🐶 AI Cracks the Code of Dog Barks

Researchers from the University of Michigan have developed an innovative AI model → to decode dog barks. By leveraging AI models initially trained on human speech, the team can discern whether a bark signals playfulness, aggression, and even identify the dog's age, breed, and sex.

The study utilized the Wav2Vec2 model, which was fine-tuned with dog barks after being pre-trained on nearly 1,000 hours of human speech. This repurposed model surpassed those trained solely on dog vocalizations, achieving up to 70% accuracy in emotion classification tasks.

The research opens new avenues for understanding animal communication by applying human speech analysis techniques to animal sounds, overcoming the challenge of limited animal vocalization data.

@durovs_code