Audio Event Timing QA - Earliest Onset: leaderboard
Four-option questions asking which of four sound events in a real-world Freesound clip (event timings from the TACOS corpus) starts first; the correct event starts at least one second before every other option. From a study of temporal-reasoning failures in large audio-language models (Kulkarni et al., 2026); the model hears only the audio, with no caption, and the score is accuracy in percent over 528 questions, higher is better.
Metric: Accuracy (%; which of four sound events starts first, audio only; 528 questions). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Audio Flamingo 3 | 60.42 | |
| 2 | Kimi-Audio-7B-Instruct | 57.95 | #1436 |
| 3 | DeSTA2.5-Audio | 50.38 | #1758 |
| 4 | Qwen2-Audio-7B-Instruct | 30.87 |
Interactive version: theaggregate.ai/benchmark?slug=audio-event-timing-qa-earliest-onset · How It Works · Data refreshed daily, snapshot 2026-11-12.