A Dataset of Plant Sounds Under Stressful Conditions

This dataset was collected by researchers from Tel Aviv University in Israel in 2023 as part of the paper “Sounds emitted by plants under stress are airborne and informative"release.

Experiments have shown that in addition to showing changes in color, smell and shape, plants under stress also make sounds that can be remotely recorded and classified. By developing a machine learning model, researchers have successfully identified the state of plants, including dehydration and damage such as broken stems, based solely on the sounds they make.

Details of the directories in this dataset:

• Dry Tomato: Tomato plants were not watered for 4-6 days until the soil moisture in the pots dropped below 5%. Recordings were conducted in a soundproof box.
• Dry Tobacco: Tobacco plants were not watered for 4-6 days until the soil moisture in the pots dropped below 5%. Recordings were conducted in a soundproof box.
• Cut Tomato: Tomato plants with moist soil were cut before the recording started. The plant part that was cut and disconnected from the roots was recorded. The recording was performed in a soundproof box.
• Cut Tobacco: Tobacco plants with moist soil were cut before the recording began. The plant parts that were cut and disconnected from the roots were recorded. The recordings were made in a soundproof box.
• Empty Pot: Sounds recorded from an empty pot in a soundproof box. These sounds are generated by the microphone itself and are recorded from only one microphone.
• Greenhouse Noises: Sounds recorded from an empty greenhouse, captured by two microphones, and originating from unknown sources inside and outside the greenhouse.

The files in the dataset are in WAV format and can be opened with any audio processing program.