Saved in:
Bibliographic Details
Main Authors: Gibier, Marcel, Celton, Nolwenn, Duroselle, Raphaël, Serrano, Pierre, Boeffard, Olivier, Bonastre, Jean-François
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2511.14307
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915625600811008
author Gibier, Marcel
Celton, Nolwenn
Duroselle, Raphaël
Serrano, Pierre
Boeffard, Olivier
Bonastre, Jean-François
author_facet Gibier, Marcel
Celton, Nolwenn
Duroselle, Raphaël
Serrano, Pierre
Boeffard, Olivier
Bonastre, Jean-François
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2511_14307
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Audio Question Answering with GRPO-Based Fine-Tuning and Calibrated Segment-Level Predictions
Gibier, Marcel
Celton, Nolwenn
Duroselle, Raphaël
Serrano, Pierre
Boeffard, Olivier
Bonastre, Jean-François
Sound
Machine Learning
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.
title Audio Question Answering with GRPO-Based Fine-Tuning and Calibrated Segment-Level Predictions
topic Sound
Machine Learning
url https://arxiv.org/abs/2511.14307