Audio Question Answering with GRPO-Based Fine-Tuning and Calibrated Segment-Level Predictions

Published in arXiv preprint arXiv:2511.14307, 2025