# Pre-Decoding Acoustic Triage for Budgeted Vision-Language Captioning of Untrimmed Egocentric Video

arXiv: https://arxiv.org/abs/2608.22359 · HTML: https://mlsanity.com/paper/2608.22359/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 31 of its 38 references that are in the map.
- First posted on arXiv 23 Aug 2026.

## Builds on (most cited first, 30 of 31)

- [Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond](https://mlsanity.com/paper/2308.12966/) (2023)
- [Ego4D: Around the World in 3,000 Hours of Egocentric Video](https://mlsanity.com/paper/2110.07058/) (2021)
- [Video-LLaVA: Learning United Visual Representation by Alignment Before Projection](https://mlsanity.com/paper/2311.10122/) (2023)
- [Scaling Egocentric Vision: The EPIC-KITCHENS Dataset](https://mlsanity.com/paper/1804.02748/) (2018)
- [Object Hallucination in Image Captioning](https://mlsanity.com/paper/1809.02156/) (2018)
- [AST: Audio Spectrogram Transformer](https://mlsanity.com/paper/2104.01778/) (2021)
- [PANNs: Large-Scale Pretrained Audio Neural Networks for Audio Pattern Recognition](https://mlsanity.com/paper/1912.10211/) (2019)
- [LLaVA-PruMerge: Adaptive Token Reduction for Efficient Large Multimodal Models](https://mlsanity.com/paper/2403.15388/) (2024)
- [BEATs: Audio Pre-Training with Acoustic Tokenizers](https://mlsanity.com/paper/2212.09058/) (2022)
- [ActionFormer: Localizing Moments of Actions with Transformers](https://mlsanity.com/paper/2202.07925/) (2022)
- [BSN: Boundary Sensitive Network for Temporal Action Proposal Generation](https://mlsanity.com/paper/1806.02964/) (2018)
- [Efficient Training of Audio Transformers with Patchout](https://mlsanity.com/paper/2110.05069/) (2021)
- [HTS-AT: A Hierarchical Token-Semantic Audio Transformer for Sound Classification and Detection](https://mlsanity.com/paper/2202.00874/) (2022)
- [Adaptive Keyframe Sampling for Long Video Understanding](https://mlsanity.com/paper/2502.21271/) (2025)
- [Audio-Visual Segmentation](https://mlsanity.com/paper/2207.05042/) (2022)
- [Unified Multisensory Perception: Weakly-Supervised Audio-Visual Video Parsing](https://mlsanity.com/paper/2007.10558/) (2020)
- [Frame-Voyager: Learning to Query Frames for Video Large Language Models](https://mlsanity.com/paper/2410.03226/) (2024)
- [Q-Frame: Query-aware Frame Selection and Multi-Resolution Adaptation for Video-LLMs](https://mlsanity.com/paper/2506.22139/) (2025)
- [BOLT: Boost Large Vision-Language Model Without Training for Long-form Video Understanding](https://mlsanity.com/paper/2503.21483/) (2025)
- [Slow-Fast Auditory Streams For Audio Recognition](https://mlsanity.com/paper/2103.03516/) (2021)
- [VideoLLM-MoD: Efficient Video-Language Streaming with Mixture-of-Depths Vision Computation](https://mlsanity.com/paper/2408.16730/) (2024)
- [Epic-Sounds: A Large-scale Dataset of Actions That Sound](https://mlsanity.com/paper/2302.00646/) (2023)
- [End-to-End Temporal Action Detection with 1B Parameters Across 1000 Frames](https://mlsanity.com/paper/2311.17241/) (2023)
- [MDP3: A Training-free Approach for List-wise Frame Selection in Video-LLMs](https://mlsanity.com/paper/2501.02885/) (2025)
- [A Comparison of Five Multiple Instance Learning Pooling Functions for Sound Event Detection with Weak Labeling](https://mlsanity.com/paper/1810.09050/) (2018)
- [EgoSound: Benchmarking Sound Understanding in Egocentric Videos](https://mlsanity.com/paper/2602.14122/) (2026)
- [Chapter-Llama: Efficient Chaptering in Hour-Long Videos with LLMs](https://mlsanity.com/paper/2504.00072/) (2025)
- [Detect Any Sound: Open-Vocabulary Sound Event Detection with Multi-Modal Queries](https://mlsanity.com/paper/2507.16343/) (2025)
- [Multi-Iteration Multi-Stage Fine-Tuning of Transformers for Sound Event Detection with Heterogeneous Datasets](https://mlsanity.com/paper/2407.12997/) (2024)
- [Prototype based Masked Audio Model for Self-Supervised Learning of Sound Event Detection](https://mlsanity.com/paper/2409.17656/) (2024)

## Abstract (arXiv)

Automatically analyzing hours-long egocentric video is increasingly essential for progress monitoring, quality control, and safety in logistics, construction, and manufacturing. Yet current pipelines that process short, fixed-size windows with a vision-language model (VLM) are prohibitively expensive because cost scales with the number of model calls. To reduce this cost, prior work proposes triage policies to select which windows merit a VLM invocation. However, these policies either sample uniformly or rank windows using visual features, which ironically requires the video decoding that the budget constraints are meant to avoid. We propose audio-first triage: select windows using the lightest modality, scored before any video frame is decoded, so the approach composes naturally with token compression or quantization. The novelty lies in the objective, not the representation: rather than a per-frame sound-event detector, we train the selector to trigger once per action. This objective shift improves action coverage by 4.0-10.8 percentage points across all evaluated call rates, using frozen AudioSet-pretrained features without domain-specific sound-event labels. Using fewer than half of the available calls, the triage cuts 9-20% of VLM calls at matched coverage on EPIC-KITCHENS-100 (EK-100), surpasses uniform sampling through the mid-range on Ego4D over 247 clips, and outperforms two recent visual keyframe selectors. Code, the reference implementation and every results file this manuscript reads are at https://github.com/masjalayer/PreDecoding-AcousticTriage.