Acoustic sensors will help identify cows with low methane emission levels
Researchers used acoustic sensors attached to lightweight collars at the base of the neck of livestock animals using a strap. The devices recorded eructation sounds in 10 cows, after which artificial intelligence was trained to automatically recognize these signals. An expert listened to the recordings, counted the number of burps, and assessed their loudness, while the AI was able to correctly recognize them in 80–90% of cases compared to a human.
Over the course of three days, the detectors recorded that the cows burped slightly less than once a minute at an average loudness of two points on a scale from one to three. Studies revealed noticeable differences between individuals: some livestock animals produced sounds more frequently or intensely than others. According to the authors of the work, the coefficients of variation for frequency and intensity were 0.18, which is consistent with previous research.
The combination of sound sensors and artificial intelligence can offer farmers an inexpensive tool for monitoring livestock and selecting individuals with lower methane emissions. This technology may be combined with intraruminal gas sensors or "sniffers" to refine emission data and support breeding programs. Nevertheless, the authors emphasize the need for larger-scale studies involving a greater number of livestock animals, across different farms, on various days, seasons, and lactation periods to verify the system's performance.
A research article authored by AgResearch Senior Scientist Paul Shorten was published in the New Zealand Journal of Agricultural Research. The project was funded by AgResearch New Zealand.