Wine has always depended on the human nose. In Champagne, however, researchers are now exploring what might happen when sensory observation is supplemented by a very different kind of nose: an electronic one capable of recognising volatile signatures too elusive or inconsistent for conventional tasting.
Developed through a collaboration between Moët Hennessy’s Robert-Jean de Vogüé research centre in Épernay, the American semiconductor company Analog Devices and the University of California, Davis, the Champagne electronic nose began with a specific problem. Researchers wanted a reliable way to identify an intermittent aroma defect associated with fresh mushrooms—one that may be impossible to detect in grapes and can behave unpredictably once wine has been made.
Yet the significance of the project extends beyond one troublesome smell. By studying the volatile compounds released by biological systems, the technology could eventually provide growers and winemakers with another way of observing what is happening in grapes, musts, vineyards and fermentations.
It is an ambitious proposition, but also a revealing one. As viticulture confronts climatic conditions that can create unfamiliar biological pressures, better detection may become almost as important as better treatment.
Why the Champagne electronic nose was developed
The story begins not with artificial intelligence, but with a wine fault.
Researchers in Champagne have been studying what they describe as fresh-mushroom aromas: odours reminiscent of cultivated mushrooms that become apparent after vinification and can interfere with the aromatic profile of a wine.
The difficulty is not simply their presence. It is their unpredictability.
According to the research team, the grapes themselves may show no obvious sensory warning. Even after vinification, the aroma can seem conspicuous during one tasting and far less apparent during another. That makes conventional sensory detection an uncertain basis for managing affected lots.
The phenomenon has gained particular attention in Champagne during the past decade. Researchers associate its emergence with changing climatic conditions and with vintages in which rot pressure becomes important.
That connection led the team towards microbiology.
Research involving two doctoral projects examined both the molecules responsible for the aroma and the microorganisms involved in producing them. Work with Institut Agro Montpellier, INRAE and Comité Champagne helped the researchers characterise the chemical side of the problem, while microbiological investigation pointed towards an interaction involving Botrytis and Penicillium moulds.
The picture that emerged was not of a single organism or isolated compound acting alone, but of a biological interaction capable of producing unwanted odorant molecules under certain conditions.
For Champagne, that distinction matters. If the defect is the result of biological processes beginning before the wine reveals the problem sensorially, waiting until tasting provides confirmation may already be too late.
The more useful question becomes: can it be recognised earlier?
From human perception to an electronic fingerprint
This is where the electronic nose enters the story.
Moët Hennessy’s researchers were already working with Analog Devices when the company introduced them to sensor technology originating at MIT. Its potential was immediately relevant to the Champagne project: instead of attempting to identify the problem through conventional smelling alone, the device could analyse a pattern of volatile compounds and search for a recognisable signature.
An electronic nose does not smell as a human does. It does not experience mushroom, citrus, toast or minerality. Nor does it possess the associative memory that allows a skilled taster to connect an aroma with thousands of previous wines.
Its strength is different.
Biological activity releases volatile compounds. Taken together, those compounds can create patterns. A sufficiently sensitive sensor system can record those patterns, while computational analysis can distinguish one from another.
The technology being tested combines a small electronic chip equipped with thousands of sensing elements with onboard data handling. An artificial-intelligence layer is then used to interpret the resulting signals and turn them into something useful for researchers or operators. Analog Devices contributes expertise in electronics and algorithms, while the Champagne research centre supplies the viticultural, oenological and experimental knowledge needed to train the system.
This distinction is crucial. The machine does not begin by knowing what a problematic sample looks like. It has to learn from examples.
And Champagne happens to possess an unusually valuable archive for teaching it.
A library of old musts becomes a research asset
Long before the electronic nose project began, the Épernay research centre had established a collection of musts from previous harvests.
The practice was systematised during the mid-2000s, with must samples being isolated from 2017 onwards. Preserved material allows researchers to return to particular vintages when a new scientific question arises.
For machine learning, that historical collection has acquired another purpose.
Training a detection system requires numerous examples representing different conditions. Samples from previous harvests can therefore provide the electronic nose with a library against which volatile signatures can be studied and compared.
In effect, an archive created for conventional wine research has become training material for a new generation of analytical tools. Two electronic-nose units are already operating at the research centre, although the equipment has not yet been commercialised for wine-industry use.
For wine research, the combination is particularly powerful: preserved biological material provides the history, sensors provide measurement, and algorithms provide a way of searching for patterns within it.
How the Champagne electronic nose could help at harvest
Detection becomes especially valuable when a defect can spread through blending.
If a problematic must cannot be distinguished reliably from a healthy one, separate lots may be combined before anyone realises that something is wrong. Where the sensory threshold of an unwanted compound is very low, even a relatively small affected component may compromise a much larger volume.
An early-warning system could change the sequence of decisions.
Instead of discovering a problem only after fermentation, growers or winemakers might eventually be able to analyse samples earlier, separate suspect material and decide how individual lots should be handled.
The prototype itself has been conceived with practical deployment in mind. The research team describes an instrument roughly the size of a shoebox, using a small sample container fitted with the sensing technology. Both liquids and solids can be examined without the elaborate sample preparation associated with some laboratory analyses.
