foundation of CRRC’s approval, participants diagnosed with ASIA (National Spinal Cord Injury Statistical Center [NSCISC], 2017) classification A and B with C4, C5 injuries, combined with vocal disorders, were firstly screened by clinical medical experts such as spinal cord injury specialists, spinal surgeons, and rehabilitation specialists according to the study inclusion criteria, applied for consultation and referred to the music therapy center. The music therapy researchers conducted the initial assessment and test of the vocal function of the referred patients, and the patients meeting the inclusion criteria will be included in the study to be randomly assigned, and those who fail to meet the criteria will be excluded. All patients and their families were informed and consented to this study. Participants included in the study were randomly divided into the intervention group as the vocal intonation therapy group (n = 9) and control group, respiratory phonation group (n = 9) according to the computer-generated sequence based on routine inpatient rehabilitation therapy. Patients in the intervention group were given vocal intonation therapy (VIT) by music therapists. The training process was 30 min of one-on-one VIT conducted by music therapists, 5 times per week for 12 weeks, 60 sessions in total. The control group was given respiratory phonation therapy by respiratory physiotherapists, 30 min a day, 5 times a week, for 12 weeks, 60 sessions in total. The enrollment and allocation of participants are shown in Figure 1.

Figure 1 illustrates that 30 participants were enrolled in the study, 12 participants withdrew from the study as they failed to meet the inclusion criteria (n = 10) dropped off the trial (n = 1); private reasons (n = 1). The intervention group was treated with VIT (total n = 9, ASIA A n = 6, ASIA B n = 3; C4 n = 5, C5 = 4); the control group underwent respiratory phonation training (total n = 9, ASIA A n = 5, ASIA B n = 4; C4 n = 6, C5 = 3). Among the 9 participants in the intervention group, all the participants were able to sing the trained items during the entirety of the study period. The data analysis included a sample of 18 CSCI patients. ASIA: American Spinal Injury Association; VIT: vocal intonation training; CSCI: cervical spinal cord injury.

Patients in the control group were given respiratory phonation training (RPT). RPT is a vocal training method whereby respiratory physiotherapists use physical therapy techniques to press the abdominal cavity with external force after the patient inhales to help them produce long sounds (Katz et al., 2022). Patients in the intervention group were treated by music therapy professionals using VIT techniques. The biggest difference between RPT and VIT is that RPT uses external force to press the patient’s abdominal cavity to guide the vocalization; while VIT is a method whereby patients make their voices autonomous under the guidance of music. In the intervention group, as well as the music therapy group, first, a vocalization assessment was performed, whereby the therapist determines the location of the difficulty based on the articulation of the patient’s voice. Following this, the preparatory exercises were then started. The specific intervention steps are: (1) Vocal warming. The therapist instructs the patient to practice monophonic vocalations starting at Andante speed (approximately = 72), using the combined inhalation pattern of chest and abdomen, drawing respiratory to the waist and abdominal diaphragm. (2) Vocal intonation. After repeated repetitions in the previous step, the therapist instructs the patient to practice monophonic vocalizations after a quick inhalation. For example, “mi-ma-mi,” “li-lu-li” and other melody lines. The therapist uses 4/4 to complete the training of guiding the patient. During the guidance process, the patient’s vocal quality was always scrutinized. This step is repeated 20 times with different combinations of vowels, consonants, and melodies in different tones. As the final step of the intervention, therapist used a song to reinforce the previous VIT practice session (Figure 2).

(3) Rhythmic induced intonation. It starts with an intensive rhythm of moderate ( = 96), switches between word and tone combinations, exercises with a faster rhythm, and improves diaphragm jumping. Finally, the lyrics are read aloud with the volume of natural speech to normalize the vocalization function of oral communication (Figure 3). Patients in the control group were trained by respiratory physiotherapists in the bedside posture.

The intervention group was trained with VIT, and the control group was trained with respiratory phonation therapy at doses as previously described. The other clinical indications remained identical in both groups. Before intervention (baseline, T0), during therapy (mid-test, 6 weeks later, T1), and after therapy (evaluation, 12 weeks later, T2), using computer-aided real-time audio analyzer 2.1.6 (Adobe Systems, United States) (Reimers and Stewart, 2016) for Sing-loud Pressure Level (Sing-SPL), Speak Pressure Level (Speech-SPL), Sustained note Length (SNL), and Fundamental Frequency (F0) assessment. The measurement results in the three-time points are analyzed and compared.

