XTREME-S：评估跨语言语音表征
计算与语言
2022-04-14 v3
摘要
我们提出XTREME-S，一个新的用于评估多种语言中通用跨语言语音表征的基准。XTREME-S涵盖四个任务族：语音识别、分类、语音到文本翻译与检索。覆盖10余语系的102种语言、3个不同领域与4个任务族，XTREME-S旨在简化多语言语音表征评估，并催化“通用”语音表征学习的研究。本文描述该新基准，并使用XLS-R与mSLAM在所有下游任务上建立首个仅语音与语音-文本基线。我们阐明了设计选择并详述如何使用该基准。数据集与微调脚本已于 https://hf.co/datasets/google/xtreme_s 便捷获取。
引用
@article{arxiv.2203.10752,
title = {XTREME-S: Evaluating Cross-lingual Speech Representations},
author = {Alexis Conneau and Ankur Bapna and Yu Zhang and Min Ma and Patrick von Platen and Anton Lozhkov and Colin Cherry and Ye Jia and Clara Rivera and Mihir Kale and Daan Van Esch and Vera Axelrod and Simran Khanuja and Jonathan H. Clark and Orhan Firat and Michael Auli and Sebastian Ruder and Jason Riesa and Melvin Johnson},
journal= {arXiv preprint arXiv:2203.10752},
year = {2022}
}
备注
Minor fix: language code for Filipino (Tagalog), "tg" -> "tl"