Effect of key flavor compounds in fermented soymilk on sensory attributes: Integrating electronic sensory technology with GC–MS analysis

-nose) and HS-SPME-GC-MS with odor activity value (OAV) analysis to characterize flavor profiles of fermented mixed soymilk. The results showed that radar fingerprint profiles of the electronic nose combined with orthogonal partial least squares-discriminant analysis (OPLS-DA) can effectively distinguish the overall flavor profiles among samples. GC-MS identified and quantified 48 volatile compounds, with 35, 32, and 32 detected in samples A, B, and C, respectively. Thirteen key flavor compounds (OAV >1) were screened out, including aroma-enhancing substances such as 1-octanol, 2,3-butanedione, and acetoin (OAV >150), as well as off-flavor contributors like hexanoic acid, 1-hexanol, and decanal. Correlation network analysis indicated that concentration variations of these compounds directly drove the divergence in sensory attributes. The findings highlight the efficacy of GC-MS combined with electronic sensory technologies in evaluating flavor quality and provide valuable insights for characterizing and optimizing fermented soymilk flavor profiles.