arXiv：cs.CL· Stephen Fernandes, Narayan Choudhary·· 16 小时前
Anuvani-Diarization：覆盖 22 种语言的零样本说话人分离与标注规范研究
Anuvani-Diarization: Zero Shot Speaker Diarization and Annotation Conventions Across 22 Languages
AI 导读
研究提出 Anuvani-Diarization，将自监督端到端分割模型适配印度多语言对话，在 IndicDiarBench 全部 1,164 条录音上以零样本方式取得 16.78% 的时长加权 DER，为所测系统最低，仅次于已发表最佳结果 16.0%，并在 22 种语言中的 14 种领先三个最强对手，在六个国际基准上与母模型持平。
正文
Abstract:India's 22 scheduled languages are spoken by over a billion people in conversations that mix languages, often within one turn, overlap freely, and take place over telephones, in meetings, and in media. Knowing who spoke when, the task of speaker diarization, is essential to speech applications for such conversation: voice agents conversing with several people at once, speaker attributed transcription of meetings and calls, contact center analytics, and media monitoring. We present a systematic study of speaker diarization across all 22 languages on IndicDiarBench and introduce Anuvani-Diarization, which adapts a self-supervised end-to-end segmentation model to Indian conversation with uncertainty aware silver supervision. Evaluated zero shot alongside nine recent diarization systems, it reaches a duration weighted diarization error rate (DER) of 16.78% on all 1,164 recordings, the lowest of all systems we ran and second only to the best published result (16.0%), and leads the three strongest rivals in 14 of 22 languages. It matches its parent on six international benchmarks. We also show that the benchmark's utterance level annotation convention is a hidden variable worth 1.9 to 4.1 DER points, depending on the system, and that a convention learned from other Indian transcriptions does not transfer.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.20327 [cs.CL] |
| (or arXiv:2610.20327v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.20327 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Stephen Fernandes [view email]
[v1]
Wed, 14 Oct 2026 15:12:20 UTC (302 KB)
来源：arXiv：cs.CL · arxiv.org