KMMAU - Age: leaderboard
Speaker age recognition, a three-way choice, on native Korean speech in KMMAU (Kim et al., 2026). Accuracy in percent; higher is better.
Metric: Accuracy (%; three-way speaker age question; native Korean audio from KSS, KMSAV and the Seoul Corpus; single run at temperature 0.7). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Raon-Speech | 62 | |
| 2 | Step-Audio-2-mini | 52.2 | |
| 3 | MiniCPM-o 4.5 | 51.8 | |
| 4 | Fun-Audio-Chat | 47.5 | |
| 5 | Qwen2.5-Omni | 41.3 | #1499 |
| 6 | InteractiveOmni | 29 | |
| 7 | Audio Flamingo 3 | 21 | |
| 8 | HyperCLOVA X 8B Omni | 21 |
Interactive version: theaggregate.ai/benchmark?slug=kmmau-age · How It Works · Data refreshed daily, snapshot 2026-11-12.