LLM-Based Synthetic Ground Truth Generation for Audio-Based Emotion Classification via In-Context Learning
Researchers have developed a novel method for generating synthetic ground truth data for audio-based emotion classification, particularly within immersive virtual reality environments. This approach utilizes large language models (LLMs) and in-context learning (ICL) to adapt to streaming speech data without requiring computationally intensive fine-tuning. The system employs a retrieval-based strategy to select relevant audio demonstrations for prompt construction, aiming to overcome challenges in labeling dynamic team processes from noisy and sparse sensor data. AI