Digitizing Wine Tasting with Cantilever Photoacoustic Spectroscopy and AI

Reproducible, Automated Aroma Analysis for the Future of Winemaking

Wine tasting has traditionally combined extensive physical and chemical testing with the trained senses of expert sommeliers. But this process can be time-consuming, subjective, and difficult to reproduce. Today, the Fraunhofer Institute for Integrated Circuits IIS is taking wine analysis into the digital era through the PINOT project, powered by METTLER TOLEDO’s automated analytical systems.

By integrating an autosampler, pH meter, and UV/Vis spectrophotometer with a cantilever enhanced photoacoustic spectroscopy (PAS) cell and advanced AI, researchers are streamlining the analysis of wine aromas, creating a reliable, objective, and fully automated alternative to human tasting.

Figure 1. The wine flavor wheel depicts the array of wine flavors and their origins, highlighting the complexity and subjective nature of the traditional wine tasting process.

The Challenge: Complexity and Subjectivity in Wine Tasting

Traditional wine tasting involves layers of evaluation: chemical testing, physical measurements, and sensory panels with trained sommeliers. While indispensable, these tastings are subjective and difficult to reproduce across different wines or vintages.

At the Fraunhofer IIS, the Digital Sensory Perception (DSP) group aims to overcome this by digitizing aroma perception. As part of the PINOT project, over 20 trained panelists at the Wine Institute in Neustadt an der Weinstrasse evaluated wines, while Fraunhofer measured the same bottles in the lab with their automated setup. The goal: to bridge human perception with precise analytical data for comprehensive aroma profiling.

Figure 2. The photoacoustic setup integrates an autosampler, UV/Vis spectrophotometer, and pH meter.

The Solution: Photoacoustic Spectroscopy Meets Automation

To recreate the tasting process in the lab, researchers measure the volatile aroma compounds that define wine character. Here’s how the setup works:

Sample preparation: An 80 mL autosampler cup is filled with 12 mL of wine, replicating the wine-to-air ratio of a standardized tasting glass.
Temperature control: The cup is sealed and tempered to 21 °C to accumulate aromas in the headspace.
Headspace extraction: The autosampler handler removes the lid and transfers the headspace using a tailor-made probe into the PAS cell.
Photoacoustic analysis: A laser excites the sample; the cantilever PAS device detects the volatile fingerprint.
Acidity and color: Acidity is measured by titration, while CIELAB color is analyzed by a UV/Vis spectrophotometer.
AI integration: Data is processed with an AI algorithm trained on a reference wine database, producing a full analytical report.
Experimental gas mixing station: An experimental gas mixing station is currently being used to validate the setup, ensuring reproducibility and reliability of the PAS-based aroma analysis.

The PAS + AI system offers a modern alternative to traditional wine evaluation, with significant advantages:

Precise control: Optimal tasting conditions at 21 °C can be replicated or adjusted for experimental studies.
Reproducibility: Automated workflows ensure consistency across wines, batches, and vintages.
Objectivity: AI removes subjective bias, enabling clear comparisons and documentation.
Flexibility: Modular design allows expansion with spectrometers or density meters for alcohol content measurement.
Efficiency: Automation and AI reduce the need for expert tasters, cutting costs while handling multiple samples simultaneously.

Figure 3. Dr. Thorsten Graunke, Chief Scientist Digital Sensory Perception Smart Sensing and Electronics Division.

Conclusion and Outlook

By digitizing wine tasting, the Fraunhofer IIS and METTLER TOLEDO are making wine analysis more accessible, reproducible, and objective. In the future, a miniaturized version of the system could even monitor fermentation and aging in real time, detecting critical compounds like hydrogen sulfide, acetic acid, and acetaldehyde early in production.

Beyond wine, this approach holds promise for other industries, from food spoilage detection to recycled materials odor analysis. By combining PAS with automation and AI, laboratories can replicate sensory evaluations with unprecedented precision.

We are currently running a big wine project, the PINOT project, where we try to support wine producers and sommeliers through a new type of aroma analysis. The wines have been evaluated by more than 20 panelists at the Wine Institute in Neustadt an der Weinstrasse. We measure the volatile substances of the same bottles in our laboratory and the resulting data is used for a comprehensive aroma analysis.

Dr. Thorsten Graunke, Chief Scientist Digital Sensory Perception Smart Sensing and Electronics Division at Fraunhofer Institute for Integrated Circuits IIS (Germany).

Anika Adams

Anika Adams is Product Manager for UV/Vis Spectrophotometers at METTLER TOLEDO, where she focuses on driving innovation and optimizing customer workflows.

With a degree in business administration and experience spanning software development, mechanical engineering, and production, Anika brings a multidisciplinary perspective to the laboratory business. Her expertise lies in automating and enhancing customer processes, ensuring both product excellence and customer satisfaction.