This clinical study is a prospective, multicenter, randomized, controlled, double-blind clinical study. Sun Yat Sen Memorial Hospital of Sun Yat sen University was the central unit, and Nanjing First Hospital, Sun Yat Sen people's Hospital, Guanfzhou Panyu central Hospital and Zhuhai integrated traditional Chinese and Western medicine hospital were the cooperative units. In this study, 388 patients with chronic subjective tinnitus were recruited. In view of chronic subjective tinnitus, a common ear disease, the study gave the patients three months of treatment with transcutaneous vagus nerve stimulation paired with tailor-made notched music therapy or tailor-made notched music training alone. By comparing the changes of subjective scale scores related to tinnitus before and after treatment in patients receiving two different therapies, such as THI, VAS, BAI, BDI, PSQI, to evaluate the efficacy of the two therapies, so as to judge whether transcutaneous vagus nerve stimulation paired with tailor-made notched music therapy is better than tailor-made notched music training alone. In addition, the study will continue to follow up the patients after the treatment for one year to observe the difference in the long-term sustained efficacy of the two therapies. This clinical study will also evaluate the two therapies from the perspective of compliance and safety, and explore the factors that affect the efficacy of the two therapies.

Unified rules for filling in case report forms：

• All case report forms must be completed by designated and trained researchers.
• Two data managers independently input and proofread the data in duplicate.
• The data manager will modify, confirm and enter the data according to the scales written by participant.
• The locked data file will not be changed.
• Use medical terms / concepts to record AES . The type, degree, occurrence time, duration, treatment measures and treatment process shall be recorded in detail.

Sample size estimation:

• △ THI≥ 7 points in the comparison between groups indicates that the difference in efficacy has clinical significance, the superiority margin is established.
• The pre-test results showed that △ THI = -10.2 points, SD = 10.0 points in the two groups after the treatment.
• According to unilateral inspection standard α= 0.025, inspection efficacy 1- β= 0.80, using two sample t-test to estimate the sample size, the results show that at least 155 subjects are included in each group.
• Further estimation was made according to the loss of follow-up rate of 20%. Finally, at least 194 subjects (388 persons in total) should be included in each group.
• PASS 21.0.03 (NCSs, LLC, Kaysville, Utah, USA) is adopted for the above calculation.

Plan for missing data：

• If the subject uses the drugs, treatment or surgery prohibited in the protocol, the subject will no longer meet the conditions of the clinical study and will be deemed to have withdrawn from the study.
• If the event occurs before 3 months of treatment, the subject's data will no longer be used for statistical analysis of subsequent data.
• If the event occurs in the follow-up period after the completion of treatment, the data of the subject before and after treatment shall be retained. The data in the follow-up period shall be regarded as missing data and filled with corresponding methods for full analysis set analysis.
• For the missing data caused by the loss of follow-up, this study will use the method of multiple imputation to estimate the missing value.
• The research subjects who failed in screening will provide corresponding treatment according to their own conditions and clinical guidelines. These subjects will not be included in the clinical study.

Statistical analysis plan:

When considering the influence of baseline, the continuous variables were analyzed by covariance analysis, and the qualitative indicators were tested by CMH test or logistic regression.

Primary analysis： Using covariance analysis to compare the different changes of THI scores between two groups after 3-month treatment, controlled for age and baseline THI.

Secondary analysis： Using covariance analysis to compare the different changes of VAS, BAI, BDI, PSQI scores between two groups after 3-month treatment, controlled for age and baseline values corresponding to each scale.

Using a repeated measure ANOVA to compare the different changes of THI, VAS, BAI, BDI, PSQI between two groups at 3, 6 and 12 follow-up visits.

Using Chi square test or Fisher exact test to compare the different efficient rate between two groups after 3-month treatment.

Exploratory analysis:

Using multiple linear regression analysis to explore the factors affecting the short-term and long -term efficacy of the two treatments, such as age, hearing loss threshold, tinnitus loudness, tinnitus frequency and so on.

Using Independent two sample t-test or nonparametric analysis to compare the differences in EEG- or fMRI-related indicators between two groups.

Safety analysis:

Using Pearson's chi-square test to compare the difference of adverse event incidence rate between two groups.

Inclusion Criteria:

• Patients with tinnitus as the main complaint: the patients subjectively feel the sound in the ear or deep part of the head without internal or external sound stimulation, and therefore seek medical treatment
• Chronic (≥ 6 months) and tonal tinnitus
• Age 18-60
• Tinnitus frequency is 125-8000 Hz
• Hearing threshold at each frequency shall not exceed 70 dB HL

Exclusion Criteria:

• Patients with conductive deafness, history of middle ear surgery, pulsatile tinnitus caused by vascular distortion and tinnitus caused by Meniere disease
• History of head trauma, central nervous system diseases, mental diseases and drug abuse

In the whole clinical research process, the subjects and researchers participating in the efficacy and safety evaluation should be in a blind state, that is, neither of them knows the specific intervention measures given to the subjects. For participant, the treatment equipment is exactly the same between the two groups. Also, 10 seconds sham stimulation at the beginning of treatment is set in the control group for minimizing risk of participants being able to guess treatment allocation. The randomized allocation is completed by a unified randomization center. Therefore, the investigators in each center just know "A" or "B" for the allocation information of each participant. For outcomes assessors, they are only responsible for baseline and follow-up assessment during the study.