Introduction Music is an effective medium for eliciting and regulating emotions and has been increasingly applied in therapeutic contexts. Yet the absence of standardized and validated music stimulus databases limits reproducibility and application in psychological and clinical research. This study aimed to develop a culturally inclusive therapeutic music database and to examine its affective validity and reliability. Methods A total of 234 participants rated 87 instrumental excerpts from Chinese and Western traditions, spanning classical, traditional, and popular genres, along six dimensions: valence, arousal, expressiveness, familiarity, liking, and perceived tempo. Results Descriptive analyses indicated moderate to high ratings across dimensions, and reliability testing confirmed strong internal consistency across repeated evaluations (test–retest rs = 0.74–0.89, p s < 0.001). Correlation analyses demonstrated a coherent internal structure among the six dimensions. Exploratory factor analysis further supported a unidimensional affective–perceptual factor (KMO = 0.75, p < 0.001), explaining 79.2% of the variance. Cluster analysis yielded three distinct categories: Positive–Energizing ( n = 27), Neutral–Relaxing ( n = 19), and Negative–Reflective ( n = 14), which aligned significantly with expert-defined classifications [χ 2 (4) = 55.9, p < 0.001, Cramér’s V = 0.57]. Discussion Based on these results, a final set of 60 validated excerpts was retained to form a standardized therapeutic music library. This resource offers a multidimensional, cross-culturally grounded, and empirically validated tool to advance emotion research, support cross-cultural comparisons, and guide the design of evidence-based music interventions in psychological and clinical practice.