SEA-SpeechBench - Speaker Recognition (SEA Prompt): leaderboard
Metric: Macro-F1 (%; whether two clips come from the same speaker, on self-constructed same and different speaker pairs, refusals count as errors; short clips of at most 30 s from curated public Southeast Asian speech corpora, up to 1,000 sampled items per dataset; instruction prompt in the native Southeast Asian language). Source: arxiv.org. Saturation forecast: Around January 2028. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Flash | 51.54 |
| 2 | Qwen3 Omni 30B A3B Instruct | 44.28 |
| 3 | Voxtral-Mini-3B-2507 | 38.67 |
| 4 | Gemma 3n E4B (IT) | 34.13 |
| 5 | Phi-4 Multimodal Instruct | 32.73 |
| 6 | gemma-3n-E2B-it | 32.07 |
| 7 | Qwen2-Audio-7B-Instruct | 31.66 |
| 8 | SeaLLMs-Audio-7B | 31.05 |
Interactive version: theaggregate.ai/benchmark?slug=sea-speechbench-speaker-recognition-sea-prompt · How It Works · Data refreshed daily, snapshot 2026-09-26.