Study of the sensory profile of Coffea canephora through malting/fermentation using HS-SPME-GC‐MS and synthetic sampling combined with random forest

The sensory attributes of Coffea canephora beverages depend on the chemical composition of the bean, especially to the content of volatile organic compounds (VOCs). However, the relative abundance of these compounds may vary with the stage of bean maturation. This study investigated the VOCs responsible for the sensory attributes of Coffea canephora after malting and fermentation with Saccharomyces cerevisiae using HS-SPME-GC‐MS and random forest analysis with synthetic sampling. Ninety-four VOCs were identified, of which approximately 10 % contributed to discriminating the sensory profiles of the beverage. After 64 h of fermentation, malting treatments using fructose (T2), glucose (T3), and cellulase (T4) increased the sensory scores compared to natural coffee. The maceration/fermentation of coffee berries generated intense microbial activity, favoring the generation of VOCs. The machine learning methods proved efficient in identifying VOCs. Odor activity values demonstrated that the VOCs identified by this method were relevant to the sensory profile of coffee beverage.