对比七种微调方法，发现简单版LoRA最适合患者专用失语症语音识别。
Choosing a PEFT Variant for Per-Patient Dysarthric ASR: A Single-Speaker Case Study on Two ASR Bases
- 在单个患者上测试七种低秩微调方法，聚焦失语症语音识别场景。
- LoRA与DoRA效果相近（CER 13.86% vs 13.90%），但LoRA更便宜简单。
- 4比特QLoRA反而更差，且无内存节省；少量录音即可大幅改善识别率。
针对失语症自动语音识别（ASR）的个性化适配器是主流生产架构，但参数高效微调（PEFT）变体在依赖说话人、每患者独立训练的场景中尚未系统比较。本文以一位匈牙利男性卒中后严重失语患者（S1，409条语句）为案例，对比了七种LoRA家族方法（LoRA、QLoRA、AdaLoRA、DoRA、LoHA、VeRA、VB-LoRA）在两个生产级模型（经过匈牙利语微调的Whisper-large-v3和多语言Qwen3-ASR-1.7B）上的表现。注意力投影适配器显著降低词错误率（CER）。三次随机种子下，配对置换检验显示LoRA与DoRA无显著差异（p>0.5；Whisper上分别为13.86%/13.90%，Qwen3-ASR上为28.10%/28.33%），故采用更简洁低成本的LoRA。真实4比特（NF4）QLoRA在所有种子和模型上表现更差（14.56%/30.09% CER），且在此规模无内存节省。其他方法均未超越LoRA，尽管LoHA在Whisper上仍带来18.6%的相对CER下降。全量微调更准确（11.43% CER），但仅115MB的LoRA（同时适配前馈层）即达到其性能的99.34%（仅差0.66个百分点），存储成本约为3.7%。6点注册采样网格显示，约5分钟患者音频可捕获零样本到30分钟的45.6%性能提升，10分钟和30分钟仍有进一步增益（提醒：单一患者、单一语言、严重卒中后失语症）。训练脚本与配置将在发表时开源，研究用途许可。
原文摘要 · Abstract (English)
Per-patient adapters are the preferred production architecture for dysarthric automatic speech recognition (ASR), yet parameter-efficient fine-tuning (PEFT) variants have not been compared in the speaker-dependent, per-patient regime. We present a single-speaker case study comparing seven LoRA-family methods (LoRA, QLoRA, AdaLoRA, DoRA, LoHA, VeRA, VB-LoRA) on two production bases (Whisper-large-v3 with Hungarian fine-tuning, and a multilingual Qwen3-ASR-1.7B checkpoint) for one post-stroke Hungarian male speaker (S1, 409 utterances; severe dysarthria on auditory-perceptual clinical assessment). Attention-projection adapters substantially improve CER on both bases. Across three seeds, a paired bootstrap detects no significant LoRA-DoRA difference (p>0.5; 13.86/13.90 % CER on Whisper, 28.10/28.33 % on Qwen3-ASR), so we adopt the simpler, cheaper LoRA. Real 4-bit (NF4) QLoRA is worse on every seed and both bases (14.56/30.09 % CER) with no memory saving at this scale, and LoHA, VeRA, VB-LoRA and AdaLoRA do not reach the LoRA family, though LoHA still gives an 18.6 % relative CER reduction on Whisper. On the same base, full fine-tuning is more accurate (11.43 % CER), but a 115 MB LoRA that also adapts the feed-forward blocks reaches within 0.66 pp of it at approximately 3.7 % of the per-patient storage. A 6-point enrollment grid shows about 5 min of patient audio captures 45.6 % of the zero-shot-to-30-min CER reduction, with further gains at 10 and 30 min (caveat: one speaker, one language, severe post-stroke dysarthria). Training scripts and recipes will be released, source-available under a research-use licence, on publication.
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。