低资源方言语音识别需多种子评估，避免偶然误差。
Seeds Before Objectives: Rethinking Evaluation for Low-Resource Garhwali ASR
- 在官方数据集上用多种子训练，验证结果可靠性
- 标准CTC配合w2v-BERT 2.0达47.0% WER，优于大模型
- 适合关注低资源语音识别可复现性的研究者
针对低资源方言典型语料规模下单次实验结果不可复现的问题，本文以印度喜马拉雅山区的加尔瓦利语为例，构建首个基于官方VAANI划分的可复现多种子语音识别基准，提供每种子的输出结果并进行显著性检验。重新评估潜在提升时发现：焦点CTC和音节加权目标均未在种子级别测试中优于标准CTC，音节目标甚至未能减少其针对性错误；从印地语迁移学习也未带来优于直接微调的收益。真正稳健的是基础方案：使用w2v-BERT 2.0与标准CTC，在五个种子上平均达到47.0%的词错误率（WER），优于更大的MMS-1B及同类模型；性能关键在于预训练设计而非参数量，速度增强带来小而一致的改善。多种子评估有效区分真实性能提升与随机噪声。
原文摘要 · Abstract (English)
At corpus sizes typical of low-resource dialects, single-run comparisons can yield gains that do not replicate. We show this for Garhwali, an under-resourced Indo-Aryan language of the central Himalaya, building the first reproducible multi-seed ASR benchmark on the official VAANI splits, with per-seed outputs and significance testing. Re-examining plausible gains, we find them fragile: neither Focal CTC nor a matra-weighted objective beats standard CTC under seed-level testing, the matra objective fails to cut even its targeted errors, and Hindi-to-Garhwali transfer gives no gain over direct fine-tuning. What holds up is mundane: w2v-BERT 2.0 with standard CTC reaches 47.0% WER over five seeds, beating the larger MMS-1B and comparable models; pretraining design, not parameter count, drives performance, and speed augmentation gives a small, largely consistent gain. Multi-seed evaluation on official splits separates real gains from seed noise.
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。