The Spanish company CEALVET SLu develops the first bioacoustic sensor for the poultry industry called VocalData.

VocalData is an additional tool as part of the global poultry welfare strategy. By focusing on the acoustic expression of animals, we do not define what their well-being is, but rather they vocalize it, sing it.

“Using information from VocalData, it is possible to predict production parameters and measure animal welfare in both commercial farms and hatcheries.”

How was your project born?

Our project was conceived about 4 years ago, in parallel with the application of our product Licuicel Complex, a complementary feed for early nutrition administered in the hatchery. Our need was to measure the benefit on the welfare of the treated animals, observing the improvement produced and the reduction of stress in the chicks.

In its publication this year, EFSA, in the “Bird welfare” section 2022 20(9), referred to the work of CEALVET by identifying a significant change in the acoustic parameter of frequency in stress calls emitted by birds (Ginovart-Panisello et al., 2020 Acoustic Description of the Soundscape of a Real-Life Intensive Farm and Its Impact on Animal Welfare: A Preliminary Analysis of Farm Sounds and Bird Vocalisations).

We are currently participating in the European Eurostars project “Development of new integrated technologies against multi-resistant pathogens in broiler chickens”, where bioacoustics is a measurement tool that allows us to research indicators related to animal welfare.

The technology developed by CEALVET detects and captures bird vocalizations to obtain acoustic descriptors that we later explain.

What is and acoustic descriptor?

An acoustic descriptor is a parameter obtained by automatically processing bird vocalizations and describes the vocalization or set of songs. Let’s think, for example, of calculating how many vocalizations a group of animals emits per unit of time. In this case the value obtained describes the temporal frequency of the calls of the group of animals and is considered an acoustic descriptor.

To correctly obtain acoustic data from a group of birds, is it necessary to change the handling of the animals?

Our technology is small and non-invasive. It is installed when there are no animals in the farm or in the shipping room of the hatchery, in an elevated position, and captures all vocalizations, eliminating noises external to those generated by the animals, without altering the management routine. Once installed, you will forget that animal vocalizations are continuously analyzed.

What services do you offer your customers with VocalData in hatcheries?

In hatcheries, the sensor is installed in the shipping room, and allows uninterrupted monitoring from the moment the first animal arrives until the last one is loaded onto the truck headed to the farm. The program allows you to analyze and compare the stress of the animals based on the day of birth, the handling practices of the staff, the different treatments carried out, the seasons of the year, among others, generating a register of all the animals.

What services do you offer your customers with VocalData on commercial farms?

In commercial farms, the sensor is installed in the center of the farm and allows uninterrupted monitoring, every 15 minutes, of the state of well-being expressed by the birds. Furthermore, by applying artificial intelligence models, it is possible to predict the weight of the bird 4 days in advance, based only on the animals’ vocalizations. All the data obtained is accessible through a web application that allows you to use the data to compare management between multiple farms, or reference welfare values.

What are the advantages of applying artificial intelligence to bioacoustics data?

More than 2.7 Tb of bird vocalizations were used to create the algorithms, corresponding to 17 production cycles for both Ross and Cobb chickens and one year of hatchery births. This large amount of data would be impossible to process if it weren’t for artificial intelligence, which helps us find relationships between descriptors in a supervised and automated way. The algorithms are constantly updated as they are reevaluated with new data obtained, and the error is reduced. Performing the same job manually would be impossible with such satisfactory results.

What future ideas do you currently have with bioacoustics?

The project is constantly evolving to meet the needs of our customers and improve the welfare of the birds where possible. Our challenge is that our system can generate reliable alarms based on animal behavior in order to apply preventive or corrective measures.

Would it be possible in the future to understand what animals express?

Animal welfare is actually individual, and therefore it is essential to have robust methods to measure it on the animals themselves. We must not forget that the science of animal welfare is a young science, and for this reason we believe we must concentrate our efforts by focusing more on life experience and on what they transmit to us throughout the breeding period with their vocalizations in order to be able to obtain maximum information and, in the future, greater understanding, relating it more efficiently with the production environment. We are convinced that thanks to technological advances in artificial intelligence and much more data, we will be ever closer to understanding their behavior.