Scientists are increasingly utilizing artificial intelligence models to decipher animal communication, helping bridge the gap between human observation and species-specific behavior. The Earth Species Project (ESP), a U.S.-based nonprofit, is developing generalized and custom AI tools to analyze large-scale acoustic datasets. In northern Spain, researchers Vittorio Baglione and Daniela Canestrari have collaborated with ESP since 2024 to decode vocalizations of carrion crows (Corvus corone). These birds practice cooperative breeding, necessitating highly complex social interactions. The AI model has successfully processed more than 127,000 vocalizations, enabling researchers to differentiate between adult and chick calls while synchronizing data from multiple audio loggers to track overlapping dialogues.

Initial analysis suggests that the majority of crow communication involves soft, low-amplitude murmurs, indicating a preference for close-range interactions. Ongoing efforts aim to create a comprehensive semantic map that integrates audio recordings with video data and accelerometer (device measuring movement and speed) input to correlate specific behaviors with vocalizations. This multimodal approach offers scientists a more granular understanding of how individuals coordinate complex tasks within their social structures.

The application of this technology extends to marine conservation as well. The Raincoast Conservation Foundation in Canada is partnering with ESP to study orca (Orcinus orca) behavior. By utilizing drone footage and acoustic recordings, researchers are building datasets that link vocal patterns to specific actions and environmental factors. The AI tools significantly accelerate data processing, reducing the time required to synchronize voice notes with whale audio from months to approximately one hour. While these advancements improve the efficiency of identifying and labeling dialects, researchers emphasize the goal remains understanding animals on their own terms rather than attempting human-like communication.

Scientists are increasingly utilizing artificial intelligence models to decipher animal communication, helping bridge the gap between human observation and species-specific behavior. The Earth Species Project (ESP), a U.S.-based nonprofit, is developing generalized and custom AI tools to analyze large-scale acoustic datasets. In northern Spain, researchers Vittorio Baglione and Daniela Canestrari have collaborated with ESP since 2024 to decode vocalizations of carrion crows (Corvus corone). These birds practice cooperative breeding, necessitating highly complex social interactions. The AI model has successfully processed more than 127,000 vocalizations, enabling researchers to differentiate between adult and chick calls while synchronizing data from multiple audio loggers to track overlapping dialogues.

Initial analysis suggests that the majority of crow communication involves soft, low-amplitude murmurs, indicating a preference for close-range interactions. Ongoing efforts aim to create a comprehensive semantic map that integrates audio recordings with video data and accelerometer (device measuring movement and speed) input to correlate specific behaviors with vocalizations. This multimodal approach offers scientists a more granular understanding of how individuals coordinate complex tasks within their social structures.

The application of this technology extends to marine conservation as well. The Raincoast Conservation Foundation in Canada is partnering with ESP to study orca (Orcinus orca) behavior. By utilizing drone footage and acoustic recordings, researchers are building datasets that link vocal patterns to specific actions and environmental factors. The AI tools significantly accelerate data processing, reducing the time required to synchronize voice notes with whale audio from months to approximately one hour. While these advancements improve the efficiency of identifying and labeling dialects, researchers emphasize the goal remains understanding animals on their own terms rather than attempting human-like communication.