According to the International Standard Organization (ISO), the U.S. Environmental Protection of America (EPA) (Geneid et al., 2020), the Chinese National Acoustic Environmental Quality Standard (GB 3096-2008) (Molina et al., 2021), and the Social Life Environmental Noise Standard (GB 22337-2008), office ambient noise for sound acquisition are controlled to a control standard of less than 30 dBA (Molina et al., 2021). Data was collected from both sets using a computer voice test system at three-time points before the intervention (T0, baseline), 6 weeks after (T1), and 12 weeks (T2). Formulas were used to calculate the standard deviation of the mean and normal distribution for each group, and two-way ANOVA was used to analyze differences between groups, time effects, and differences in time interactions between groups. Data from 18 patients with vocal dysfunction following CSCI were analyzed using SPSS 22.0 (SPSS Statistical Software Inc., Chicago, Illinois, United States) who completed this part of the study. Before analysis, discrete data intervals were analyzed to screen for missing and outlier values, ensure the accuracy of data entry, and determine the specific effects of interventions. Sensitive and private information was blurred out, and all statistical data were kept confidential.

Data on vocal function were collected at three-time points, before the intervention (T0, baseline), 6 weeks after (T1), and after 12 weeks (T2). These include Sing-SPL (dBA), Speech-SPL (dBA), SNL (s), and sound frequency (F0). Multivariate ANOVA was used to analyze two groups of patients T0, T1, and T2 data. The results showed that in the vocal function test, the Sing-SPL of the intervention group at the T2 time point (T2 = 54.33 ± 11.30, p < 0.0001), Speech-SPL (T2 = 47.83 ± 11.30, p = 0.0029), SNL (T2 = 11.19 ± 3.25, p < 0.0001), sound frequency (F0, T2 = 305.89 ± 80.39, p < 0.0001) possessed significant differences between groups compared with the control group. (In the analysis and comparison of the Sing-SPL, Speech-SPL, SNL, F0, the intervention group patients had a very significant inter-group difference compared with the control group and had a significant difference in time effect). The results of the statistical analysis of the two sets of data are as follows (see Table 2 and Figure 4).

In the analysis of Sing-SPL results in the intervention group patients and control group patients, it was found that there was no significant difference in Sing-SPL values between the two groups of patients in the baseline test at the T0 time point, which indicated that the two groups of patients were evenly distributed and met the requirements of randomized controls. However, after the end of all interventions, at the T2 time point (after 12 weeks), there was a significant difference in the results of Sing-SPL acoustic analysis between the two groups (Figure 9).

In the analysis of The Speed-SPL results of patients in the intervention group and the control group, it was found that there was no significant difference in the Speed-SPL values of the two groups of patients in the baseline test at the T0 time point, which indicated that the two groups of patients were evenly distributed and met the requirements of the randomized control. However, after the end of the entire intervention, at the T2 time point (after 12 weeks), there was a significant difference in the acoustic analysis results of Speech-SPL between the two groups. Due to a large amount of data, a total of 6 patients with the most obvious changes in the intervention group and the control group were taken as an example, and the individual differences in Speech-SPL of the two groups were shown (Figure 10).

In this study, patients with CSCI were trained to sing short melodies to long melodic vowels and consonants using VIT step by step (vowel pronunciation function and consonant pronunciation function), focusing on the output of vocal function during singing. This was manifested by an increase in Sing-SPL (T1) time point after 6 weeks compared to the baseline (T0) time point results (Figure 5), at which point there was a statistical difference between the two groups of patients receiving different training methods. Among them, the vocal pressure level of the intervention group with CSCI who received VIT was significantly higher than that of the control group who received respiratory phonation training. Their vocal function was significantly improved in the accumulation of 6 weeks. In the second course of training that followed, the vocal function of the patients in the intervention group continued to improve with the guidance of effective training methods and the increase of training time. Their vocal function improved more obviously compared to the control group after 12 weeks (Figure 9). This shows that in the cumulative time effect of the same amount of training, the vocal function of patients with CSCI in different groups has obvious differences in intervention methods: the sound pressure value of the vocal function of the patients trained in VIT has been significantly increased, and the vocal function has been significantly improved.

