SCENEBench (Audio Understanding) - Code-Switched Transcription (Mandarin): leaderboard
Metric: Mean similarity (%) between the model's transcription and the reference sentence over 884 synthetic code-switched clips: English DailyTalk turns with about 30 percent of the words translated into Mandarin and voiced by multilingual TTS; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen2-Audio-7B-Instruct | 61 | #859 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=scenebench-audio-understanding-code-switched-transcription-mandarin · How It Works · Data refreshed daily, snapshot 2026-10-11.