A recent study has found that artificial intelligence (AI) can accurately detect human emotions from audio clips as short as 1. 5 seconds, matching human performance. The researchers used over 1, 500 audio clips from English and German databases to train the AI models. The study focused on six emotional categories: joy, fear, neutral, anger, sadness, and disgust.

Various features, such as pitch tracking, spectral bandwidth, and root mean square, were used to quantify the emotional states. The AI models, including a deep neural network and a hybrid model, outperformed a convolutional neural network. This research demonstrates the potential for AI to have cognitive empathy and advance automatic emotion recognition in various applications.