Abstract: Objective: To evaluate the degree of bitterness of soybean protein hydrolysates and to find out a more suitable method of bitterness evaluation. Methods: Sensory evaluation and electronic tongue were used to analyze the bitterness of soybean protein hydrolysate. The contents of free amino acids and the molecular mass distribution of peptides were analyzed by high performance liquid chromatography (HPLC). Partial least squares regression (PLSR) was used to study the correlation between free amino acids, sensory bitterness intensity and electronic tongue response, and a prediction model for bitterness value was established. Results: The electronic tongue was superior to the sensory evaluation, and could clearly distinguish the difference in bitterness value among samples. Sample F1 had the strongest bitterness value, and its contents of bitter free amino acids and peptides with molecular mass less than 1 000 Da were higher than those of other samples. There was a good linear relationship between the sensory evaluation of samples with different concentrations and the electronic tongue evaluation, and the regression equation describing this relationship was obtained as follows: Y = 0.484X1 + 0.537X2 + 0.509X3 + 0.522X4, where Y is predicted sensory bitterness intensity, and X1, X2, X3 and X4 are electronic tongue sensor responses for bitterness, umami taste, richness and salt taste, respectively. The determination coefficients (R2) for the calibration and validation sets were 0.93 and 0.97, and the root mean square error of calibration (RMSEC) were 0.18 and 0.10, respectively, indicating that the model has good fitness and high reliability. Conclusion: Both sensory evaluation and electronic tongue model have high reliability. Electronic tongue evaluation can be used instead of sensory evaluation to evaluate the bitterness of plant protein hydrolysates. This study can provide a new evaluation method for further studies on the bitterness of plant protein hydrolysates.

Key words: bitter peptide; electronic tongue; sensory evaluation; partial least squares regression model; bitter taste evaluation