2025 SMSI Bannerklein

P49 - From Whisky to Aroma Investigating mixture data for odor prediction

Event
16. Dresdner Sensor-Symposium 2022
2022-12-05 - 2022-12-07
Dresden
Band
Poster
Chapter
Smart Sensors/Edge Computing/Künstliche Intelligenz in der Sensorik
Author(s)
S. Singh, A. Strube, A. Grasskamp, H. Haugh - Fraunhofer Institute for Process Engineering and Packaging IVV, Freising/D, S. Saloman, T. Scholz, B. Saha, S. Hettenkofer - Fraunhofer Institute for Integrated Circuits IIS, Erlangen/D, T. Gorges - Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen/D
Pages
224 - 229
DOI
10.5162/16dss2022/P49
ISBN
978-3-9819376-7-1
Price
free

Abstract

Hedonic impression of food products largely depends on consistency, taste, and smell. While process control over other factors is quite feasible, perception of smell is not something easily predictable or measurable as volatile organic compounds (VOCs) are immensely diverse in odorant quality and detection threshold. Furthermore, investigating key aroma compounds in food products is costly in time and effort. Therefore, it is desirable to develop a system for the efficient and reliable interpretation of the decisive aroma components that influence consumer satisfaction. Recent years have seen considerable progress in computer aided analysis of molecules for different purposes. To avail ourselves of this progress, we utilized machine learning methods to predict the aroma qualities of whisky spirits, which are classically valued for their diverse aroma profiles depending on process parameters like aging duration, cask origin and blending.

Download