Instrumental measurement of wine sensory descriptors using 1 a voltammetric electronic tongue

11 12 Abstract 13 The approach presented herein reports the application of a voltammetric electronic 14 tongue (ET), in contrast with a wine tasting sensory panel, as a tool for standardized 15 wine tasting; concretely, to achieve the discrimination of different wine DOs 16 (Denominación de Origen, a mark related to its geographical region and ensuring 17 specific quality levels) and the prediction of the global score assigned by the trained 18 sensory panel. To this aim, a voltammetric array of sensors based on bulk-modified 19 graphite and metallic electrodes was used as the sensing part, while chemometric tools 20 such as linear discriminant analysis (LDA) and artificial neural networks (ANNs) were 21 used as the qualitative and quantitative modelling tools. Departure information was the 22 set of voltammograms, which were first preprocessed employing fast Fourier transform, 23 followed by removal of less-significant coefficients employing a stepwise inclusion 24 method and pruning of the inputs. The trend, in global scores, was modelled 25 successfully with a 92.9% of correct identification for the qualitative application, and a 26 correlation coefficient of 0.830 for the quantitative one (with 14 and 20 samples for the 27 external test subset, respectively). 28 29