That portability is arguably as important as sensitivity.
Viticultural technology has limited value if every question requires samples to be sent away to a specialist laboratory and results arrive after operational decisions have already been made. A compact instrument potentially changes where analysis can happen—from the laboratory towards the winery and, eventually, the vineyard itself.
It does not solve the biological problem. Detection is not treatment.
But knowing that a problem exists, and knowing it before different lots are combined, can give a winemaker something equally fundamental: the possibility of acting before the fault becomes harder to manage.
Climate change is creating unfamiliar wine risks
The project also illustrates a less discussed consequence of climate change.
Much of the wine world’s conversation about warming concentrates on harvest dates, sugar accumulation, acidity and drought. Those are fundamental concerns, but changing weather patterns also influence microbiology.
Different combinations of humidity, heat, rainfall and vine development can alter the conditions in which fungi and other microorganisms thrive. Problems that were historically rare enough to attract little attention can become more visible when climatic circumstances begin to favour them.
That appears to be part of the concern surrounding fresh-mushroom aromas in Champagne.
Research of this kind therefore represents a form of climate adaptation that is less visible than changing grape varieties or modifying vineyard architecture. It is adaptation through observation: developing tools capable of recognising biological changes early enough for growers to respond.
The better a vineyard or winery can be measured, the less dependent decision-making becomes on discovering problems only after they have become obvious.
UC Davis broadens the testing ground
The involvement of UC Davis brings another dimension to the project.
Working with the Californian university allows the technology to be tested on American grapes and musts while avoiding many of the practical and regulatory complications involved in transporting biological material across the Atlantic.
More importantly, it opens the system to different viticultural problems.
One application investigated at UC Davis concerns smoke-related aromas associated with wildfires. These present a challenge with similarities to the Champagne mushroom problem: grapes can appear relatively normal while undesirable sensory consequences become much clearer only later in the winemaking process.
For growers in regions increasingly exposed to wildfire smoke, earlier detection could be particularly useful.
The broader lesson is that an electronic nose does not have to be designed around a single defect. Once trained to recognise different volatile patterns, the same underlying approach could potentially be adapted to multiple biological or aromatic questions.
Could electronic noses detect vine disease?
The possibilities become more speculative once the technology moves beyond known wine faults.
Researchers envisage potential applications involving vine disease, soil biology, fermentation, microorganisms and other living systems encountered throughout wine production. The principle is that biological processes create volatile signatures, and those signatures may contain information about the condition or activity of the organism producing them.
Whether every promising laboratory application will translate into practical viticulture remains to be seen.
Wine science has no shortage of technologies that perform impressively under controlled conditions but prove less transformative in the complexities of a working vineyard. Temperature, wind, soil, plant diversity and countless other variables make biological systems difficult to reduce to simple signals.
That is precisely why extensive training and validation matter.
The electronic nose should therefore be understood not as an oracle capable of diagnosing an entire vineyard from a single sniff, but as a potentially useful new category of sensor—one whose value will depend on the quality of the data behind it and the reliability with which particular signatures can be linked to particular conditions.
What an electronic nose cannot replace
Perhaps the most reassuring aspect of the research is what its developers do not expect the machine to do.
There is little suggestion that electronic noses will replace the tasting room.
A machine may become extremely good at recognising a chemical pattern. It may identify the signature associated with a particular fault more consistently than a human nose, especially when that fault is intermittent or appears at concentrations close to our sensory threshold.
But wine evaluation extends far beyond fault detection.
Blending, style and quality involve interpretation. Texture, proportion, development, pleasure and the relationship between different components of a wine cannot easily be reduced to a single analytical fingerprint.
The research team itself distinguishes between analytical assistance and the creative judgement of winemakers. Electronic sensing might characterise wine profiles or flag abnormalities, but the act of constructing a cuvée remains a human one.
That may ultimately be the technology’s most appropriate role: not replacing expertise, but giving expertise better information.
The Champagne electronic nose and a more observable vineyard
For all the sophistication of its sensors and algorithms, the Champagne electronic nose points towards a surprisingly simple ambition: understanding living systems before their problems become visible.
Wine is biological from beginning to end. Soil organisms, vines, fungi, yeasts and bacteria all participate in the chain that eventually produces what we taste in the glass. Yet much of this activity remains difficult to observe directly.
New sensing technologies offer a way of making parts of that invisible world measurable.
The researchers behind the Épernay project see this ability to understand biological systems as increasingly important to sustainable viticulture. Their view is that future solutions will require a more systemic understanding of vineyard and winery health rather than reliance solely on interventions applied after problems appear.
There is good reason to resist grand predictions. The equipment is not yet commercially available for wine use, and the distance between a successful research instrument and an everyday viticultural tool can be considerable.
Yet its underlying idea deserves attention.
Some of the most important innovations in French wine may not change what appears on the label or what a consumer sees in the vineyard. They may instead change what growers are capable of seeing before the rest of us notice anything at all.
For Champagne, a region built upon precision, blending and meticulous control of detail, that could make the electronic nose far more than a technological curiosity.
It could become another instrument for listening to what the vineyard is already saying.