This study aimed to investigate abnormal tinnitus activity by evaluating brain surface-based amplitude of low-frequency fluctuation (ALFF) changes detected by resting-state functional magnetic resonance imaging (RS-fMRI) in patients with idiopathic tinnitus before and after 24 weeks of sound therapy. We hypothesized that sound therapy could gradually return cortical local brain function to a relatively normal range. In this prospective observational study, we recruited thirty-three tinnitus patients who had undergone 24 weeks of sound therapy and 26 matched healthy controls (HCs). For the two groups of subjects, we analyzed the spontaneous neural activity of tinnitus patients by cortical ALFF and detected its correlation with clinical indicators of tinnitus. Patients’ Tinnitus Handicap Inventory (THI) scores were assessed to determine the severity of their tinnitus before and after treatment. Two-way mixed model analysis of variance and Pearson’s correlation analysis were used in the statistical analysis. Student–Newman–Keuls tests were used in the post hoc analysis. Interaction effects between the two groups and between the two scans revealing local neural activity as assessed by ALFF were observed in the bilateral dorsal stream visual cortex (DSVC), bilateral posterior cingulate cortex (PCC), bilateral anterior cingulate and medial prefrontal cortex (ACC and MPC), left temporo-parieto-occipital junction (TPOJ), left orbital and polar frontal cortex (OPFC), left paracentral lobular and mid cingulate cortex (PCL and MCC), right insular and frontal opercular cortex (IFOC), and left early visual cortex (EVC). Importantly, local functional activity in the left TPOJ and right PCC in the patient group was significantly lower than that in the HCs at baseline and was increased to relatively normal levels after treatment. The 24-week sound therapy tinnitus group demonstrated significantly higher ALFF in the left TPOJ and right PCC than in the tinnitus baseline group. Also, compared with the HC baseline group and the 24-week HC group, the 24-week sound therapy tinnitus group demonstrated slightly lower or higher ALFF in the left TPOJ and right PCC, and there were no differences between the 24-week sound therapy tinnitus and HC groups. Decreased THI scores and ALFF changes in the abovementioned brain regions were not correlated. Taken together, surface-based RS-fMRI can provide more subtle local functional activity to explain the mechanism of tinnitus treatment, and long-term sound therapy had a normalizing effect on tinnitus patients.