Audio Question Answering with GRPO-Based Fine-Tuning and Calibrated Segment-Level Predictions
This is an incremental improvement for audio question answering in a specific challenge context.
The authors tackled Audio Question Answering by combining acoustic event predictions from BEATs with a fine-tuned Qwen2.5-7B-Instruct model using GRPO, achieving 62.6% accuracy on the DCASE 2025 development set.
In this report, we describe our submission to Track 5 of the DCASE 2025 Challenge for the task of Audio Question Answering(AQA). Our system leverages the SSL backbone BEATs to extract frame-level audio features, which are then processed by a classification head to generate segment-level predictions of acoustic events, following the Audioset ontology. These segment-level predictions are subsequently calibrated before producing event-level predictions. Finally, these predictions are incorporated into a structured prompt, along with the question and candidate answers. This prompt is then fed to a fine-tuned version of Qwen2.5-7B-Instruct, trained using the GRPO algorithm with a simple reward function. Our method achieves an accuracy of 62.6 % on the development set, demonstrating the effectiveness of combining acoustic event reasoning with instruction-tuned large language models for AQA.