Speechocean762 (LMMPA) - Fluency: leaderboard
Metric: Pearson correlation (-1 to 1) of predicted sentence fluency with human ratings on 2,500 Speechocean762 test utterances; specialist Qwen-7B multimodal system trained on LibriSpeech and Speechocean762 training data; higher is better. Source: arxiv.org. 1 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | LMMPA-Qwen-7B | 0.78 |
Interactive version: theaggregate.ai/benchmark?slug=speechocean762-lmmpa-fluency · How It Works · Data refreshed daily, snapshot 2026-10-12.