Sunday, November 22, 2026
Bioacoustics

How can a plant’s musical personality be established?

Talking about a plant’s “personality” in music does not mean arbitrarily attributing it a human character: cheerful, sad, calm or nervous. A much more fruitful approach is to define its musical personality as the whole of the…

Talking about a plant’s “personality” in music does not mean arbitrarily attributing it a human character: cheerful, sad, calm or nervous. A much more fruitful approach is to define its musical personality as the set of sonic characteristics that remain recognisable over time, to which the variations in its instantaneous behaviour are superimposed.

In other words: a plant has a musical identity, and then it passes through musical states.

It is precisely this separation that makes it possible to turn plant sonification into a genuine research object.

Morphology forms the first level of personality

A succulent, a cactus, a fern or a climbing plant do not have the same mechanical architecture, the same leaf mass, the same arrangement of petioles, or the same way of occupying space.

These physical differences produce different mechanical behaviours.

A broad leaf carried by a flexible petiole has a different natural frequency from that of a small rigid leaf. A water-rich succulent has more inertia than light foliage. A fern distributes its leaflets according to a repetitive architecture. A climber, by contrast, develops a directional organisation.

Plantsound uses this morphology as a first level of translation. Artificial intelligence recognition assigns the plant to a family, then that family receives a musical profile made up of a scale, a register, a timbre, an articulation and a tempo. Plantsound’s scientific system thus describes thirteen morphological families and, for example, associates the cactus with a minor pentatonic scale, a marimba, fast, staccato notes, while other plant architectures use different musical gestures.

This profile obviously does not mean that “the cactus naturally plays the marimba”. It constitutes a coherent musical projection of its morphology.

The distinction is fundamental.

The musical profile gives the plant a perceptible identity. But it does not determine the melody it will produce.

Movement gives the temperament

Two cacti placed side by side have a comparable morphology. Yet their movements are never perfectly identical.

That is where individuality truly begins.

Plantsound optically analyses extremely small variations in the foliage. Its processing system distinguishes, in particular, the relatively rapid oscillations of the foliage and much slower drifts, associated with phenomena such as variations in turgor or certain directional movements. The processing chain thus exploits two time scales: a fast pathway and a slow pathway.

The extracted signal then drives the degrees of the scale.

Thus, two plants belonging to the same family have the same general musical vocabulary, but they do not produce the same phrase.

It is exactly like two pianists playing the same instrument and using the same scale: the material is shared, but the gesture differs.

We can then define several measurable behavioural traits:

amplitude, which describes the general intensity of micro-movements;

density, which measures the frequency with which significant events occur;

speed, which gives information about temporal dynamics;

regularity, which distinguishes periodic movements from more irregular behaviours;

persistence, which measures how long a trend remains in place;

diversity, which describes the richness of the changes observed;

response to events, which shows how the dynamics evolve after a change in light, humidity, air or temperature.

These parameters gradually form a genuine behavioural portrait.

Personality appears through repetition

A thirty-second observation mainly tells us about the present state.

To speak of personality, long-term time must be introduced.

Let us imagine that Plantsound observes the same plant every day for several weeks. The AI then no longer merely retains the notes produced. It builds a multidimensional space containing the statistical properties of its behaviour.

For each session:

P = [energy, dominant frequency, variability, slow activity, density, asymmetry, rest time, environmental response...]

Each observation becomes a point in this space.

After several dozen measurements, a cloud specific to the plant emerges.

The centre of the cloud represents its habitual behaviour.

Its spread represents its variability.

Its movements represent its state changes.

What is then obtained is something far more powerful than a simple sonification: an individual dynamic signature.

Two Monsteras identified by the AI as belonging to the same family then begin with a similar musical language, but several weeks of observation bring out two different behaviours.

One shows very stable slow activity, with long phrases and few breaks.

The other shows more rapid oscillations and abrupt changes.

Musically, the first develops a spacious and continuous personality. The second becomes more fragmented and mobile.

