Application of electronic tongue for distinguishing coffee samples and predicting sensory attributes

Abstract

Efforts have been made to predict the sensory profile of coffee samples by instrumental measurement results. The objective of the work was to evaluate the most important sensory attributes of coffee samples prepared from ground roasted coffee by electronic tongue and by sensory panel. Further aim was to predict the Arabica concentration and the main sensory attributes of the different coffee blends by electronic tongue and to analyze the sensitivity of the electronic tongue to the detection of poor quality coffee samples. Five coffee blends with known Arabica and Robusta concentration ratio, five commercially available coffee blends and a poor quality coffee were analyzed. The electronic tongue distinguished the coffee samples according to the Arabica and Robusta content. The sensory panel was able to discriminate the samples based on global aroma, bitterness and coffee aroma intensity (p < 0.01). The Arabica concentration was predicted from the electronic tongue results by PLS with close correlation and low prediction error. Models were developed to predict sensory attributes of the tested coffee samples from the results obtained by the electronic instrument.