多语言语音识别系统在对话任务中实现29.27%的通话级错误率，优于基线。
An Omnilingual-ASR-Based Speech-LLM System for the 2nd MLC-SLM Challenge
- 分步处理：分割-聚类-识别，使用大模型与LoRA微调。
- 开发集通话级错误率达29.27%，评测集50.23%。
- 基于嵌入的聚类优于仅靠转录标记的方案，适合多语种场景。
我们提交了第二届MLC-SLM挑战赛任务1的系统：一个级联的去重-识别架构，结合DiariZen-Large-s80（WavLM-Large）分割、基于CAM++嵌入的双说话人聚类，以及经过LoRA适配的omniASR LLM 7B v2识别器，测试时无需先验分割或说话人标签。在官方开发集（150段对话，21种语言/口音类别）上，系统取得29.27%的宏平均通话级通话错误率（tcpMER），相比官方基线的79.15%有显著提升；在评测集上得分为50.23%。我们还分析了两项工程选择对tcpMER的显著影响：第一，基于嵌入的说话人聚类优于仅依赖ASR <sc> 转录标记的端到端式方案；第二，重叠感知分割虽提升去重召回率，但因重叠语音被重复转录，反而导致tcpMER升高。
原文摘要 · Abstract (English)
We describe our submission to Task 1 of the 2nd MLCSLM Challenge: a cascaded diarization-then-recognition system that combines DiariZen-Large-s80 (WavLM-Large) segmentation, CAM++ embedding-based two-speaker clustering, and a LoRA-adapted omniASR LLM 7B v2 recognizer, with no oracle segmentation or speaker labels at test time. On the official Development set (150 conversations, 21 language/accent categories) the system attains a macro tcpMER of 29.27%, versus 79.15% for the official baseline; on the Evaluation set it scores 50.23%. We also analyze two engineering choices that substantially affect tcpMER. First, embedding-based speaker clustering outperforms an end-to-end-style alternative that assigns speakers from ASR <sc> turn markers alone. Second, overlap-aware segmentation, although intended to raise diarization recall, increases tcpMER because overlapped speech is transcribed twice.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。