一种语言无关的分层LoRA-MoE架构用于基于CTC的多语言ASR
A Language-Agnostic Hierarchical LoRA-MoE Architecture for CTC-based Multilingual ASR
AI总结：
本文提出基于CTC架构的轻量多语言ASR系统，通过分层LoRA-MoE框架实现语言无关解码，实验表明在资源受限场景下提升解码效率。
摘要：
大规模多语言ASR（mASR）模型如Whisper在性能上表现优异，但计算和延迟成本高，限制了其在资源受限的边缘设备上的部署。本文提出了一种轻量且语言无关的多语言ASR系统，基于CTC架构并结合领域自适应。具体而言，我们引入了语言无关的分层LoRA-MoE（HLoRA）框架，集成到mHuBERT-CTC模型中，通过LID-posterior驱动的LoRA路由实现端到端解码。分层设计包括多语言共享LoRA用于学习语言不变的声学表示，以及语言特定的LoRA专家用于建模语言依赖性特征。所提出的路由机制在推理过程中无需先验语言身份信息或显式语言标签，实现真正的语言无关解码。在MSR-86K和MLC-SLM 2025挑战数据集上的实验表明，HLoRA在性能上与两阶段推理方法相当，同时将RTF分别减少11.7%和8.2%，从而提高了低资源mASR应用的解码效率。
英文摘要：
Large-scale multilingual ASR (mASR) models such as Whisper achieve strong performance but incur high computational and latency costs, limiting their deployment on resource-constrained edge devices. In this study, we propose a lightweight and language-agnostic multilingual ASR system based on a CTC architecture with domain adaptation. Specifically, we introduce a Language-agnostic Hierarchical LoRA-MoE (HLoRA) framework integrated into an mHuBERT-CTC model, enabling end-to-end decoding via LID-posterior-driven LoRA routing. The hierarchical design consists of a multilingual shared LoRA for learning language-invariant acoustic representations and language-specific LoRA experts for modeling language-dependent characteristics. The proposed routing mechanism removes the need for prior language identity information or explicit language labels during inference, achieving true language-agnostic decoding. Experiments on MSR-86K and the MLC-SLM 2025 Challenge datasets demonstrate that HLoRA achieves comparable performance to two-stage inference approaches while reducing RTF by 11.7% and 8.2%, respectively, leading to improved decoding efficiency for low-resource mASR applications.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
