arXiv：cs.CL· Sangmin Lee, Yejin Jeon, David Ifeoluwa Adelani·· 16 小时前
ELARSpeech 发布 30K 小时、覆盖 675 种低资源语言的多语言语音数据集
ELARSpeech: A 30K-Hour Dataset Across 675 Underrepresented Languages for Long-Tail Multilingual Speech Technologies
AI 导读
ELARSpeech 是来自 Endangered Languages Archive、覆盖 675 种低资源语言的 30202 小时语音数据集。实验显示该数据集可将 ASR 与 TTS 能力扩展至这些语言，对未标注数据做伪标注还能带来进一步提升；语音表征分析也暴露出现有多语言模型迁移到语言长尾时的局限。数据集、训练好的 ASR 和 TTS 模型及完整数据构建流程均已公开。
正文
Abstract:Massively multilingual speech modeling has made remarkable progress in recent years, yet existing training resources and benchmarks still underrepresent much of the linguistic long tail. To address this gap, we introduce ELARSpeech, a large-scale multilingual speech resource comprising 30,202 hours of speech across 675 underrepresented languages, curated from the Endangered Languages Archive. Combining human-annotated and unlabeled recordings, the corpus spans diverse recording conditions, domains, and language families. Beyond corpus construction, our experiments show how ELARSpeech supports extending ASR and TTS capabilities to diverse underrepresented languages, with further benefits from pseudo-labeling unlabeled data. At the same time, our speech representation analysis reveals remaining limitations in transferring existing multilingual models to the linguistic long tail. Together, ELARSpeech provides both a resource and a testbed for extending multilingual speech technology further into the linguistic long tail. We publicly release the dataset, all trained ASR and TTS models, and the complete data construction pipeline to facilitate research on long-tail multilingual speech technologies.
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Sound (cs.SD) |
| Cite as: | arXiv:2610.19388 [cs.CL] |
| (or arXiv:2610.19388v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.19388 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Sangmin Lee [view email]
[v1]
Wed, 14 Oct 2026 01:32:49 UTC (456 KB)
来源：arXiv：cs.CL · arxiv.org