Enhance the capabilities of artificial intelligence. Experts utilized this method to anticipate the scents emanating from whiskey and distinguish whether it was produced in the United States or Scotland.
This study is a crucial advancement towards a computerized system that can forecast the intricate aromas of whiskey based on its molecular makeup. Traditionally, expert panels evaluate woody, smoky, buttery, or caramel hints to ensure consistency in scent across different batches of the same product.
“One of the key advantages of AI is its unwavering consistency,” noted Dr. Andreas Glaskampf, the lead researcher at the Fraunhofer Institute for Process Engineering and Packaging in Freising, Germany.
“While trained professionals may introduce subjectivity, AI complements human olfaction by offering efficiency and consistency,” he continued.
Detecting the aroma of whiskey poses a significant challenge due to the complex chemical blends that form the dominant notes in spirits. These chemicals interact in the nose, often masking each other to produce a distinct aroma profile, making it hard to predict a whiskey’s scent from its chemical composition.
In the recent research, scientists collected data on the chemical composition and aroma of 16 American and Scottish whiskies, such as Jack Daniel’s, Maker’s Mark, Laphroaig, and Talisker, from an expert panel. This data trained an AI algorithm to predict the five primary aromas and origins of a drink based on its molecular components.
One algorithm showed over 90% accuracy in distinguishing between American and Scottish whiskies, excelling in identifying each drink’s five dominant notes compared to individual human experts. Further details can be found here.
Compounds like menthol and citronellol aided in detecting American whiskey’s caramel-like aroma, while methyl decanoate and heptanoic acid were essential in identifying Scotch’s smoky or medicinal notes.
Beyond whiskey, researchers are exploring applications in various fields, from spotting counterfeit products based on scent variations to innovating the recycling process for plastics with less noticeable odors in new products.
Dr. William Peverer, a senior lecturer in chemistry at the University of Glasgow, believes this AI-based approach could offer more reliable consistency compared to human taste panels, ensuring a stable flavor profile batch after batch.
While this study focused on a limited number of whiskies, the performance of AI with larger datasets and its adaptation to evolving flavor notes as the spirit ages in barrels remain to be seen. External factors like consumption environment may also influence flavor perception, requiring further research according to Dr. Peverer.
Source: www.theguardian.com