In this study, patients used vocalization methods to adjust vocal habits when practicing speech function. After adapting to 4 weeks of 30 min/day of training, most patients showed good behavioral changes that resulted in significant changes in speech function, which was manifested by an increase in Speech-SPL after 6 weeks (T1) compared to the baseline (T0) results. There was a statistical difference between the two groups of patients who received different training methods (Figure 6). Among them, the Speech-SPL of the intervention group CSCI patients who received VIT was significantly higher than that of the control group, which showed that after cervical spinal cord injury, patients with vocal dysfunction were trained in VIT, and their vocal function was significantly improved in the accumulation of 6 weeks. There was a more obvious improvement after 12 weeks, and the difference was more obvious (Figure 10). This shows that in the cumulative time effect of the same amount of training, the vocal function of patients with CSCI in different groups possesses obvious differences in intervention methods: the sound pressure value of a vocal function of patients trained in music therapy has been significantly increased, and the vocal function has been significantly improved.

In the SNL test, the patients in the intervention group significantly increased the duration of the continuous tone after 12 weeks, indicating that when practicing singing short melodies (up and down the third chord, Figure 2) and long melodies (three-degree melodic intervals overlapping up and down, Figure 2), long-term sound output could be performed under the support of the thoracic and abdominal resonance cavity. Patients undergo VIT therapy in conjunction with adjustments to respiratory function to support the output of vocalization. After 6 weeks of 30 min/day of training, the patients in the intervention group improved SNL in terms of the length of music and the length of speech output. The SNL in the intervention group of patients receiving VIT (T1) was significantly longer than the control group (T1) trained in respiratory pronunciation, which indicates that patients with vocal dysfunction after CSCI were trained in vocal methods, and their SNL improvement was significantly improved after 6 weeks. In the second course of training that followed, the SNL (T2) of the patients in the intervention group continued to improve with the guidance of effective training methods and the increase of training time, and the difference was more conspicuous after 12 weeks (Figure 7). This shows that in the cumulative time effect of the same amount of training, there are obvious differences in the intervention methods of SNL in different groups of CSCI patients: patients who received VIT had significantly increased their pronunciation duration and improved significantly.

In the test of F0, the amplitude of the F0 of the patients in the intervention group increased significantly after 12 weeks, indicating that after VIT exercises and singing exercises, it was conducive to the vocal function output of high-density amplitude under the support of the thoracic and abdominal resonance cavity. Patients undergo VIT and singing training in conjunction with adjusting breathing function, while using more conserving breath to support vocal output. After adapting to 6 weeks of 30 min/day of training, the patients in the intervention group showed a change in the amplitude of sound frequency. Still, there was no significant statistical difference at that time. After continuing the training with the same amount of time accumulation for 12 weeks, the output of the sound frequency of the experimental group increased significantly. Specifically, the intervention group F0 significantly improved the accumulation of time at 12 weeks compared with the control group (Figure 8). After completing 12 weeks of therapeutic training, the F0 (T2) of the intervention group patients increased significantly with the guidance of vocal training methods and the increase in training time. This shows that in the cumulative time effect of the same amount of training, the vocal function of different groups of CSCI patients has obvious differences in the intervention method: the sound waveform amplitude of patients who received VIT vocal training and singing training was significantly enhanced, and the width of the sound frequency was significantly increased.

Clinicians can introduce songs gradually in order of difficulty (based on the length and pitch range of phrases). For example, a simpler song supplemental training “call-respond-chakra” singing format allows patients participating in a group practice to learn and practice together. During the first few weeks of the treatment process, the therapist can use songs with short phrases and songs with adequate breathing intervals. Actual clinical experience demonstrates that many patients with CSCI (or the general population with vocal dysfunction) initially have little confidence in their vocal function, so there is little emphasis on the quality of sound or pitch accuracy during actual VIT training. In the initial VIT singing training, the training method is mainly used to improve respiratory function and sound presentation and bring a pleasant emotional experience. Patients can sing with the accompaniment of an instrument or in karaoke style under background music recorded in the soundtrack to improve their vocal skills. According to the experience of clinical VIT plus singing training, the main recommended songs are (1) “Farewell”; (2) “Orchid Grass”; (3) “Hawthorn Tree”; (4) “Kangding Love Song”; (5) “Country Road”; (6) “The Country Road Takes Me Home”; (7) “Friendship lasts for a long time”; (8) “The Wind Blows the Wheat Waves”; (9) “Once Upon a Time”; (10) “Crooked Moon.”