The musical personality is therefore not programmed by the software: it emerges from the statistics of the measured behaviour.

Character must be distinguished from mood

This distinction opens up a particularly interesting architecture.

Character corresponds to properties that persist for several days or weeks.

State corresponds to variations measured over a few seconds or a few minutes.

Musically, the two levels do not act on the same parameters.

Character can gradually determine the preferred register, the average density, the ambitus, the amount of silence, the length of the notes or the harmonic richness.

The instantaneous state, by contrast, drives the melody, the intensity, the acceleration, certain changes in texture and the punctual events.

Thus a plant retains its sonic identity while producing different music at each session.

This is probably one of the most important criteria of successful sonification: recognisable without being repetitive.

The environment becomes a third dimension

A personality never exists independently of its surroundings.

The plant responds to light, water, temperature, air movement, atmospheric pressure and day-night cycles.

Plantsound already incorporates this principle: the foliage signal drives the notes, the recognition of the plant structures its profile and certain environmental data, notably atmospheric pressure and the UV index, play a part in the sound texture.

This separation makes important scientific progress possible.

It becomes possible to study not only how a plant moves, but how it modifies its behaviour when its environment changes.

Personality then becomes partly relational.

One plant may show strong sensitivity to light and weak sensitivity to small air movements. Another, morphologically similar, may respond more to mechanical changes than to light variations.

These differences provide a new layer of personality.

One can even define a sensitivity vector:

S = [response to light, water response, thermal response, mechanical response, circadian response]

Each plant then obtains a true “reactivity map”.

Taking AI from recognition to learning the individual

Today, the artificial intelligence in Plantsound is used in particular to recognise the morphology of the plant and assign it a suitable musical profile. The chain described in the scientific dissertation combines recognition by MobileCLIP-S0, the musical profile and then the translation of the measured movement into scale degrees.

The next step is to give this intelligence a memory.

The AI then learns the plant itself.

After several sessions, it knows its usual activity level, its calm periods, its dominant mechanical frequencies, its daily rhythm and its characteristic reactions.

It no longer compares only a cactus with a fern.

It compares this cactus today with the same cactus from thirty days ago.

It is at this precise moment that the notion of personality becomes scientifically interesting.

An increase in activity is no longer interpreted as an absolute value. It is assessed relative to the individual’s habitual behaviour.

A naturally very mobile plant may remain in its normal state with high energy. Another, usually extremely stable, produces an exceptional event with the same energy.

Individualisation therefore completely changes the musical meaning of the signal.

Music becomes a form of temporal portrait

We then arrive at a much richer definition of “plant music”.

The plant does not directly produce notes.

It produces movements.

These movements contain a mechanical, physiological and environmental organisation.

Plantsound transforms this organisation into perceptible musical structures.

Morphology provides the voice.

Habitual behaviour provides the character.

The current activity provides the gesture.

The environment provides the context.

The AI’s memory provides the individual identity.

From there, listening to a plant over several months takes on a new dimension. We no longer hear a succession of independent compositions. We hear a sonic identity that is built up, stabilises, evolves with the seasons, passes through periods of activity and rest and, despite these transformations, retains certain characteristic traits.

Music then becomes a kind of audible phenotype: not a literal translation of the plant, but a temporal representation of the particular way in which this organism occupies its environment and responds to it.

And this is probably where plant sonification becomes most interesting: when it stops asking “what music does a plant play?” and asks a much more precise question:

what is it in a plant’s behaviour that remains sufficiently stable to be recognisable by ear?

This question turns plant music into a genuine research programme.

To find out more, visit the app https://plantsound.fr

LR

La Rédaction

Directeur de la publication
Newsletter

Don't miss the next article.

Get every new article, or the Monday digest.

The NEURAL.RE newsletter

The science of the brain and sound, in your inbox.

Every new article as soon as it is published, or the best of the week on Monday morning. Free, no ads, unsubscribe in one click.