KMMAU - Gender: leaderboard
Speaker gender recognition, a two-way choice, on native Korean speech in KMMAU (Kim et al., 2026). Accuracy in percent; higher is better.
Metric: Accuracy (%; two-way speaker gender 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 | Qwen2.5-Omni | 98.9 | #1499 |
| 2 | Raon-Speech | 93.7 | |
| 3 | MiniCPM-o 4.5 | 91.1 | |
| 4 | Step-Audio-2-mini | 88.9 | |
| 5 | Fun-Audio-Chat | 87.8 | |
| 6 | Audio Flamingo 3 | 57 | |
| 7 | InteractiveOmni | 46.3 | |
| 8 | HyperCLOVA X 8B Omni | 30 |
Interactive version: theaggregate.ai/benchmark?slug=kmmau-gender · How It Works · Data refreshed daily, snapshot 2026-11-12.